Decoding Cancer Evolution: Single-Cell Analysis of Clonal Dynamics and Therapeutic Resistance

Abstract

This article provides a comprehensive overview of how single-cell technologies are revolutionizing our understanding of clonal evolution in cancer. Aimed at researchers, scientists, and drug development professionals, it explores the foundational principles of tumor heterogeneity, details cutting-edge methodological approaches like scDNA-seq and multi-omics integration, and addresses key troubleshooting challenges in data analysis. By examining validation strategies and clinical applications, particularly in tracking therapy-resistant clones and informing adaptive treatment regimens, the content bridges genomic discoveries with translational impact, offering a roadmap for leveraging single-cell resolution to overcome treatment failure in oncology.

The Evolutionary Engine: Unraveling Clonal Diversity and Tumor Heterogeneity

The fundamental understanding of cancer has been revolutionized by the application of evolutionary biology principles. The conceptual framework, first articulated by Peter Nowell in 1976, posits that tumorigenesis is an evolutionary process whereby cancers originate from a single neoplastic cell and evolve through a process of selection for somatic alterations, leading to the proliferation and survival of the most aggressive clones [1]. This Darwinian model rests upon three key pillars: variation (genetic and epigenetic heterogeneity within cell populations), heredity (clonal propagation of advantageous traits), and selection (differential fitness imposed by microenvironmental pressures and therapeutic interventions) [1]. While grounded in somatic selection, contemporary research reveals that a strict Darwinian model alone is insufficient to fully explain cancer evolution, necessitating integration with concepts such as macroevolutionary jumps, neutral evolution, and cellular plasticity [1] [2].

The clinical ramifications of cancer evolution are profound, driving therapeutic resistance, metastatic progression, and ultimately patient mortality. This whitepaper examines the principles of clonal evolution and selection through the lens of single-cell analysis, providing researchers with both theoretical frameworks and practical methodologies for investigating these dynamics in cancer systems.

Theoretical Foundations of Clonal Evolution

Historical Models of Tumor Evolution

Historically, tumor evolution was viewed as a linear succession of clonal cell divisions, where alterations accrue in progenitor cells in a stepwise fashion and endow cells with strong selective advantages, enabling previous clones to be outcompeted [1]. This linear model has been largely supplanted by the concept of branched evolution, supported by single-cell sequencing data across multiple cancer types [1] [3]. In branched evolution, multiple subclones derived from a common ancestor diverge and expand simultaneously with differing fitness levels, resulting in intratumoral heterogeneity (ITH) [1]. A consequence of branched tumor evolution is ITH—the coexistence of molecularly and phenotypically distinct subclones within a tumor [1].

Beyond Strict Darwinism: Expanding Evolutionary Concepts

Recent evidence suggests several non-Darwinian and post-Darwinian concepts must be incorporated into cancer evolutionary models:

• Macroevolutionary jumps: Single catastrophic genomic events such as chromothripsis (massive chromosomal shattering and rearrangement) and whole-genome doubling can drive tumor evolution in rapid bursts, contradicting Darwin's central thesis of gradualism [1].
• Neutral evolution: In some tumors, selective pressures may be minimal, with clonal dynamics driven primarily by genetic drift rather than selective advantage [1].
• Cellular plasticity: Cancer cells demonstrate remarkable phenotypic plasticity, transitioning between states without genetic alteration, enabling adaptation to therapeutic pressures [2].
• Extended Evolutionary Synthesis (EES): This framework integrates Darwinian selection with additional concepts including niche construction, developmental plasticity, and extra-genetic inheritance, providing a more comprehensive model of cancer evolution [2].

Table 1: Key Evolutionary Models in Cancer Biology

Evolutionary Model	Core Principles	Clinical Implications
Linear Evolution	Sequential acquisition of driver mutations; selective sweeps	Limited clonal diversity; simpler therapeutic targeting
Branched Evolution	Divergent subclones; spatial and temporal heterogeneity	Therapeutic resistance; sampling bias in biopsies
Neutral Evolution	Genetic drift without selective advantage; mutation accumulation	Reduced selective pressure; different therapeutic approaches
Macroevolution	Single catastrophic events (chromothripsis, WGD)	Rapid progression; genomic instability
Extended Evolutionary Synthesis	Integration of plasticity, niche construction, non-genetic inheritance	Multi-dimensional therapeutic strategies

Methodological Approaches for Tracking Clonal Evolution

Single-Cell Sequencing Technologies

The resolution of clonal architecture requires single-cell approaches, as bulk sequencing obscures cellular heterogeneity and provides only averaged genomic signals [4]. Several technological platforms enable deconvolution of clonal structure:

• Single-cell whole-genome sequencing (scWGS): Directly reveals clonal composition based on copy number alterations and structural variants at diagnosis, enabling evolutionary tracking over time [5]. The DLP+ platform represents a high-throughput, tagmentation-based shallow scWGS approach that identifies copy-number alterations, SVs, and complex rearrangements at 0.5-Mb resolution [5].
• Single-cell RNA sequencing (scRNA-seq): Enables simultaneous interrogation of genotype and phenotype, revealing transcriptional states associated with drug resistance [5] [6]. Recent computational advances like scClone allow mutation detection and clonal inference from scRNA-seq data, despite technical challenges including expression drop-out and allelic imbalance [4].
• Spatial transcriptomics: Maps clonal distributions within histological context, revealing geographical relationships between subclones and microenvironmental elements [4].

The CloneSeq-SV Workflow for Evolutionary Tracking

A sophisticated methodology for monitoring clonal dynamics combines scWGS with targeted deep sequencing of clone-specific structural variants (SVs) in cell-free DNA (cfDNA) [5]. This CloneSeq-SV approach exploits tumor clone-specific SVs as highly sensitive endogenous cfDNA markers, enabling relative abundance measurements and evolutionary analysis of co-existing clonal populations throughout therapy [5].

Diagram 1: CloneSeq-SV workflow for tracking evolution

Computational Tools for Clonal Deconvolution

Computational methods are essential for reconstructing evolutionary trajectories from single-cell data. The scClone toolkit exemplifies this approach, integrating variant detection and genotype inference for scRNA-seq and spatial transcriptomic data while addressing technical artifacts like expression drop-out and allelic imbalance [4]. This pipeline enables:

• Direct processing of raw sequencing reads to detect somatic mutations
• Imputation of technical drop-outs to recover missing genetic information
• Interactive visualization of clonal structures and evolutionary relationships
• Integration of single-cell transcriptomic annotations with mutational signatures [4]

Table 2: Quantitative Framework for Clonal Evolution Analysis

Parameter	Measurement Approach	Biological Interpretation
Variant Allele Frequency (VAF)	Deep sequencing; duplex error correction	Clonal abundance; subclonal architecture
Structural Variant (SV) Error Rate	Off-target patient controls; duplex vs. simplex sequencing	Assay specificity; optimal marker selection
Clone-Specific SV Abundance	Targeted capture; longitudinal cfDNA monitoring	Clonal dynamics under therapeutic pressure
Copy Number Variation (CNV)	scWGS; inference from scRNA-seq	Genome instability; phylogenetic relationships
Transcriptional Signature Association	Integrated scRNA-seq/genotype analysis	Phenotypic consequences of genetic evolution

Experimental Evidence and Case Studies

Clonal Evolution in Ovarian Cancer

Application of CloneSeq-SV to 18 patients with high-grade serous ovarian cancer (HGSOC) over multi-year periods revealed that drug resistance typically arose from selective expansion of a single or small subset of clones present at diagnosis [5]. This research demonstrated:

• Drug-resistant clones frequently showed distinctive genomic features including chromothripsis, whole-genome doubling, and high-level amplifications of oncogenes such as CCNE1, RAB25, MYC, and NOTCH3 [5].
• Phenotypic analysis of matched single-cell RNA sequencing data indicated pre-existing and clone-specific transcriptional states such as upregulation of epithelial-to-mesenchymal transition and VEGF pathways, linked to drug resistance [5].
• Structural variants demonstrated superior signal-to-noise characteristics in cfDNA analyses compared to SNVs, with error rates orders of magnitude lower, enabling more sensitive detection of minimal residual disease [5].

Evolutionary Tracking in Ewing Sarcoma

Single-cell RNA sequencing of Ewing Sarcoma tumors demonstrated significant transcriptional heterogeneity and clonal evolution prior to treatment [6]. Analysis revealed:

• Conserved gene expression programs related to proliferation and Ewing sarcoma gene targets that correlated with overall survival [6].
• Copy-number analysis identified subclonal evolution within patients prior to treatment, indicating early Darwinian selection [6].
• An immunosuppressive microenvironment with complex intercellular communication among tumor and immune cells [6].

Progression Trajectories in Head and Neck Squamous Cell Carcinoma

Comprehensive scRNA-seq profiling across the normal-to-malignant continuum in HNSCC identified the transcriptional development trajectory of malignant epithelial cells and a tumorigenic epithelial subcluster regulated by TFDP1 [3]. Key findings included:

• Infiltration of POSTN+ fibroblasts and SPP1+ macrophages gradually increased with tumor progression, shaping a desmoplastic microenvironment that reprogrammed malignant cells [3].
• During lymph node metastasis, exhausted CD8+ T cells with high CXCL13 expression strongly interacted with tumor cells to acquire more aggressive phenotypes of extranodal expansion [3].
• Malignant epithelial cells in primary and recurrent tumors displayed distinct features, providing a foundation for precise selection of targeted therapy for tumors at different stages [3].

Quantitative Frameworks and Research Reagents

Research Reagent Solutions for Clonal Evolution Studies

Table 3: Essential Research Reagents for Clonal Evolution Studies

Reagent/Category	Specific Examples	Research Application
Single-Cell Platforms	10X Genomics; C1 platform	High-throughput cell capture and barcoding
Hybrid Capture Probes	Patient-bespoke SV flanking probes	Clone-specific marker enrichment in cfDNA
Enzyme Inhibitors	Competitive reversible inhibitors; IC50 determination	Quantitative assessment of drug response
Viability Assays	Cell Titer Glo (CTG) ATP measurement	Cellular viability post-treatment
Cell Sorting Markers	CD138 (myeloma); EPCAM/CDH1 (epithelial)	Target population isolation

Quantitative Biology Framework

Quantitative approaches to cancer biology enable modeling of evolutionary dynamics and therapeutic responses. Key methodologies include:

• Michaelis-Menten kinetics: Models enzyme-ligand or enzyme-substrate binding and catalysis for target engagement studies [7]. Reaction velocity is described as v = ([S]Vmax)/([S] + Km), where Vmax represents maximum velocity and Km the substrate concentration at half-maximal velocity [7].
• Dose-response modeling: The 4-parameter logistic nonlinear regression model (4PL) describes sigmoid-shaped response patterns for inhibitor compounds [7]. The IC50 (inhibitor concentration yielding 50% inhibition) serves as a key metric for compound potency [7].
• Criteria for robust concentration-response curves: Well-defined top and bottom plateau values established using sufficient inhibitor concentration ranges; minimum of 8-10 concentration data points equally spaced; three biological replicates per data point [7].

Diagram 2: Computational analysis pipeline

Clinical Translation and Therapeutic Implications

The understanding of cancer as a Darwinian system has profound implications for clinical oncology:

• Evolution-informed adaptive therapy: Rather than maximum tolerated dose approaches that inevitably select for resistant clones, adaptive strategies aim to maintain sensitive populations that suppress resistant subclones [5] [2].
• Targeting clonal vulnerabilities: Resistant clones frequently depend on specific amplified oncogenes or pathways (e.g., ERBB2 amplification in HGSOC), creating therapeutic opportunities [5].
• Early intervention strategies: Cancer precursor clones harboring driver mutations emerge early, sometimes during the first two decades of life, providing opportunities for early intervention [2].
• Liquid biopsy monitoring: Clone-specific structural variants in cfDNA enable non-invasive tracking of clonal dynamics during treatment, potentially guiding therapeutic adjustments before radiographic progression [5].

The integration of evolutionary principles into oncology practice represents a paradigm shift from reactive to proactive cancer management, leveraging an understanding of Darwinian dynamics to outmaneuver cancer's adaptive strategies.

Cancer operates as a complex Darwinian system governed by principles of variation, heredity, and selection. Single-cell technologies have revolutionized our ability to dissect clonal architecture and evolutionary trajectories, revealing both canonical Darwinian dynamics and non-Darwinian processes including macroevolutionary jumps, cellular plasticity, and neutral evolution. The integration of these concepts through frameworks like the Extended Evolutionary Synthesis provides a more comprehensive understanding of tumor progression, metastasis, and therapeutic resistance.

Methodological advances in single-cell sequencing, computational analysis, and liquid biopsy monitoring are translating evolutionary principles into clinical applications. By tracking clonal dynamics in real-time and understanding the selective pressures shaping tumor evolution, researchers and clinicians can develop evolution-informed therapeutic strategies that anticipate and circumvent resistance mechanisms. The future of oncology lies in embracing cancer's evolutionary nature to develop more durable and effective control strategies.

Clonal evolution describes the process by which cancer cells accumulate genetic mutations over time, passing them to their descendants and generating intratumor heterogeneity (ITH). This genetic diversity, driven by genomic instability, provides the substrate for natural selection, allowing subpopulations of cells with advantageous mutations to expand [8]. Understanding the patterns of this evolution—monoclonal, linear, and branched—is critical for deciphering tumor progression, therapeutic resistance, and relapse [9]. Single-cell multiomics technologies have revolutionized this field by enabling researchers to dissect ITH at an unprecedented resolution, moving beyond the limitations of bulk sequencing which averages signals across diverse cell populations [10] [11]. These technologies allow for the simultaneous analysis of genomic, transcriptomic, and epigenomic landscapes within individual cells, revealing the dynamic clonal architecture and complex phylogenetic trajectories that underlie cancer pathogenesis and treatment failure [12] [11].

Defining the Core Evolutionary Patterns

Cancer evolution can be categorized into several fundamental patterns based on the phylogenetic relationships of subclones. The three primary patterns are monoclonal, linear, and branched evolution.

• Monoclonal Evolution: This pattern is characterized by the dominance of a single, genetically uniform clone at the time of sampling. While most cells share a core set of driver mutations, minor genetic deviations may exist in individual cells without forming stable, expanding subclones [12].
• Linear Evolution: In this model, evolution follows a sequential, step-wise path. A founding clone acquires a new driver mutation, which confers a selective advantage and allows it to outcompete its predecessors, leading to a clonal expansion. This process repeats, resulting in a succession of dominant clones that replace one another over time [12] [13].
• Branched Evolution: This pattern involves the divergence of a clone into multiple distinct lineages, resulting in a tree-like phylogenetic structure. Different subclones evolve in parallel, often harboring unique sets of mutations. This creates a highly heterogeneous tumor where multiple subclones can co-exist, compete, and be subject to independent selection pressures [12] [13].

Quantitative Landscape of Evolutionary Patterns

The prevalence and clinical impact of different evolutionary patterns are revealed through large-scale single-cell studies. The table below summarizes key quantitative findings from recent research in acute myeloid leukemia (AML) and high-grade serous ovarian cancer (HGSOC).

Table 1: Prevalence and Characteristics of Evolutionary Patterns in Human Cancers

Cancer Type	Evolution Pattern	Prevalence	Key Genomic Features	Clinical/Experimental Context
Complex Karyotype AML (CK-AML) [12]	Monoclonal	2 of 8 cases (25%)	Inversions at 3q generating RPN1–MECOM fusion; low intrapatient karyotype heterogeneity.	Diagnosis or salvage samples.
	Linear	3 of 8 cases (38%)	Step-wise acquisition of structural variants.	Diagnosis or salvage samples.
	Branched Polyclonal	3 of 8 cases (38%)	Highest intrapatient karyotype heterogeneity; ongoing karyotype remodeling.	Diagnosis or salvage samples.
High-Grade Serous Ovarian Cancer (HGSOC) [5]	Branched at Diagnosis, Reduced at Relapse	Predominant model	Pre-existing resistant clones with chromothripsis, whole-genome doubling, amplifications (e.g., CCNE1, MYC).	Pre-treatment tissue; drug resistance arose from selective expansion of a subset of pre-existing clones.

The distribution of these patterns has direct clinical implications. In a study of CK-AML, branched evolution was associated with the highest levels of intratumor karyotype heterogeneity [12]. Furthermore, research in HGSOC indicates that drug-resistant clones frequently pre-exist at diagnosis within a branched architecture and are selectively expanded by therapy, leading to a reduction in clonal complexity at relapse [5].

Methodologies for Dissecting Evolutionary Patterns

Single-Cell Multiomics Technologies

The delineation of clonal patterns relies on advanced single-cell technologies that link genotype to phenotype.

• Single-Cell DNA Sequencing (scDNA-seq): This is the gold standard for directly profiling mutations, such as copy number variations (CNVs) and single nucleotide variants (SNVs), in individual cells. Methods like single-cell whole-genome sequencing (scWGS) are used to reconstruct clonal phylogenies and define subclonal structure based on allele-specific copy number alterations and structural variants (SVs) [10] [5] [11].
• Single-Cell RNA Sequencing (scRNA-seq): This technique characterizes the transcriptome of individual cells, allowing for the identification of distinct cell states and functional phenotypes. When combined with genotypic data, it can connect evolutionary subtypes to specific transcriptional programs, such as epithelial-to-mesenchymal transition (EMT) or stem-like states [10] [14] [11].
• Cellular Indexing of Transcriptomes and Epitopes by Sequencing (CITE-seq): A multiomics approach that simultaneously measures single-cell transcriptomes and surface protein levels using oligonucleotide-labeled antibodies. This provides a powerful link between cellular genotype, transcriptional state, and immunophenotype [12] [11].
• Single-Cell Strand Sequencing (Strand-seq): A specialized technique for haplotype-resolved sequencing that is particularly effective for discovering complex structural variants—such as chromothripsis and breakage-fusion-bridge cycles—in a single cell. It can be coupled with scNOVA (single-cell nucleosome occupancy and genetic variation analysis) to integrate structural variant discovery with epigenomic features [12].

Key Experimental and Analytical Workflows

The following diagram illustrates a generalized workflow for a single-cell multiomics study designed to reconstruct clonal evolution patterns.

Diagram 1: Single-Cell Multiomics Workflow for Clonal Evolution Analysis. This workflow outlines the key steps from tumor sampling to the classification of evolutionary patterns and their clinical correlation.

Specific methodologies are tailored to answer distinct biological questions. For instance, the CloneSeq-SV workflow was developed to track clonal dynamics in patient blood samples over time. This method involves performing scWGS on a pretreatment tumor to identify clone-specific structural variants, which are then used as highly specific endogenous markers to track the abundance of individual clones over the therapeutic time course via targeted deep sequencing of cell-free DNA (cfDNA) [5]. Another powerful approach is the Luria-Delbrück experimental design, used to distinguish transient transcriptional heterogeneity from clonally stable epigenetic memory. In this design, single cells are isolated and expanded into clonal populations. If a transcriptional state is stably transmitted to daughter cells, the inter-clonal variance will recapitulate the single-cell heterogeneity of the founding population, indicating epigenetic memory [14].

Table 2: Essential Research Reagents and Platforms for Single-Cell Clonal Analysis

Reagent / Platform	Function	Key Utility in Clonal Evolution
10x Genomics Chromium [10]	High-throughput single-cell partitioning and barcoding.	Enables scalable scRNA-seq and scDNA-seq for profiling thousands of cells from a single tumor.
Oligonucleotide-labeled Antibodies (CITE-seq) [12] [11]	Simultaneous profiling of surface protein expression and transcriptome in single cells.	Links clonal genotypes (from integrated data) to immunophenotypes, identifying surface markers of subclones.
Hybrid Capture Probes (for cfDNA) [5]	Target enrichment of clone-specific mutations in cell-free DNA.	Enables longitudinal, non-invasive tracking of clonal dynamics in patient plasma via CloneSeq-SV.
Tn5 Transposase (scATAC-seq) [10]	Profiling genome-wide chromatin accessibility in single cells.	Reveals epigenetic heterogeneity and regulatory programs associated with different subclones.
Phyolin [13]	Computational constraint programming tool.	Classifies a tumor's evolutionary history as linear or branched from scDNA-seq data.

Visualizing Evolutionary Relationships and Pathways

The core evolutionary patterns can be conceptualized through phylogenetic trees that map the accumulation of mutations. The following diagram illustrates the key differences between monoclonal, linear, and branched architectures.

Diagram 2: Phylogenetic Trees of Core Evolutionary Patterns. This diagram visualizes the three primary patterns of clonal evolution. Colored edges represent new driver mutations that confer a selective advantage, leading to clonal expansion.

Beyond genetics, clonal evolution is influenced by the tumor microenvironment and epigenetic regulation. Single-cell analyses have revealed that different subclones can occupy distinct niches and exhibit unique transcriptional and epigenetic states. For example, in colorectal cancer cells, a continuum of epithelial-to-mesenchymal (EMT) transcriptional identities can be stably maintained clonally, indicating a form of non-genetic (epigenetic) memory that diversifies the tumor population [14]. Furthermore, the derepression of retrotransposons like LINE-1 in cancer cells can act as a source of genomic instability, causing double-strand breaks and insertional mutagenesis, thereby fueling ITH and influencing clonal fitness [8].

Clinical Implications and Therapeutic Perspectives

The patterns of clonal evolution have profound implications for cancer diagnosis, treatment, and monitoring.

• Therapy Resistance: Branched evolution is a major driver of therapy resistance. The co-existence of multiple genetically distinct subclones within a tumor means that a therapy targeting one subclone may leave others unharmed, leading to relapse. In HGSOC, resistance frequently arises from the selective expansion of a small subset of pre-existing clones that often harbor genomic features like chromothripsis or specific oncogene amplifications (e.g., CCNE1, MYC) [5]. Similarly, in CK-AML, subclones frequently display ongoing karyotype remodeling and varied drug-response profiles [12].
• Minimal Residual Disease (MRD) and Relapse: Single-cell multiomics enables highly sensitive monitoring of MRD. By identifying and tracking clone-specific markers (e.g., structural variants) in cfDNA, clinicians can detect the early expansion of resistant clones long before clinical relapse becomes apparent [10] [5].
• Evolution-Informed Therapy: Understanding a tumor's evolutionary trajectory opens the door to novel treatment strategies. These include adaptive therapy, which aims to control rather than eliminate a tumor by maintaining sensitive clones that can suppress the growth of resistant ones, and extinction therapy, which seeks to simultaneously target multiple evolutionary pathways to prevent escape [9]. The identification of subclone-specific vulnerabilities, such as a dependency on BCL-xL in a specific leukemic stem cell subclone [12], provides a rationale for personalized combination treatments designed to counteract branched evolution and preempt resistance.

The architectural blueprints of cancer—monoclonal, linear, and branched evolution—provide a critical framework for understanding tumor development and therapeutic failure. Single-cell multiomics technologies have been instrumental in delineating these patterns, revealing a complex landscape of genetic and non-genetic heterogeneity. The integration of genomic, transcriptomic, and epigenomic data at cellular resolution is transforming our approach to cancer treatment, moving the field toward evolution-informed, adaptive, and highly personalized therapeutic strategies that anticipate and counteract the evolutionary maneuvers of cancer.

In the paradigm of cancer evolution, genomic instability serves as the fundamental engine that generates diversity upon which natural selection can act. Among the various forms of instability, three macroscopic genomic alterations—chromothripsis, whole-genome doubling (WGD), and somatic copy-number variations (CNVs)—function as powerful drivers of clonal diversity and tumor adaptation. Chromothripsis, or "chromosome shattering," represents a catastrophic single-genomic event where tens to hundreds of chromosomal rearrangements occur in a single crisis [12] [15]. Whole-genome duplication involves the duplication of the entire chromosome complement, providing a permissive background for extensive genomic exploration [16] [17]. CNVs, comprising recurrent focal and arm-level gains and losses, represent the most common form of somatic genetic variation in cancer genomes [18]. When studied through the resolving lens of single-cell analysis, the interplay of these mechanisms reveals a complex landscape of clonal architecture, with profound implications for therapeutic resistance, immune evasion, and metastatic progression. This technical review synthesizes current understanding of how these drivers collectively shape tumor evolution, providing methodologies for their investigation and quantitative frameworks for their clinical interpretation.

Molecular Mechanisms and Genomic Consequences

Chromothripsis: Catastrophic Rearrangement as an Evolutionary Catalyst

Chromothripsis arises through a single catastrophic event involving chromosomal shattering and subsequent error-prone repair, generating complex genomic rearrangements with significant evolutionary potential. The molecular triggers primarily involve chromosome missegregation during mitosis, which can lead to micronucleus formation [19]. Within these micronuclei, premature chromosome condensation and compromised DNA repair create an environment where numerous double-strand breaks occur simultaneously [19]. The repair of these shattered fragments occurs primarily through non-homologous end joining (NHEJ) mechanisms, evidenced by minimal microhomology at breakpoint junctions and sensitivity to DNA-PKcs and PARP inhibition [19].

The genomic consequences of chromothripsis are profound, with three distinct rearrangement profiles identified in experimental models:

• Single-fragment excision: A chromothriptic event excises a single fragment containing an oncogene (e.g., DHFR), which circularizes into double-minute chromosomes [19].
• Multi-fragment reassembly: Multiple non-contiguous chromosomal fragments (up to 17 fragments spanning 80 Mb) reassemble into circular extrachromosomal DNA elements [19].
• Chromosomal loss with retention: An entire chromosome copy is lost except for a rescued fragment that circularizes into an amplifiable element [19].

These rearrangements facilitate rapid gene amplification under therapeutic selection, with continuing structural evolution through successive chromothriptic events enabling increased drug tolerance [19]. In acute myeloid leukemia with complex karyotype (CK-AML), chromothripsis generates extensive intratumoral heterogeneity through oscillating copy-number states and complex translocation networks that reshape the genomic landscape [12].

Whole-Genome Doubling: A Permissive Platform for Genomic Exploration

Whole-genome doubling represents a macro-evolutionary transition that fundamentally alters the genomic landscape of cancer cells. The cellular mechanisms driving WGD include:

• Mitotic slippage: Cells enter mitosis but fail to undergo proper chromosome segregation or cytokinesis, often due to persistent activation of the spindle assembly checkpoint [17].
• Cytokinesis failure: Nuclear division completes but cytoplasmic division fails, generating binucleated tetraploid cells [17].
• Endoreduplication: Successive rounds of genome replication occur without intervening mitoses [17].

These processes are enabled by loss of critical tumor suppressors, particularly TP53, which normally prevents the proliferation of tetraploid cells through cell cycle arrest or apoptosis [17] [20]. The evolutionary advantages of WGD are multifaceted. First, it provides a buffer against deleterious mutations by restoring heterozygosity in regions with extensive loss-of-heterozygosity, effectively rescuing cells from the negative fitness consequences of haploidization [17]. Second, WGD promotes chromosomal instability (CIN) through ongoing chromosomal missegregation, increasing cellular diversity and adaptive potential [16]. Third, WGD induces profound chromatin reorganization characterized by loss of chromatin segregation (LCS), wherein boundaries between chromatin compartments and topologically associated domains become blurred, potentially enabling oncogenic reprogramming [20].

Single-cell sequencing has revealed that WGD is not a single historical event but rather a dynamic, ongoing process in tumor evolution. In high-grade serous ovarian cancer (HGSOC), multiple WGD multiplicities coexist within individual patients, with 40 of 41 patients exhibiting cells with different WGD histories simultaneously [16]. This ongoing genome doubling generates continuous diversity that fuels tumor evolvability.

Somatic Copy-Number Variations: Recurrent Genomic Imbalances with Functional Impact

Somatic CNVs represent the most prevalent form of large-scale genomic alteration in cancer, with distinct patterns of selection across tumor types. The generation of CNVs occurs through multiple mechanisms:

• Breakage-fusion-bridge (BFB) cycles: Telomere loss leads to repeated fusion and breakage of sister chromatids, generating amplifications and deletions [12].
• Focal amplifications and deletions: Unequal crossing over or replication errors create small regions of gain or loss containing oncogenes or tumor suppressor genes.
• Arm-level and whole-chromosome events: Mis-segregation during mitosis produces large-scale aneuploidies with significant transcriptomic consequences.

CNVs contribute to tumor evolution through dosage effects on key cancer pathways. In non-small cell lung cancer (NSCLC), clone-specific analysis reveals that metastasis-seeding clones are enriched for losses affecting tumor suppressor genes (e.g., TP53, RB1) and amplifications affecting cell cycle regulators (e.g., CCND1) [18]. The temporal ordering of CNV acquisition follows distinct evolutionary patterns—monoclonal, linear, and branched—with branched evolution associated with higher intratumoral heterogeneity and potentially worse clinical outcomes [12] [18].

Table 1: Quantitative Impact of Genomic Diversity Drivers Across Cancer Types

Driver Mechanism	Prevalence in Human Cancers	Key Associated Alterations	Evolutionary Consequences
Chromothripsis	20% across various cancers [15]	TP53 mutations (frequent) [15], Complex structural variants [12]	Rapid oncogene amplification [19], Extrachromosomal DNA formation [19]
Whole-Genome Doubling	30-40% of solid cancers [17], 52% in Japanese cancer cohort [15]	TP53 loss [17] [20], Ongoing chromosomal instability [16]	Increased clone copy-number diversity [18], Altered chromatin segregation [20]
Somatic CNVs	Near-ubiquitous in solid tumors	Arm-level gains/losses, Focal amplifications/deletions [18]	Lineage diversification, Subclonal selection under therapy [18]

Single-Cell Methodologies for Resolving Clonal Architecture

Experimental Workflows for Single-Cell Multiomics Analysis

Advanced single-cell technologies now enable simultaneous profiling of genomic, transcriptomic, and epigenomic features within individual cells, providing unprecedented resolution of clonal architecture.

Figure 1: Single-Cell Multiomics Workflow for Clonal Evolution Analysis

The scNOVA-CITE framework exemplifies this integrated approach, coupling single-cell Strand-seq for haplotype-resolved structural variant detection with CITE-seq for simultaneous transcriptome and surface protein profiling [12]. This methodology enables direct correlation of genetic alterations with phenotypic states at single-cell resolution. For WGD analysis, the Direct Library Preparation (DLP+) protocol enables high-throughput single-cell whole-genome sequencing (scWGS) with median coverage of 0.060× per cell, sufficient for copy-number profiling and WGD detection [16]. Critical to this analysis is the inference of WGD multiplicity—the number of WGD events in each cell's evolutionary history—based on allele-specific copy-number profiles [16].

For computational inference of clonal relationships from bulk multi-sample data, the ALPACA (Allele-Specific Phylogenetic Analysis of Copy Number Alterations) method leverages phylogenetic trees reconstructed from single-nucleotide variant frequencies as a scaffold to guide inference of SCNA evolution [18]. This approach accurately infers clone-specific copy numbers and evolutionary timing of events, outperforming previous methods like HATCHet and MEDICC2 in benchmarking studies [18].

The Scientist's Toolkit: Essential Research Reagents and Platforms

Table 2: Essential Research Reagents and Platforms for Single-Cell Clonal Analysis

Reagent/Platform	Function	Application Example
Strand-seq [12]	Haplotype-resolved structural variant detection	Identifying complex rearrangement patterns in CK-AML [12]
CITE-seq [12]	Simultaneous transcriptome and surface protein profiling	Linking genetic subclones to immunophenotypic states [12]
DLP+ scWGS [16]	High-throughput single-cell whole-genome sequencing	Resolving WGD multiplicity in HGSOC [16]
10x Genomics Single-Cell Platform	Partitioning cells into nanoliter-scale droplets	Generating single-cell libraries for RNA/DNA sequencing
ShatterSeek [15]	Computational detection of chromothripsis	Identifying chromothriptic events from WGS data [15]
ALPACA Algorithm [18]	Inference of clone-specific copy numbers	Reconstructing SCNA evolution in NSCLC [18]

Functional and Clinical Implications

Evolutionary Trajectories and Tumor Adaptation

The interplay between chromothripsis, WGD, and CNVs generates distinct evolutionary patterns with significant clinical implications. Single-cell multiomics analysis of CK-AML reveals three predominant clonal evolution patterns: monoclonal growth (single dominant clone), linear evolution (stepwise acquisition of alterations), and branched polyclonal evolution (multiple competing subclones) [12]. These patterns exhibit different levels of karyotypic heterogeneity, with branched evolution associated with the highest diversity and ongoing karyotype remodeling [12].

WGD timing creates distinct evolutionary modes that shape subsequent tumor development. Analysis of HGSOC reveals three predominant patterns: (1) early fixation followed by considerable diversification, (2) multiple parallel WGD events on a pre-existing background of copy-number diversity, and (3) evolutionarily late WGD in small clones and individual cells [16]. These different temporal patterns influence the rate of copy-number alteration acquisition and the overall evolutionary trajectory of the tumor.

The relationship between these genomic drivers and the tumor immune microenvironment represents another critical dimension of cancer evolution. In HGSOC, WGD-high tumors exhibit cell-cycle dysregulation, STING1 repression, and immunosuppressive phenotypic states despite increased chromosomal missegregation [16]. This contrasts with predominantly diploid tumors, where chromosomal instability triggers inflammatory signaling and cGAS-STING pathway activation [16]. This suggests that WGD not only drives genomic evolution but also shapes the immune contexture, potentially influencing response to immunotherapy.

Therapeutic Implications and Biomarker Potential

The genomic diversity generated by chromothripsis, WGD, and CNVs creates significant therapeutic challenges but also reveals potential vulnerabilities. In CK-AML, single-cell multiomics enables dissection of subclone-specific drug-response profiles, identifying potential LSC-targeting therapies such as BCL-xL inhibition [12]. Similarly, in longitudinal MCL analysis, multiomic profiling reveals how minor clones present at diagnosis acquire different mutations and CNVs, leading to relapse through diverse evolutionary paths [21].

Clone-specific copy-number diversity has emerged as a significant prognostic factor across cancer types. In NSCLC, increased clone copy-number diversity is associated with reduced disease-free survival, with higher SCNA rates observed in tumors with polyclonal metastatic dissemination and extrathoracic metastases [18]. This suggests that metrics of clonal diversity may provide superior prognostic information compared to traditional bulk sequencing approaches.

Table 3: Clinical Associations of Genomic Diversity Drivers

Driver Mechanism	Prognostic Association	Potential Therapeutic Implications
Chromothripsis	Worse overall survival in multiple cancers [15]	Possible sensitivity to DNA repair inhibitors (PARPi) [19]
Whole-Genome Doubling	Poor outcome in most solid tumors [17]	Targeting of WGD-associated vulnerabilities (e.g., G1 checkpoint)
High Clone CNV Diversity	Reduced disease-free survival in lung cancer [18]	Combination therapies addressing multiple subclones simultaneously

Chromothripsis, whole-genome doubling, and somatic copy-number variations represent complementary mechanisms driving clonal diversity in cancer evolution. Through catastrophic genomic restructuring, genome-wide duplication, and recurrent regional imbalances, these processes generate the functional heterogeneity that enables tumor adaptation to therapeutic pressures and microenvironmental constraints. Single-cell multiomics technologies now provide the resolution necessary to dissect these complex evolutionary dynamics, linking genotypic alterations to phenotypic consequences at unprecedented resolution. The clinical translation of these insights requires development of analytical frameworks that incorporate clonal diversity metrics into prognostic models and therapeutic strategies. As these approaches mature, they promise to transform cancer management from its current focus on bulk tumor characteristics to a more nuanced approach that addresses the dynamic, heterogeneous nature of malignant evolution.

The failure of cancer therapy is often a consequence of drug resistance. For years, the prevailing model attributed this treatment failure primarily to acquired resistance, where therapeutic pressure induces new genetic mutations or adaptive phenotypes in cancer cells, granting them survival advantages. However, a paradigm shift is underway, driven by advanced single-cell analyses that reveal a more complex reality: a significant proportion of drug-resistant cells pre-exist within the untreated tumor at the time of diagnosis [22] [5]. This phenomenon, termed pre-existing or intrinsic resistance, fundamentally changes our understanding of tumorigenesis and therapeutic failure.

This whitepaper synthesizes recent evidence demonstrating that tumors are not monoclonal entities but are composed of multiple, genotypically distinct subpopulations, or clones, which engage in dynamic evolution [23]. Within this heterogeneous landscape, certain clones already possess genetic alterations or phenotypic states that confer resistance before any therapeutic intervention [22]. The administration of treatment acts as a powerful selective agent, wiping out the drug-sensitive majority while allowing these pre-adapted, resistant minorities to expand, ultimately leading to disease relapse [5]. Understanding the evidence for, and mechanisms of, this pre-existing resistance is critical for developing evolution-informed treatment strategies that can anticipate and circumvent therapeutic failure.

Quantitative Evidence of Pre-existing Resistant Clones

Advanced genomic studies tracking clonal dynamics from diagnosis through relapse provide direct quantitative evidence for pre-existing resistance. The following table summarizes key findings from seminal studies that have shaped this understanding.

Table 1: Key Genomic Features of Pre-existing Resistant Clones Identified in Clinical Studies

Cancer Type	Study/Method	Key Finding on Pre-existing Resistance	Identified Genomic Features in Resistant Clones
High-Grade Serous Ovarian Cancer (HGSOC)	CloneSeq-SV (scWGS + cfDNA tracking) [5]	Drug resistance typically arose from selective expansion of a single or small subset of clones present at diagnosis.	Chromothripsis, Whole-genome doubling, High-level amplifications of CCNE1, RAB25, MYC, NOTCH3
Acute Myeloid Leukemia (AML)	Whole-genome sequencing pre-/post-relapse [22]	Relapsed tumors showed novel mutations and increased transversion mutations, suggesting therapy-induced DNA damage selected for pre-existing variants.	Pre-existing genetic heterogeneity in primary tumors; expansion of resistant subclones post-therapy.
Colorectal Cancer (Model System)	Genetic Barcoding & Mathematical Modeling [24]	Inferred distinct evolutionary routes: a stable pre-existing resistant subpopulation (SW620 cells) vs. phenotypic switching (HCT116 cells).	Pre-existing resistance fraction (ρ) parameter quantified; distinct phenotype dynamics without new genetic alterations.

The application of the CloneSeq-SV method to high-grade serous ovarian cancer (HGSOC) offers a particularly compelling case. This approach combines single-cell whole-genome sequencing (scWGS) of pre-treatment tumor tissue with targeted deep sequencing of clone-specific structural variants (SVs) in longitudinal cell-free DNA (cfDNA) [5]. This powerful combination allows for the direct observation of clonal populations over a therapeutic time course. The study found that at relapse, the tumor's clonal complexity was often reduced, a hallmark of selective pressure where one or a few pre-existing clones—characterized by disruptive genomic events like chromothripsis and amplifications of oncogenes such as CCNE1 and MYC—outcompete others [5]. This suggests that the genomic "seeds" of therapeutic failure are sown early in tumor development.

Table 2: Phenotypic States Associated with Pre-existing Resistance

Phenotypic State	Functional Role in Pre-existing Resistance	Associated Mechanisms
Cancer Stem Cells (CSCs)	Subpopulation with self-renewal capacity; intrinsically more resistant to chemo/radiotherapy [22].	Upregulated drug efflux (ABC transporters); enhanced DNA repair; resistance to p53-induced apoptosis.
Epithelial-to-Mesenchymal Transition (EMT)	Morphological change linked to a more mesenchymal, drug-tolerant state [22].	Transcriptional upregulation by Snail/Slug; linked to CSC self-renewal programs.
Slow-Cycling Persister Cells	A drug-tolerant state characterized by reduced proliferation, allowing survival during therapy [24].	Non-genetic phenotypic plasticity; can stochastically progress to full resistance.

Complementing these clinical observations, experimental evolution models in colorectal cancer cell lines have quantified the dynamics of pre-existing resistance. Using genetic barcoding, researchers inferred that in SW620 cells, resistance to 5-Fu chemotherapy was driven by the expansion of a stable pre-existing resistant subpopulation [24]. This was contrasted with HCT116 cells, where resistance emerged through phenotypic switching into a slow-growing resistant state. Mathematical modeling of these experiments introduced the "pre-existing resistance fraction" (ρ) as a key parameter to quantify the initial proportion of resistant cells, providing a framework to measure this phenomenon [24].

Experimental Protocols for Detecting Pre-existing Clones

Identifying and characterizing pre-existing resistant clones requires a multifaceted approach, leveraging cutting-edge sequencing technologies and computational tools. Below are detailed methodologies for key experiments cited in this field.

CloneSeq-SV Methodology for Clonal Tracking in Patients

The CloneSeq-SV protocol is designed for longitudinal tracking of tumor clone dynamics in patient blood samples, offering a non-invasive window into clonal evolution [5].

• Pre-treatment Tissue Processing & Single-cell WGS (scWGS):

  • Sample Collection: Fresh tumor tissue is collected during primary debulking surgery or diagnostic biopsy.
  • Single-cell Dissociation: Tissue is dissociated into a single-cell suspension.
  • Library Preparation & Sequencing: scWGS data is generated using a high-throughput, tagmentation-based shallow sequencing approach (e.g., DLP+). This provides data for identifying copy-number alterations, structural variants (SVs), and complex rearrangements.
  • Clonal Phylogeny Reconstruction: Single-cell phylogenetic trees are constructed from allele-specific copy-number alterations using tools like MEDICC2. Clones are defined as divergent clades from these trees.

• Identification of Clone-Specific SVs:

  • Pseudobulk Analysis: Cells from each defined clone are merged to create clone-specific pseudobulk data.
  • High-Resolution CNV/SV Calling: A hidden Markov model (HMM)-based copy-number caller (e.g., HMMclone) is used on pseudobulk data to generate high-resolution copy-number profiles and precisely identify SVs specific to each clone.

• Longitudinal cfDNA Tracking with Bespoke Panels:

  • Probe Design: Patient-bespoke hybrid-capture probes are designed targeting the breakpoint sequences of truncal and clone-specific SVs.
  • Plasma Collection & cfDNA Extraction: Serial blood samples are collected from patients over the therapeutic course. Plasma is separated, and cfDNA is extracted.
  • Duplex Sequencing: cfDNA libraries are prepared and subjected to duplex error-corrected sequencing using the custom probes to achieve ultra-deep, high-fidelity sequencing.
  • Variant Calling & Clonal Abundance Quantification: The abundance of clone-specific SVs is measured in each longitudinal cfDNA sample, allowing the relative proportion of each clone to be tracked over time.

Genetic Barcoding forIn VitroEvolution Studies

This protocol uses heritable genetic barcodes to trace cell lineages and infer resistance dynamics in controlled laboratory settings [24].

• Generation of Barcoded Cell Pool:

  • A diverse library of lentiviral vectors, each containing a unique DNA barcode sequence, is generated.
  • The target cancer cell line (e.g., SW620, HCT116) is infected at a low multiplicity of infection (MOI) to ensure most cells receive a single, unique barcode.
  • Cells are expanded to create a stable, highly diverse barcoded pool.

• Experimental Evolution with Periodic Treatment:

  • The barcoded pool is split into multiple replicate populations.
  • Replicates are exposed to periodic cycles of chemotherapy (e.g., 5-Fluorouracil), interspersed with recovery periods in drug-free media.
  • The total population size is monitored throughout the experiment.

• Lineage Tracing and Population Sampling:

  • At predetermined timepoints (e.g., after each treatment cycle), a known number of cells is sampled from each replicate population.
  • Genomic DNA is extracted, and the barcode regions are amplified via PCR and sequenced to high depth.
  • The relative abundance of each barcode is quantified at each timepoint.

• Mathematical Modeling of Phenotype Dynamics:

  • A mathematical framework is applied to the barcode abundance data and population size data.
  • The model infers the dynamics of sensitive and resistant phenotypes without direct measurement, estimating parameters like the pre-existing resistance fraction (ρ), phenotype switching rates (μ), and fitness costs (δ).

Technical Visualization of Research Workflows

CloneSeq-SV Workflow for Clonal Tracking

The following diagram illustrates the integrated experimental and computational workflow of the CloneSeq-SV method for detecting and tracking pre-existing resistant clones.

Evolutionary Models of Resistance Emergence

This diagram contrasts the classic acquired resistance model with the pre-existing resistance model and its evolutionary dynamics, as revealed by single-cell and lineage tracing studies.

The Scientist's Toolkit: Essential Reagents and Research Solutions

Successfully researching pre-existing resistance requires a suite of specialized reagents and tools. The following table details key solutions for implementing the methodologies discussed in this whitepaper.

Table 3: Essential Research Reagents and Tools for Studying Pre-existing Resistance

Research Solution	Specific Function	Application in Pre-existing Resistance Research
Single-cell Whole Genome Sequencing (scWGS) Kit	Enables low-coverage whole-genome sequencing from single cells to assess copy number variations (CNVs) and structural variants (SVs).	Defining clonal architecture and identifying clone-specific genomic markers (e.g., chromothripsis, amplifications) in pre-treatment tumors [5] [4].
Duplex Sequencing Technology	An error-corrected sequencing method that significantly reduces false-positive mutation calls by tracking both strands of a DNA molecule.	Ultra-sensitive detection of tumor-derived DNA and clone-specific SVs in patient cfDNA, crucial for accurate low-frequency variant detection [5].
Genetic Barcoding Library (Lentiviral)	A diverse pool of viral vectors containing unique DNA barcode sequences for heritable lineage tracing.	Labeling individual tumor cell lineages to track their expansion or contraction during in vitro or in vivo therapy, quantifying pre-existing resistance fractions [24].
Computational Tool: scClone	A computational toolkit that detects somatic mutations and infers clonal structure directly from scRNA-seq data.	Associating cell genotypes (clones) with phenotypes (e.g., EMT state, pathway expression) from single-cell transcriptomes to identify resistant subpopulations [4].
Patient-Bespoke Hybrid Capture Panels	Custom-designed oligonucleotide probes targeting patient-specific genomic breakpoints for deep sequencing.	Enriching for clone-specific SVs in cfDNA for highly sensitive and specific longitudinal monitoring of clonal abundances [5].
Mathematical Modeling Framework	A set of computational models (e.g., phenotype transition models) to infer resistance dynamics from lineage tracing data.	Inferring parameters like pre-existing resistance fraction (ρ) and phenotypic switching rates (μ) from experimental evolution data [24].

The convergence of evidence from clinical tracking studies and experimental models solidifies the concept that pre-existing resistant clones are a fundamental cause of therapeutic failure in cancer. The ability to identify these clones at diagnosis—through the detection of specific genomic hallmarks like chromothripsis and oncogene amplification, or through the inference of resistant phenotypic states—presents a transformative opportunity for oncology.

Moving forward, the challenge and promise lie in translating this knowledge into clinical action. This involves the development of "evolution-informed" diagnostic and treatment strategies. For instance, pre-treatment tumor profiling using single-cell or deep bulk sequencing could identify high-risk patients whose tumors harbor resistant clones at the outset. For these patients, upfront combination therapies designed to target both the dominant sensitive population and the pre-existing resistant minority could be deployed to prevent clonal expansion and relapse [5]. Furthermore, the non-invasive monitoring of clonal dynamics via cfDNA provides a tool for dynamically adapting therapy in response to the earliest signs of resistant clone expansion, moving cancer care towards a more proactive and personalized paradigm. Ultimately, defeating cancer requires not only killing the cells that are present today but also anticipating and eliminating the cells that are destined to cause relapse tomorrow.

Next-Generation Tools: Single-Cell Multi-Omics for Tracking Clonal Dynamics

Intratumoral heterogeneity (ITH) is "fuel to the fire" of cancer evolution, providing the cellular variation upon which natural selection operates [25]. For decades, the prevalent opinion was that tumors follow a strictly linear evolutionary trajectory. However, multi-region sequencing has revealed that the frequent coexistence of subclones with different driver alterations is common across tumor types [25]. Understanding this clonal dynamics is crucial, as treatment resistance, relapse, and metastasis often coincide with the expansion of new clones harboring genomic alterations that confer survival advantages [25] [26].

Single-cell sequencing technologies have emerged as powerful tools to dissect this heterogeneity at the ultimate resolution of individual cells, offering a transformative window into the dynamic process of tumor evolution [25]. This technical guide provides an in-depth examination of four core technologies—scDNA-seq, scRNA-seq, Strand-seq, and CITE-seq—within the context of studying clonal evolution in cancer research, equipping researchers and drug development professionals with the knowledge to deploy these methods effectively.

The following table summarizes the core characteristics, applications, and limitations of each technology in the context of clonal evolution studies.

Table 1: Comprehensive Comparison of Single-Cell Sequencing Technologies

Technology	Primary Molecular Profile	Key Applications in Clonal Evolution	Key Advantages	Principal Limitations
scDNA-seq	Genomic DNA: Copy Number Alterations (CNAs), Single Nucleotide Variants (SNVs)	Reconstruction of clonal phylogenies, identification of subclonal genomic alterations, inference of tumor evolutionary history [26] [27]	Direct interrogation of the genetic drivers of evolution; can be combined with proliferation inference (e.g., SPRINTER algorithm) [27]	Cannot directly link genotype to phenotype; lower throughput than scRNA-seq; requires whole-genome amplification [25]
scRNA-seq	Whole transcriptome: mRNA expression, non-coding RNA, fusion transcripts	Linking genotypic subclones to phenotypic states (e.g., stemness, drug resistance), characterizing tumor microenvironment interactions, identifying transcriptional programs driving evolution [25] [6] [28]	Reveals functional consequences of genetic heterogeneity; identifies cell states and plasticity; high-throughput droplet methods available [25] [28]	Does not directly measure genomic alterations; inference of CNAs from RNA data is indirect and lower resolution [6]
Strand-seq	DNA template strands: Chromosomal rearrangements, structural variants (SVs), haplotype resolution	Resolving complex karyotypes, detecting chromothripsis and breakage-fusion-bridge cycles, precise mapping of SVs in a haplotype-aware manner [12]	Unmasks balanced SVs and complex rearrangements missed by CNA profiling; provides haplotype-phased genomic information [12]	Lower genome coverage (~0.017x); specialized protocol and analysis; technically challenging [12]
CITE-seq	Multiplexed: Transcriptome (RNA) + Surface Proteome (Antibody-Derived Tags)	High-dimensional immunophenotyping alongside cell state analysis, linking surface marker-defined populations to transcriptional subtypes and clonal identities [12]	Adds a robust protein-level dimension to transcriptomic classification; helps validate transcriptional states at the protein level [12]	Limited to surface markers with available antibodies; does not directly measure genomic alterations [12]

Experimental Workflows and Methodologies

Core Single-Cell Sequencing Protocol

While each technology has its specific requirements, they share common foundational steps in single-cell analysis. The workflow begins with single-cell isolation, which can be achieved through various methods including flow cytometry, micromanipulation, or microfluidic chips [29]. Following isolation, cells are lysed to release their molecular contents (DNA or RNA). Due to the minimal starting material, the target molecules must undergo amplification—whole-genome amplification for DNA or cDNA synthesis for RNA. Unique Molecular Identifiers (UMIs) are incorporated at this stage to tag individual molecules before polymerase chain reaction (PCR) amplification, enabling accurate quantification by distinguishing biological replicates from PCR duplicates [25]. The final wet-lab steps involve library construction and preparation for high-throughput sequencing. The crucial dry-lab phase encompasses data analysis, including quality control, sequence alignment, molecular quantification, and advanced analyses such as differential expression, clustering, and phylogenetic reconstruction to decipher clonal relationships [29].

Technology-Specific Methodological Details

scRNA-seq protocols are broadly divided into full-length transcript approaches (e.g., SMART-Seq2, SMART-Seq3) and 3'/5' end-counting methods (e.g., MARS-Seq2, CEL-Seq2, Drop-Seq) [25]. Full-length protocols enable identification of alternative transcript isoforms, fusion events, and SNVs, typically detecting a larger number of transcripts per cell but at a higher cost and with potential for higher amplification noise. In contrast, 3'/5' end-counting methods, particularly droplet-based techniques like the 10x Genomics platform, offer higher throughput at a reduced cost per cell and straightforward UMI integration for more accurate transcript quantification, making them suitable for large-scale atlas construction [25] [29].

scDNA-seq methods face the challenge of whole-genome amplification from minimal DNA. Techniques include multiple displacement amplification (MDA) for superior SNV detection and degenerate oligonucleotide-primed PCR (DOP-PCR) for better CNV detection [25]. Modern platforms like DLP+ represent "direct library preparation" methods that offer a balanced performance for both CNV and SNV detection without preamplification, enabling accurate genomic and evolutionary characterization [25] [27].

Strand-seq is a specialized scDNA-seq technique that sequences DNA template strands from individual cells. Libraries are prepared without strand orientation bias, allowing the determination of inheritance patterns for each of the two parental homologs [12]. This enables the detection of sister chromatid exchanges and the phasing of haplotypes, which is crucial for resolving complex structural variants.

CITE-seq begins with the staining of a single-cell suspension with antibodies conjugated to oligonucleotide barcodes. These antibodies bind to cell surface proteins. Cells are then processed through a standard scRNA-seq workflow (typically droplet-based). During the reverse transcription step, both cellular mRNA and the antibody-derived tags (ADTs) are captured, barcoded with the same cell barcode, and incorporated into the same sequencing library [12]. The ADTs and mRNA are subsequently deconvoluted bioinformatically based on their distinct sequences.

Essential Research Reagents and Materials

Successful execution of single-cell clonal evolution studies requires careful selection of reagents and platforms. The following table details key solutions and their functions.

Table 2: Key Research Reagent Solutions for Single-Cell Clonal Evolution Studies

Category	Reagent / Solution	Critical Function	Application Notes
Cell Isolation & Handling	Microfluidic Chips (e.g., 10x Genomics)	High-throughput single-cell partitioning and barcoding in nanoliter droplets [25]	Essential for profiling thousands of cells; ideal for scRNA-seq and CITE-seq
	Fluorescence-Activated Cell Sorting (FACS)	High-accuracy single-cell dispensing into plate formats based on surface markers [27]	Enables pre-selection of specific populations; used for full-length scRNA-seq (SMART-Seq)
Nucleic Acid Processing	Unique Molecular Identifiers (UMIs)	Random DNA barcodes that tag individual molecules pre-amplification to correct for PCR bias [25]	Crucial for accurate digital quantification in scRNA-seq; enables distinction of biological duplicates from PCR duplicates
	Transposase Enzyme (e.g., Tn5)	Fragments DNA and simultaneously adds adapter sequences for "tagmentation"-based library prep [25] [27]	Core component of DLP+ and other modern scDNA-seq methods; streamlines library construction
Library Preparation Kits	SMART-Seq3/4 Kit	For full-length, plate-based scRNA-seq with high sensitivity and UMI support [25] [29]	Optimal for detecting splice variants, fusions, and SNVs in the transcriptome
	DLP+ Reagent Kit	For single-cell whole-genome sequencing without preamplification [5] [27]	Used for high-resolution CNV/SV detection and clonal phylogeny reconstruction in cancer
Antibody Reagents	CITE-seq Antibody Panels	Oligo-tagged antibodies targeting cell surface proteins (e.g., CD45, CD3, CD19) [12]	Allows simultaneous protein and RNA measurement; requires validation of antibody specificity
Bioinformatic Tools	SPRINTER Algorithm	Infers clone-specific proliferation rates from scDNA-seq data by identifying S- and G2-phase cells [27]	Reveals proliferation heterogeneity among clones; links genetics to functional growth properties
	scGAL Tool	Jointly analyzes scDNA-seq and scRNA-seq data to refine clonal copy number substructure [30]	Uses adversarial learning to reduce technical noise in copy number data using gene expression

Research Applications in Clonal Evolution

Resolving Complex Karyotypes and Evolutionary Patterns in AML

Single-cell multiomics approaches have proven particularly valuable in deciphering the clonal complexity of cancers like acute myeloid leukemia with complex karyotype (CK-AML). By coupling Strand-seq (scNOVA) with CITE-seq, researchers can link complex structural variant landscapes with transcriptional and immunophenotypic states [12]. This integrated analysis has revealed three distinct patterns of clonal evolution in CK-AML: monoclonal growth, linear evolution, and branched polyclonal evolution [12]. For instance, a 2024 study identified that 75% of CK-AML samples harbored multiple subclones that frequently displayed ongoing karyotype remodeling, with some cases showing extensive chromothripsis and breakage-fusion-bridge cycles [12]. This level of resolution is critical for understanding how genetic heterogeneity contributes to therapeutic failure.

Tracking Clonal Dynamics in Solid Tumors and Metastasis

In solid tumors like high-grade serous ovarian cancer (HGSOC), the CloneSeq-SV method combines scDNA-seq on pretreatment tissues with targeted sequencing of clone-specific structural variants in cell-free DNA from patient blood [5]. This approach enables non-invasive monitoring of clonal population dynamics throughout treatment, revealing that drug resistance typically arises from selective expansion of a small subset of clones present at diagnosis [5]. Similarly, in non-small cell lung cancer (NSCLC), the SPRINTER algorithm applied to scDNA-seq data from 14,994 cells demonstrated widespread clone proliferation heterogeneity and revealed that high-proliferation clones have increased metastatic seeding potential and contribute more significantly to circulating tumor DNA (ctDNA) shedding [27]. These findings establish a direct link between clonal evolutionary dynamics and clinically observable phenomena like metastasis and ctDNA load.

Integrating Multiomics to Link Genotype to Phenotype

A significant challenge in cancer evolution lies in understanding how genetic alterations manifest in functional cellular phenotypes. Multi-technology approaches are essential here. For example, clonealign is a computational method that assigns scRNA-seq cells to clones defined by scDNA-seq data, enabling the identification of clone-specific dysregulated biological pathways that would be invisible from either analysis alone [30]. Similarly, scGAL uses a hybrid model to jointly analyze independent single-cell copy number and gene expression data from the same cell line, exploiting the correlation between copy number alterations and gene expression to provide a more refined indication of clonal substructure [30]. These integrated analyses help bridge the critical gap between a clone's genotype and its phenotypic behavior, such as stemness, drug resistance, or metastatic propensity.

The study of clonal evolution in cancer provides a critical framework for understanding how tumors adapt under therapeutic pressure, a process predominantly driven by the expansion of pre-existing, treatment-resistant cellular subpopulations. In high-grade serous ovarian cancer (HGSOC), the most common and lethal form of ovarian cancer, relapse after initial treatment response remains almost universal due to the emergence of drug resistance [31] [32]. Existing methods for monitoring cancer dynamics have largely failed to distinguish between treatment-sensitive and treatment-resistant cell populations, creating a fundamental barrier to predicting and preventing disease recurrence [33].

To address this challenge, researchers have developed CloneSeq-SV, a novel approach that leverages somatic structural variants (SVs) as highly sensitive clonal markers to track tumor evolution through blood tests [31] [34]. This method represents a significant advancement in cancer single-cell analysis research by combining single-cell whole-genome sequencing with targeted deep sequencing of clone-specific genomic structural variants in time-series cell-free DNA [5] [35]. The technology exploits tumor clone-specific structural variants as endogenous molecular barcodes, enabling relative abundance measurements and evolutionary analysis of co-existing clonal populations throughout the therapeutic timeline [5] [36].

This technical guide examines the core principles, methodological framework, and research applications of CloneSeq-SV, positioning it within the broader context of clonal evolution research in cancer single-cell analysis. By providing a comprehensive overview of its experimental protocols, analytical capabilities, and research utilities, we aim to equip scientists and drug development professionals with the knowledge to implement and extend this transformative approach in oncology research.

Background and Rationale

The Challenge of Clonal Evolution in HGSOC

High-grade serous ovarian cancer is characterized by extensive genomic instability and substantial intra-tumoral heterogeneity, creating a diverse ecosystem of cellular subpopulations with varying treatment sensitivities [31] [32]. While initial treatments with surgery, platinum-based chemotherapy, and maintenance therapies often achieve initial clinical response, the selective pressure exerted by these interventions inevitably promotes the expansion of resistant clones, ultimately leading to disease recurrence [5]. This recurring pattern reflects a deeper biological challenge: tumors contain diverse cell populations from their inception, with some possessing inherent resistance mechanisms that pre-date any therapeutic intervention [32] [37].

Traditional monitoring approaches, including imaging and conventional biomarker assays, lack the resolution to detect the dynamic changes in clonal composition that underlie treatment response and resistance emergence. Even serial tumor biopsies present substantial practical and clinical limitations, including invasiveness, sampling bias, and inability to capture the full spatial heterogeneity of the disease [5]. The development of CloneSeq-SV addresses these limitations by enabling non-invasive, high-resolution tracking of clonal dynamics through a simple blood draw, providing unprecedented insight into the evolutionary trajectories of treatment-resistant populations [31] [34].

Structural Variants as Ideal Clonal Markers

Structural variants - large-scale genomic rearrangements including translocations, inversions, deletions, and amplifications - represent particularly advantageous markers for clonal tracking in cancers characterized by genomic instability, such as HGSOC [5]. Unlike single nucleotide variants (SNVs), SVs exhibit orders of magnitude lower error rates in cell-free DNA detection assays, significantly enhancing signal-to-noise ratio and enabling confident detection of tumor DNA even from single events without requiring extensive error correction [5]. Furthermore, SVs frequently associate with high-level amplifications, resulting in elevated per-cell copy numbers that can further enhance detection sensitivity despite being less numerous than SNVs [5].

The unique breakpoint sequences created by structural variants, where distal chromosomal loci become juxtaposed, provide highly specific markers that are remarkably resistant to sequencing errors that commonly cause false positives in cfDNA analyses [5]. This specificity, combined with the relatively low background error rate, makes SVs exceptionally well-suited for tracking minimal residual disease and early relapse detection, where tumor DNA fraction in circulation is typically very low.

Core Methodology and Workflow

The CloneSeq-SV methodology integrates two complementary technological approaches: single-cell whole-genome sequencing of tumor tissue and targeted deep sequencing of structural variants in longitudinally collected cell-free DNA. The complete workflow encompasses sample processing, computational analysis, and evolutionary modeling, as detailed below.

Experimental Workflow and Design

Figure 1: The comprehensive CloneSeq-SV workflow integrates single-cell tumor sequencing with longitudinal blood-based monitoring to enable high-resolution tracking of clonal evolution.

Single-Cell Whole Genome Sequencing

The initial tissue characterization phase begins with collecting fresh tumor samples during primary debulking surgeries or diagnostic laparoscopic biopsies [5]. Following tissue dissociation into single-cell suspensions, CD45+ immune cells are depleted through flow sorting to enrich for malignant cells [16]. Libraries are prepared using the DLP+ protocol, a high-throughput, tagmentation-based shallow scWGS approach that enables identification of copy-number alterations, SVs, and complex rearrangements at 0.5-Mb resolution [5] [16]. In the foundational study, this process generated scWGS data from 21,916 tumor cells (range 232-2,094 cells per patient) with mean coverage of 0.088× (range 0.003-0.349× per cell) [5].

Clonal Decomposition and Phylogenetic Reconstruction

Computational analysis of scWGS data begins with inference of clonal composition based on allele-specific copy number profiles [5]. Single-cell phylogenetic trees are constructed using MEDICC2 with allele-specific copy-number alterations at 0.5-Mb resolution [5]. Clones are defined based on divergent clades from these phylogenetic trees, followed by merging cells from each clone to recompute copy-number profiles at 10-kb resolution using HMMclone, a novel hidden Markov model-based copy-number caller that improves the resolution of pseudobulk clone-specific copy-number profiles and enables more precise matching between copy number and SVs [5].

To identify clone-specific endogenous genomic markers, structural variants and single-nucleotide variants are called from patient-level pseudobulk data, then genotyped in individual cells [5]. The distribution of mutation-positive cells across the phylogenetic tree distinguishes truncal from clone-specific events, with truncal mutations (e.g., TP53 mutations) distributed uniformly across all clones, while clone-specific SNVs and SVs show non-random, clone-restricted distributions [5].

Cell-free DNA Analysis and Clonal Tracking

For longitudinal monitoring, researchers design patient-bespoke hybrid capture probes with 60-base-pair flanking sequence on either side of breakpoints or point mutations [5]. These probes are incorporated into a cfDNA duplex error-corrected sequencing assay, achieving mean raw coverage of 14,137× and mean consensus duplex coverage of 919× [5]. The exceptional specificity of SVs enables highly sensitive detection - error rates for SVs were negligible even in uncorrected sequencing (below 1×10⁻⁷), compared to substantially higher error rates for SNVs (4×10⁻⁴ for uncorrected sequencing) [5]. This improved signal-to-noise ratio allows confident detection of tumor DNA without requiring error correction, though duplex sequencing provides additional validation [5].

Table 1: Key Sequencing Metrics and Performance Characteristics of CloneSeq-SV

Parameter	Single-Cell WGS	cfDNA Sequencing	Performance Metrics
Cells Sequenced	21,916 total (232-2,094 per patient)	N/A	Comprehensive cellular sampling
Coverage Depth	0.088× mean per cell	919× mean consensus duplex coverage	Balanced breadth vs. depth
SV Identification	54 average per patient (9-233 range)	High-confidence detection	Specific clone-specific markers
Error Rates	N/A	SV: <1×10⁻⁷; SNV: 4×10⁻⁴ (uncorrected)	Superior SV signal-to-noise ratio
Tumor Fraction Correlation	N/A	R=0.95 vs. TP53 mutations (p<10⁻¹⁰)	High quantification accuracy

Research Reagent Solutions

Implementation of the CloneSeq-SV methodology requires specialized reagents and computational tools, each serving specific functions in the analytical pipeline. The following table details essential research reagents and their applications in the protocol.

Table 2: Essential Research Reagents and Computational Tools for CloneSeq-SV Implementation

Reagent/Tool	Category	Function in Protocol	Technical Specifications
DLP+ Protocol	Library Preparation	Single-cell whole-genome sequencing	Tagmentation-based; 0.5-Mb resolution for SVs and CNA
MEDICC2	Computational Algorithm	Phylogenetic tree reconstruction	Processes allele-specific CNA at 0.5-Mb resolution
HMMclone	Computational Algorithm	Copy-number calling	Hidden Markov Model; 10-kb resolution for pseudobulk profiles
Hybrid Capture Probes	Molecular Biology	SV target enrichment	Patient-bespoke; 60-bp flanking breakpoints
Duplex Sequencing	Sequencing Method	Error correction	Molecular barcoding; reduces sequencing errors
Flow Cytometry	Cell Sorting	Immune cell depletion	CD45+ antibody-based negative selection

Key Research Findings and Applications

Genomic Features of Drug-Resistant Clones

Application of CloneSeq-SV to 18 HGSOC patients from diagnosis through recurrence revealed that drug-resistant clones frequently harbor distinctive genomic features that may contribute to their survival advantage under therapeutic pressure [5] [32]. These features include:

• Oncogene Amplifications: Resistant clones frequently showed high-level amplifications of known oncogenes including CCNE1, RAB25, MYC, NOTCH3, and ERBB2 [5] [35]. These amplifications potentially drive proliferative advantages and therapeutic resistance mechanisms.

• Whole-Genome Doubling (WGD): WGD events occurred in many resistant populations, with single-cell analyses revealing WGD as an ongoing mutational process that promotes evolvability and dysregulated immunity in HGSOC [16]. WGD-associated tumors exhibited increased cell-cell diversity and higher rates of chromosomal missegregation [16].

• Chromothripsis: This catastrophic genomic event, where chromosomes shatter and reassemble haphazardly, was frequently observed in resistant clones, potentially creating novel genomic rearrangements that confer survival advantages [5] [34].

• Transcriptional Pre-Programming: Matched single-cell RNA sequencing data indicated pre-existing and clone-specific transcriptional states such as upregulation of epithelial-to-mesenchymal transition and VEGF pathways, linked to drug resistance [5] [35].

Evolutionary Dynamics and Clonal Selection

The longitudinal tracking capability of CloneSeq-SV has provided unprecedented insight into the evolutionary dynamics of HGSOC under therapeutic pressure. Several key patterns have emerged from these analyses:

• Pre-Existing Resistance: Drug-resistant clones were consistently present at diagnosis, even before treatment initiation, indicating that resistance mechanisms are inherent to a subset of tumor cells rather than exclusively acquired during therapy [31] [32] [37].

• Selective Expansion: During treatment, drug-sensitive clones are progressively eliminated, while resistant populations expand through positive selection, ultimately dominating the recurrent tumor ecosystem [5] [33].

• Reduced Clonal Complexity: The evolutionary trajectory typically progresses toward reduced clonal complexity at relapse, with recurrence typically dominated by a single or small subset of high-fitness clones [5] [38].

• Polyclonal Resistance: While frequently dominated by a single expanding clone, drug resistance demonstrated polyclonal characteristics in most cases, with multiple resistant subpopulations exhibiting varying genomic features and survival advantages [38].

Clinical Correlation and Therapeutic Implications

The research findings from CloneSeq-SV analyses have significant implications for therapeutic strategy development and clinical trial design:

• Evolution-Informed Adaptive Therapy: The ability to track clonal dynamics in real-time suggests opportunities for evolution-informed adaptive treatment regimens that could preemptively target expanding resistant clones [5] [35].

• Target Vulnerability Identification: The distinctive genomic features of resistant clones represent potential therapeutic vulnerabilities. For example, ERBB2-amplified clones showed exceptional response to ERBB2-targeted therapy (trastuzumab deruxtecan), resulting in durable remission in one documented case [31] [34] [33].

• Predictive Biomarker Development: Clone-specific genomic features could serve as predictive biomarkers for treatment selection, enabling more personalized therapeutic approaches based on the evolving genomic landscape of each patient's disease [32] [33].

Table 3: Distinctive Genomic Features of Drug-Resistant Clones Identified by CloneSeq-SV

Genomic Feature	Frequency in Resistant Clones	Potential Functional Significance	Therapeutic Implications
CCNE1 Amplification	Frequent	Cell cycle dysregulation	CDK2 inhibition potential
ERBB2 Amplification	Documented in case study	Enhanced proliferative signaling	Trastuzumab deruxtecan response
Whole-Genome Doubling	Common	Increased genomic instability	PARP inhibitor sensitivity potential
Chromothripsis	Frequent	Catabolic genomic restructuring	General genomic instability targeting
NOTCH3 Amplification	Observed	Altered developmental signaling	NOTCH pathway inhibition
RAB25 Amplification	Observed	Vesicular trafficking alteration	Pathway-specific targeting

Technical Validation and Performance Metrics

Analytical Validation

The CloneSeq-SV methodology has undergone rigorous technical validation to establish its reliability and accuracy for both research and potential clinical applications:

• Specificity Verification: Application of patient-specific probes to 'off-target' patients in which no detection was expected demonstrated the exceptional specificity of SV detection, with erroneous read support observed for only a single event across all patients [5].

• Quantification Accuracy: Tumor fraction estimates derived from truncal SVs showed high correlation with estimates from truncal TP53 mutations (R=0.95, P<10⁻¹⁰, Pearson correlation), validating the quantitative accuracy of SV-based monitoring [5].

• Sensitivity Assessment: With typical sequencing parameters (1,000× coverage, 100 mutations), the theoretical detection limit for CloneSeq-SV is approximately 1×10⁻⁵, with SV error rates falling well below this threshold for both duplex and uncorrected sequencing (1×10⁻⁷) [5].

Comparison with Alternative Approaches

CloneSeq-SV offers distinct advantages over existing methods for monitoring clonal dynamics in cancer:

• Superior to SNV-Based Approaches: The error rate for SVs is orders of magnitude lower than for SNVs, providing enhanced signal-to-noise ratio that enables more confident detection of low-frequency variants [5].

• Non-Invasive Advantage: Unlike tumor biopsies, which provide only a single snapshot of a specific anatomical site, CloneSeq-SV enables comprehensive monitoring of clonal dynamics through blood-based collection, capturing spatial and temporal heterogeneity [31] [32].

• Single-Cell Resolution: Traditional bulk sequencing methods average signals across diverse cellular populations, while CloneSeq-SV maintains single-cell resolution for initial clonal decomposition, enabling more precise phylogenetic reconstruction [5].

Research Implementation Guidelines

Protocol Optimization Considerations

Successful implementation of CloneSeq-SV requires careful attention to several technical considerations:

• Sample Quality Control: Ensure high-quality single-cell suspensions with minimal dissociation-induced stress responses, as cellular viability significantly impacts single-cell library quality [5] [16].

• Sequencing Depth Optimization: Balance coverage depth with cost considerations, as the DLP+ protocol is optimized for shallow sequencing (0.088× mean coverage) while maintaining variant detection sensitivity [5].

• Longitudinal Sampling Frequency: Establish regular intervals for blood collection throughout the therapeutic journey, from diagnosis through treatment and recurrence, to capture critical evolutionary transitions [5] [32].

Computational Infrastructure Requirements

The computational demands of CloneSeq-SV analysis necessitate substantial bioinformatics resources:

• Data Storage: scWGS data from thousands of cells per patient requires extensive storage capacity, with subsequent cfDNA sequencing adding substantial additional data volume [5].

• Processing Pipelines: Implementation of specialized algorithms including MEDICC2 for phylogenetic reconstruction and HMMclone for copy-number calling requires dedicated computational expertise [5].

• Visualization Tools: Custom visualization approaches are necessary to interpret complex evolutionary patterns and communicate findings effectively to multidisciplinary research teams [5] [36].

Future Research Directions

The development of CloneSeq-SV opens numerous avenues for future research advancement and methodological refinement:

• Expansion to Other Cancer Types: While initially developed for HGSOC, the core principles of CloneSeq-SV could be applied to other cancer types characterized by high genomic instability, such as triple-negative breast cancer, pancreatic ductal adenocarcinoma, and hepatocellular carcinoma [31] [32].

• Integration with Multi-Omics Approaches: Combining SV-based clonal tracking with transcriptomic, epigenetic, and proteomic analyses could provide deeper insights into the functional states and regulatory mechanisms of treatment-resistant clones [36] [16].

• Clinical Trial Integration: Implementation of CloneSeq-SV within adaptive clinical trial designs could enable real-time therapeutic adjustments based on evolving clonal dynamics, potentially improving patient outcomes through more personalized treatment approaches [5] [35].

• Automated Analysis Pipelines: Development of streamlined, automated bioinformatics pipelines would increase the accessibility of CloneSeq-SV to broader research communities, accelerating adoption and application [5].

CloneSeq-SV represents a significant methodological advancement in cancer single-cell analysis research, providing an unprecedentedly detailed view of clonal evolution under therapeutic pressure. By leveraging somatic structural variants as sensitive clonal markers, this approach enables researchers to decipher the complex evolutionary trajectories that underlie treatment resistance, offering new opportunities for therapeutic intervention and personalized treatment strategies. As the methodology continues to evolve and expand to new cancer types, it holds substantial promise for transforming our understanding and management of cancer evolution.

Clonal evolution is the driving force behind intra-tumor heterogeneity, therapy resistance, and cancer progression. For years, cancer research has been constrained by a fundamental disconnect: the ability to trace genetic lineages (genotype) separately from understanding the functional cellular states they produce (phenotype). Single-cell multi-omics technologies now bridge this divide by enabling simultaneous measurement of multiple molecular layers from individual cells. Within this technological landscape, Genotyping of Transcriptomes for multiple targets and sample types (GoT-Multi) and SCClone represent complementary advanced frameworks specifically engineered to reconstruct clonal architecture and link it to transcriptional phenotypes. These tools are transforming our understanding of how distinct subclonal genotypes within the same tumor can either diverge into unique phenotypic states or paradoxically converge on similar transcriptional programs to mediate therapy resistance [39] [40]. This technical guide explores their methodologies, applications, and integration into cancer research and drug development pipelines.

Technology Deep Dive: GoT-Multi

Core Principles and Workflow

GoT-Multi is a high-throughput, single-cell multi-omics platform that enables the co-detection of multiple somatic genotypes alongside whole transcriptomes from the same cell. A significant advancement over its predecessor, GoT-Multi is compatible with formalin-fixed paraffin-embedded (FFPE) tissues, vastly expanding its applicability to vast archival clinical sample repositories [39] [40].

The methodology involves several sophisticated steps:

• Single-Cell Isolation and Barcoding: Single-cell suspensions from frozen or FFPE samples are partitioned into nanoliter-scale droplets, where each cell is lysed and the released nucleic acids are tagged with cell-specific barcodes.
• Whole Transcriptome Amplification: mRNA is reverse-transcribed and amplified to construct sequencing libraries that capture gene expression profiles.
• Multiplexed Genotyping: A key innovation of GoT-Multi is its ability to target a custom panel of 27 or more mutations simultaneously through an efficient multiplexed PCR approach, while maintaining compatibility with whole transcriptome analysis [39].
• Library Sequencing and Analysis: The barcoded libraries are sequenced, and an ensemble-based machine learning pipeline optimizes genotyping accuracy by integrating information across multiple mutations to confidently assign somatic genotypes to each cell [39].

GoT-Multi Experimental Protocol

The following table outlines the key steps for implementing GoT-Multi in a research setting:

Table 1: Detailed Experimental Protocol for GoT-Multi

Step	Description	Key Considerations
Sample Preparation	Process frozen or FFPE tissue into single-cell suspensions.	For FFPE samples, optimize de-crosslinking and digestion to maximize viability and nucleic acid recovery.
Panel Design	Design multiplex PCR primers for target mutations of interest.	Include positive and negative controls; validate panel sensitivity and specificity on control samples.
Library Preparation	Use the GoT-Multi workflow for single-cell partitioning, barcoding, cDNA synthesis, and targeted genotyping.	Use unique molecular identifiers (UMIs) to correct for amplification biases and enable accurate transcript quantification [41].
Sequencing	Sequence on Illumina platforms (or equivalent).	Aim for ~50,000 reads per cell for transcriptomes; ensure sufficient coverage for genotyping panels.
Data Processing	Demultiplex samples, align reads, and quantify gene expression and mutation counts.	Use the ensemble-based machine learning pipeline provided by the method for accurate single-cell genotyping [39].
Clonal Analysis	Cluster cells based on mutation profiles and correlate with transcriptional states.	Bioinformatic tools like Weighted-Nearest Neighbor analysis can help integrate multimodal data [42].

Application in Cancer Research

Applied to Richter transformation—an aggressive progression of chronic lymphocytic leukemia (CLL) to large B cell lymphoma—GoT-Multi revealed profound insights into clonal dynamics. The technology profiled tens of thousands of cells, reconstructing clonal architectures and linking them to distinct transcriptional programs. A key finding was that distinct subclonal genotypes, including those conferring therapy resistance, could converge on similar inflammatory transcriptional states. Other subclones independently activated proliferative programs and MYC-driven pathways, suggesting multiple convergent evolutionary paths to aggression [39] [40]. This demonstrates the power of GoT-Multi to uncover non-genetic resistance mechanisms that would be invisible to DNA sequencing alone.

Technology Deep Dive: SCClone

Core Principles and Workflow

While GoT-Multi integrates transcriptomes with targeted genotyping, SCClone is a computational method designed to accurately infer subclonal populations from single-cell DNA sequencing (scDNA-seq) data, which is particularly plagued by technical noise [43].

SCClone addresses critical technical artifacts in scDNA-seq:

• Allele Dropout (ADO): False-negative errors where heterozygous sites are miscalled as homozygous.
• False-Positive (FP) Errors: Falsely calling a homozygous site as heterozygous.
• Missing Data: Caused by non-uniform coverage and ADO, often exceeding 50% of sites in scDNA-seq data [43].

The algorithm employs a probability mixture model for binary mutation data and uses an Expectation-Maximization (EM) algorithm to directly learn subclonal mutational profiles and error rates from the observed data. This approach provides faster convergence compared to Markov Chain Monte Carlo (MCMC)-based methods. Furthermore, SCClone incorporates a novel model selection scheme based on inter-cluster variance to determine the optimal number of subclones present in a sample [43].

SCClone Analysis Protocol

The typical workflow for using SCClone involves the following steps:

Table 2: Detailed Analysis Protocol for SCClone

Step	Description	Key Considerations
Input Data	Prepare a binary Genotype Matrix (GTM) of N cells by M genomic loci.	Data is typically derived from scDNA-seq variant calling pipelines. States are: presence (1), absence (0), or unobserved (NA).
Data Preprocessing	Perform quality control to filter low-quality cells and mutations.	Remove cells with extremely high missing data rates or mutation counts inconsistent with the population.
Model Initialization	Initialize parameters for the probability mixture model, including putative number of subclones K.	The number of subclones K can be explored over a range; the model selection will help identify the optimum.
EM Algorithm Execution	Run the EM algorithm to cluster cells into subclones and estimate FP/FN error rates.	The E-step calculates the probability of each cell belonging to each subclone; the M-step updates subclone genotypes and error parameters.
Model Selection	Use the inter-cluster variance criterion to select the optimal number of subclones K.	This step prevents overfitting and ensures the model reflects the true biological complexity.
Output & Validation	Generate subclone assignments for each cell and the representative genotype for each subclone.	Validate results with orthogonal methods if possible (e.g., fluorescence in situ hybridization, bulk sequencing).

Performance and Applications

Extensive evaluations on simulated and real datasets demonstrate that SCClone achieves superior performance in inferring clonal composition compared to other state-of-the-art methods, particularly on data with high rates of false negatives and other technical noise [43]. By providing a robust reconstruction of clonal architecture from error-prone scDNA-seq data, SCClone establishes a reliable genetic foundation upon which other omics layers can be integrated.

The Scientist's Toolkit: Essential Research Reagents and Materials

Successfully implementing these multi-omics approaches requires a suite of specialized reagents and platforms. The following table details key components for building a single-cell multi-omics workflow.

Table 3: Key Research Reagent Solutions for Single-Cell Multi-Omics

Reagent/Material	Function	Example Use Case
FFPE Tissue Sections	Archival clinical samples for genomic analysis.	GoT-Multi enables genotyping and transcriptomics from these widely available but challenging samples [39].
Single-Cell Partitioning & Barcoding Kit	Creates nanoliter-scale droplets to isolate single cells and label their nucleic acids with cell barcodes.	10X Genomics Chromium Next GEM Chip Kits are widely used for high-throughput single-cell library preparation [44] [41].
Single-Cell Multiome ATAC + Gene Expression Kit	Allows simultaneous assay of gene expression (RNA-seq) and chromatin accessibility (ATAC-seq) from the same nucleus.	Used in studies like the hepatoblastoma analysis to link epigenetics and transcription [45].
Unique Molecular Identifiers (UMIs)	Short random nucleotide sequences that tag individual molecules before PCR amplification to correct for amplification bias and enable accurate quantification.	Critical for counting transcript copies in scRNA-seq and mitigating errors in scDNA-seq [41].
Multiplex PCR Primer Panels	Custom-designed primers to amplify specific genomic loci of interest (e.g., known driver mutations).	Core to the GoT-Multi genotyping step, allowing parallel detection of dozens of mutations [39].
Cell Hashing Antibodies	Antibodies conjugated to sample-specific barcodes that label cells from different samples, allowing sample multiplexing.	Enables pooling of samples before single-cell processing, reducing costs and batch effects [42] [41].

Visualizing Workflows and Logical Relationships

To effectively capture the logical structure and experimental flow of these integrated analyses, the following diagrams were created using Graphviz DOT language.

GoT-Multi Workflow Integration

The diagram below illustrates the integrated workflow of the GoT-Multi technology, from sample input to biological insight.

SCClone Computational Pipeline

The diagram below outlines the computational steps of the SCClone algorithm for inferring subclones from noisy single-cell DNA sequencing data.

Discussion and Future Perspectives in Cancer Research

The integration of technologies like GoT-Multi and SCClone is pivotal for advancing a unified understanding of clonal evolution. GoT-Multi directly links genotype to transcriptional phenotype, revealing mechanisms of resistance and progression. SCClone provides a robust foundation by accurately deciphering the complex clonal architecture from genetically noisy data. Together, they enable researchers to ask and answer previously intractable questions: Do genetically distinct subclones occupy unique niches in the tumor microenvironment? How does cellular plasticity contribute to relapse?

Future developments will likely focus on increasing the scalability and multiplexing capabilities of these platforms, integrating additional omics layers such as proteomics (via CITE-seq) [42] and chromatin accessibility (ATAC-seq) [45] [44], and improving computational methods to reconstruct more complex evolutionary lineages. As these tools become more accessible, they will undoubtedly reshape our strategies for early cancer detection, monitoring of minimal residual disease, and the design of combination therapies that target both the genetic drivers and the phenotypic vulnerabilities of resistant subclones.

The management of cancer is increasingly moving towards precision medicine, guided by a deeper understanding of intratumoral heterogeneity and clonal evolution. While single-cell DNA sequencing reveals complex mutational histories and branching phylogenetic patterns in cancers like acute myeloid leukemia (AML) [46] [47], its routine clinical application for monitoring remains challenging. Analysis of cell-free DNA (cfDNA), particularly the tumor-derived component (circulating tumor DNA or ctDNA), has emerged as a powerful, non-invasive liquid biopsy tool for tracking these clonal dynamics in real-time [48] [49]. This whitepaper details the clinical translation of cfDNA analysis for monitoring minimal residual disease (MRD) and therapy response, framing it within the critical context of clonal evolution studies. MRD refers to the residual cancer cells that persist after treatment at levels undetectable by conventional methods, serving as the primary reservoir for eventual relapse [50] [51]. The ability to detect MRD and monitor therapeutic efficacy through cfDNA analysis provides an unprecedented opportunity to guide treatment decisions, identify emerging resistance, and ultimately improve patient outcomes [52].

Clinical Utility of cfDNA in MRD Detection

Prognostic Value and Clinical Impact

The presence of ctDNA post-treatment is a robust biomarker of residual disease and predicts future recurrence with high accuracy. A recent meta-analysis of 95 studies demonstrated that a positive MRD test result confers an average odds ratio (OR) for relapse/recurrence of 3.5 in hematological cancers and 9.1 in solid cancers compared to patients with negative MRD tests [51]. This quantitative relationship between ctDNA detection and clinical outcomes underscores its prognostic power.

The clinical applications of cfDNA-based MRD monitoring are multifaceted:

• Early Relapse Detection: cfDNA analysis can identify molecular relapse months before clinical or radiographic recurrence, providing a critical window for therapeutic intervention [52].
• Treatment Guidance: MRD status can inform decisions on treatment escalation, de-escalation, or modification. For instance, in chronic myeloid leukemia (CML), loss of major molecular response triggers re-initiation of tyrosine kinase inhibitor therapy [51].
• Assessment of Heterogeneity: As a liquid biopsy, cfDNA captures tumor heterogeneity better than a single-site tissue biopsy, providing a more comprehensive snapshot of the overall disease burden and clonal architecture [49].

Clinical Validity Across Cancer Types

Table 1: Performance of cfDNA-Based MRD Testing Across Cancers

Cancer Type	Typical Assay	Key Molecular Target(s)	Reported Positive Predictive Value (PPV)	Clinical Context
Non-Small Cell Lung Cancer (NSCLC)	Targeted NGS	EGFR, ALK, ROS1, etc.	Varies by assay	Management of locally advanced (stage IIIb), recurrent, or metastatic disease when tissue is insufficient [53]
Metastatic Breast Cancer	PCR or NGS	PIK3CA, AKT1, PTEN, ESR1	Not fully established	Identify candidates for alpelisib, capivasertib plus fulvestrant, or elacestrant therapy [53]
Metastatic Prostate Cancer	NGS	BRCA1/2, other HRR genes	Not fully established	Identify candidates for PARP inhibitors or PD-1 inhibitors when tissue is insufficient [53]
Acute Myeloid Leukemia (AML)	NGS, dPCR, MPFC	Mutations in NPM1, FLT3, IDH1/2, etc.	<60% [51]	Assessment during/after remission induction; pre-transplant
Colorectal Cancer	NGS	KRAS, APC, TP53, PIK3CA	Varies by assay	Post-surgical monitoring; detection of recurrence [54] [51]

Technical Methodologies for cfDNA Analysis

Core Workflow and Experimental Protocol

The standard end-to-end workflow for cfDNA-based MRD analysis involves several critical steps, each requiring rigorous optimization.

Sample Collection and Processing:

• Blood Draw: Collect 10-20 mL of peripheral blood into cell-stabilizing tubes (e.g., Streck Cell-Free DNA BCT or PAXgene Blood ccfDNA tubes) to prevent genomic DNA contamination from lysed white blood cells.
• Plasma Separation: Centrifuge blood within a few hours of collection (e.g., 1600 × g for 20 min at 4°C) to separate plasma from cellular components. Transfer the supernatant and perform a second, high-speed centrifugation (e.g., 16,000 × g for 10 min) to remove any remaining cells and debris.
• cfDNA Extraction: Isolate cfDNA from plasma using commercial silica-membrane column or magnetic bead-based kits (e.g., QIAamp Circulating Nucleic Acid Kit, MagMAX Cell-Free DNA Isolation Kit). Elute in a low-volume buffer (e.g., 20-50 µL). Quantify yield using fluorometry (e.g., Qubit dsDNA HS Assay).

cfDNA Analysis - Key Methodologies:

• Tumor-Informed Assays: Considered the most sensitive approach for MRD detection, especially in early-stage disease [52].
  • Protocol: First, perform whole-exome or comprehensive genomic profiling of the primary tumor tissue to identify patient-specific somatic mutations (SNVs, indels). Then, design a custom, patient-specific panel (e.g., using multiplex PCR or hybrid capture) to track these mutations in plasma cfDNA.
  • Advantage: High sensitivity and specificity by focusing on mutations confirmed in the tumor.
• Tumor-Agnostic Assays: Do not require prior tissue sequencing.
  • Protocol: Analyze plasma cfDNA using fixed panels targeting recurrent mutations (e.g., in genes like KRAS, EGFR, PIK3CA) or epigenetic signatures like methylation patterns.
  • Advantage: Faster turnaround and applicable when tumor tissue is unavailable.
• ddPCR for Targeted Detection:
  • Protocol: Design TaqMan assays for a specific, known mutation. Partition the extracted cfDNA sample into thousands of nanoliter-sized droplets. Perform endpoint PCR amplification and analyze droplets for fluorescence to absolutely quantify the number of mutant and wild-type DNA molecules.
  • Application: Ideal for monitoring known, recurrent mutations (e.g., ESR1 mutations in breast cancer).
• Next-Generation Sequencing (NGS):
  • Protocol: Prepare sequencing libraries from cfDNA. For MRD, use either hybrid capture-based methods (enrich for large genomic regions) or amplicon-based methods (e.g., Safe-SeqS, TAm-Seq) that are highly efficient for low-input DNA. Sequence to very high depth (>50,000x coverage) to detect variants at allele frequencies as low as 0.01%.
  • Application: Provides a broad view of multiple mutations simultaneously, capturing more clonal diversity.

The following diagram illustrates the core workflow and the two primary assay strategies for cfDNA-based MRD detection.

Analytical Considerations and Sensitivity

The limit of detection (LOD) is a critical parameter for MRD assays. Tumor-informed NGS assays can achieve a sensitivity of up to (10^{-6}) (detecting one mutant molecule in a background of one million wild-type molecules), which is superior to tumor-agnostic approaches [50] [52]. Key factors influencing sensitivity and performance include:

• cfDNA Input: The quantity and quality of extracted cfDNA directly impact assay sensitivity. Low cfDNA yields from early-stage disease patients can be a major challenge [54].
• Sequencing Depth: Ultra-deep sequencing (>50,000x coverage) is required to confidently identify low-frequency variants.
• Variant Calling: Sophisticated bioinformatics pipelines are essential to distinguish true somatic mutations from technical artifacts introduced during amplification and sequencing. Error-suppression methods (e.g., unique molecular identifiers - UMIs) are routinely employed.
• Clonal Hematopoiesis: Somatic mutations originating from age-related hematopoietic clones can be detected in cfDNA and confound interpretation, necessitating careful variant filtering or complementary testing [54].

Table 2: Comparison of Key cfDNA Analysis Technologies for MRD

Technology	Typical Sensitivity	Throughput	Key Advantages	Key Limitations
Droplet Digital PCR (ddPCR)	0.01% - 0.1%	Medium	Absolute quantification; high sensitivity for known targets; low cost per assay.	Limited to 1-3 targets per reaction; requires a priori knowledge of mutation.
Tumor-Informed NGS	(10^{-5}) - (10^{-6})	Low to High	Ultra-sensitive; tracks multiple patient-specific mutations; captures heterogeneity.	Requires tumor tissue; longer turnaround time; higher cost; complex data analysis.
Tumor-Agnostic NGS	0.1% - 1%	High	No tissue required; faster; standardized panel.	Lower sensitivity, especially in early-stage disease; may miss clonal variants.
Methylation-Based NGS	<0.1%	High	Tissue-of-origin mapping; high specificity for cancer signal.	Complex assay development and data analysis; evolving standards.

Table 3: Key Research Reagent Solutions for cfDNA-Based MRD Studies

Reagent / Material	Function	Example Products / Assays
Cell-Free DNA Collection Tubes	Stabilizes blood cells to prevent lysis and preserve the native cfDNA profile for up to several days.	Streck Cell-Free DNA BCT, PAXgene Blood ccfDNA Tube
cfDNA Extraction Kits	Isolate high-purity, short-fragment cfDNA from plasma with high recovery and minimal contamination.	QIAamp Circulating Nucleic Acid Kit (Qiagen), MagMAX Cell-Free DNA Isolation Kit (Thermo Fisher), cfDNA/cfRNA Preserve Kit (Norgen Biotek)
Library Preparation Kits	Prepare sequencing libraries from low-input, fragmented cfDNA. Must be compatible with UMIs.	KAPA HyperPrep Kit (Roche), ThruPLEX Plasma-seq Kit (Takara Bio), AVENIO cfDNA Library Prep Kit (Roche)
Target Enrichment Panels	Enrich for cancer-specific genomic regions via hybrid capture or multiplex PCR.	AVENIO ctDNA Analysis Kits (Roche), Signatera (Natera), Guardant Reveal (Guardant Health)
ddPCR Assays	Pre-designed or custom assays for absolute quantification of specific mutations.	Bio-Rad ddPCR Mutation Detection Assays, Thermo Fisher QuantStudio Absolute Q Digital PCR Assays
Unique Molecular Identifiers (UMIs)	Short random nucleotide sequences added to each DNA molecule pre-amplification to tag and correct for PCR and sequencing errors.	Integrated in various library prep kits (e.g., from Roche, Takara Bio, Bio-Rad)

Interpreting cfDNA Data in the Context of Clonal Evolution

The true power of cfDNA analysis lies not just in detecting MRD, but in interpreting the data to understand the underlying clonal dynamics. Single-cell sequencing studies in AML have revealed that tumors are composed of multiple subclones with linear and branching evolutionary patterns [46] [47]. cfDNA profiling reflects this complexity.

• Variant Allele Frequency (VAF) Tracking: The changing VAF of different mutations in serial cfDNA samples can reveal clonal selection pressures. The emergence of a subclone harboring a specific mutation (e.g., an ESR1 mutation in breast cancer on aromatase inhibitor therapy) under therapeutic pressure is a classic example of clonal evolution driving resistance [53] [48].
• Discordant Findings: cfDNA analysis may detect mutations not found in the initial tumor biopsy due to spatial heterogeneity or clonal evolution post-treatment. This necessitates a flexible diagnostic approach.
• Therapeutic Resistance: Monitoring the rise of mutations known to confer resistance (e.g., KRAS mutations in colorectal cancer patients on anti-EGFR therapy) allows for early intervention and therapy switching before clinical progression is evident [54].

The following diagram illustrates how clonal architecture and therapy shape the cfDNA profile, providing insights into tumor evolution and resistance.

Current Challenges and Future Directions

Despite its promise, the clinical implementation of cfDNA for MRD faces several hurdles.

• Sensitivity in Early-Stage Disease: The very low tumor fraction in early-stage cancers post-resection poses a significant challenge, as the ctDNA concentration can fall below the LOD of even the most sensitive assays [54] [52].
• Standardization and Validation: Lack of standardization across pre-analytical (blood collection, processing), analytical (assay protocols), and post-analytical (bioinformatics, reporting) steps hinders widespread adoption and comparison between studies [49].
• Clinical Utility Trials: While retrospective data strongly links ctDNA presence with poor outcomes, prospective randomized trials demonstrating that acting on MRD results improves survival are still needed for many cancer types to establish definitive clinical utility [52] [51].
• Cost and Accessibility: The sophisticated technology and bioinformatics expertise required, particularly for tumor-informed assays, currently limit broad accessibility.

Future directions focus on overcoming these challenges:

• Novel Biosources: Investigating peritoneal fluid, stool, and urine as alternative sources of cfDNA that may have higher local tumor DNA concentration [54].
• Multi-Omics Integration: Combining mutation analysis with epigenetic markers (methylation, fragmentomics) to improve sensitivity and specificity for cancer detection and tissue-of-origin assignment [54] [49].
• Ultra-Sensitive Assays: Continued development of new technologies and chemistry to push the LOD even lower.
• Liquid Biopsy in Clinical Trials: Increasing use of cfDNA analysis as a biomarker in clinical trials to stratify patients, monitor response, and understand resistance mechanisms, accelerating drug development [49].

Navigating Technical Challenges: From Data Noise to Biological Interpretation

Single-cell RNA sequencing (scRNA-seq) has revolutionized our ability to study cellular heterogeneity, particularly in complex biological systems like cancer. However, several technical challenges impede the full exploitation of this technology in deciphering clonal evolution and intratumoral heterogeneity. This technical guide addresses three major hurdles—allelic imbalance analysis, transcriptional drop-outs, and RNA editing detection—within the context of cancer single-cell analysis, providing researchers with current methodologies and analytical frameworks to overcome these limitations.

The prevalence of scRNA-seq in research settings is growing rapidly, with the market projected to expand from US$1.63 billion in 2024 to US$6.65 billion by 2034, reflecting a compound annual growth rate of 15.05% [55]. This growth is largely driven by the technology's critical applications in oncology, which accounted for approximately 42% of the market share in 2024 [55]. As research into clonal evolution intensifies, overcoming technical challenges becomes paramount for accurate biological interpretation.

Allelic Imbalance Analysis in Single Cells

Biological Significance and Technical Challenges

Allele-specific expression (ASE) analysis provides powerful insights into cis-regulatory mechanisms in diploid organisms, revealing how genetic and epigenetic variations influence the exclusive or preferential expression of a particular allele [56]. In cancer research, ASE can uncover subclonal regulatory heterogeneity and inform our understanding of how specific alleles contribute to clonal expansion and dominance.

Current ASE analysis pipelines face notable limitations, including a lack of end-to-end solutions, restricted options for multi-omics integration, and insufficient support for single-cell sequencing technologies [56]. A systematic review of 26 cutting-edge ASE pipelines revealed that most fail to automate preprocessing, integrate multi-omic data, and support high-throughput single-cell sequencing [56]. These gaps significantly impact their utility in clonal evolution studies where multi-omic integration is essential.

DAESC: A Specialized Statistical Framework

The DAESC (Differential Allelic Expression using Single-Cell data) method has been developed specifically for differential ASE analysis using scRNA-seq data from multiple individuals [57]. This framework addresses two critical challenges in cross-individual single-cell studies: haplotype switching and sample repeat structure.

Table 1: Key Features of DAESC Framework

Feature	DAESC-BB	DAESC-Mix
Statistical Model	Beta-binomial with individual-specific random effects	Beta-binomial mixture model with implicit haplotype phasing
Sample Size Requirement	Applicable regardless of sample size	Requires larger sample sizes (N ≥ 20)
Haplotype Switching Handling	No implicit phasing	Accounts for haplotype switching through latent variables
Best Use Cases	General differential ASE analysis	Scenarios where expression-increasing allele can be on either haplotype

DAESC employs a beta-binomial regression model that can test differential ASE against any independent variable, including cell type, continuous developmental trajectories, genotype, or disease status [57]. The method accounts for non-independence between cells from the same individual through random effects, addressing the sample repeat structure inherent to scRNA-seq data [57].

Experimental Protocol for Single-Cell ASE Analysis

A robust single-cell ASE analysis protocol involves:

• Sample Preparation: Process 105+ individuals when possible to ensure sufficient statistical power for DAESC-Mix applications [57].
• Library Construction: Use platform-specific chemistries (e.g., 10x Genomics Chromium GEM-X assays) to minimize allelic bias during amplification [55].
• Sequencing: Aim for adequate depth to cover heterozygous SNPs; typically 50,000 reads/cell provides good allele coverage.
• Variant Calling: Identify heterozygous transcribed SNPs (tSNPs) using matched DNA sequencing when available.
• ASE Quantification: Apply DAESC framework to test for differential ASE across conditions or cell states.
• Validation: Use orthogonal methods such as single-molecule RNA FISH for confirmation of key findings.

Simulation studies demonstrate that DAESC maintains robust type I error control and achieves high power for differential ASE detection, particularly in scenarios with low linkage disequilibrium between eQTLs and tSNPs [57].

Embracing and Overcoming Transcriptional Drop-Outs

The Nature and Impact of Drop-Outs

Drop-out events represent a fundamental characteristic of scRNA-seq data where genes expressed at low or moderate levels in one cell are not detected in another cell of the same type [58]. These events occur due to low mRNA quantities in individual cells, inefficient mRNA capture, and stochastic gene expression [58]. In cancer studies, drop-outs can obscure rare subclones and complicate trajectory analyses aimed at reconstructing clonal evolution.

The impact of drop-outs on downstream analyses is profound. Research shows that while cluster homogeneity (cells in a cluster being of the same type) is maintained under increasing dropout rates, cluster stability (cell pairs consistently being in the same cluster) decreases significantly [59]. This instability makes sub-populations within cell types increasingly difficult to identify because "similar cells are close to each other in space" assumption breaks down [59].

Co-Occurrence Clustering: Leveraging Drop-Out Patterns

Rather than treating drop-outs as noise to be eliminated, an alternative approach embraces drop-outs as useful signals by analyzing their patterns [58]. The co-occurrence clustering algorithm operates on binarized scRNA-seq data (zero vs. non-zero) and identifies cell populations based on coordinated absence of gene expression.

Table 2: Co-Occurrence Clustering Workflow

Step	Process	Outcome
1	Binarization of count matrix	Conversion of expression values to 0 (dropout) or 1 (expressed)
2	Gene-gene co-occurrence calculation	Identification of genes with similar dropout patterns across cells
3	Gene pathway identification	Clustering of co-occurring genes into pathway signatures
4	Pathway activity calculation	Percentage of detected genes in each pathway per cell
5	Cell-cell graph construction	Euclidean distances based on pathway activity representation
6	Community detection and cluster merging	Identification of cell clusters with distinct dropout patterns

This method has demonstrated effectiveness in identifying major cell types in Peripheral Blood Mononuclear Cells (PBMC), with the binary dropout pattern proving as informative as quantitative expression of highly variable genes for cell type identification [58].

Experimental Considerations for Dropout Management

When designing experiments where dropout events may impact conclusions:

• Cell Number: Sequence sufficient cells (typically 2-3× more than required) to compensate for information loss from dropouts.
• Platform Selection: Choose technologies with higher capture efficiency (e.g., SMART-seq4 for full-length transcript coverage) when studying rare cell populations.
• Spike-Ins: Use external RNA controls to quantify technical noise and distinguish it from biological zeros.
• Multi-Omic Integration: Combine with DNA-based single-cell assays to guide interpretation of missing expression values.

For clonal evolution studies specifically, implement cross-validation by comparing clustering results from both quantitative expression and dropout patterns to ensure identified subpopulations are robust to technical artifacts.

RNA Editing Detection in Single Cells

Biological Significance in Cancer Context

RNA editing, particularly adenosine-to-inosine (A-to-I) deamination, represents a crucial post-transcriptional modification that increases transcriptome diversity [60]. In cancer research, RNA editing profiles can distinguish cell types and states within tumors, providing insights into functional heterogeneity and clonal dynamics.

Single-cell studies of human brain cortex cells have revealed that RNA editing levels per cell show a bimodal distribution, distinguishing major brain cell types [60]. Unlike the unimodal distribution observed in bulk tissue, single-cell analysis reveals an "all or nothing" pattern where editing penetrance varies substantially between individual cells [60]. This heterogeneity likely exists in cancer cells as well and may contribute to phenotypic diversity within tumors.

Methodological Framework for Editing Detection

Accurate identification of RNA editing events requires careful experimental design and computational analysis:

Diagram 1: RNA Editing Detection Workflow. This diagram outlines the standard pipeline for identifying RNA editing events from sample collection to final verification.

The critical requirement for reliable RNA editing detection is obtaining matched transcriptome and DNA resequencing data from the same sample [61]. This approach enables distinction between true RNA editing events and genomic polymorphisms or sequencing errors.

Single-Cell RNA Editing Protocol

For single-cell RNA editing analysis in cancer samples:

• Sample Collection: Process fresh tumor samples to maintain RNA integrity; consider single-cell dissociation protocols that preserve neuronal RNA where applicable [61].
• Matched DNA-RNA Sequencing: Ideally, perform scRNA-seq on tumor cells and scDNA-seq on matched normal cells from the same patient.
• Library Preparation: Use whole-transcriptome amplification methods that minimize sequence-dependent bias (e.g., SMART-seq4) [58].
• Sequencing Depth: Aim for minimum coverage of 10× at candidate editing sites for reliable detection [60].
• Variant Calling: Use specialized tools like REDItools with parameters optimized for single-cell data [60].
• Quality Filtering: Exclude cells with fewer than 1 million uniquely aligned reads and mapping rates <70% [60].
• Validation: Employ Sanger sequencing of both RNA and matched DNA for confirmation of high-priority editing sites [61].

In application to brain cortex cells, this approach revealed that editing activity in recoding sites was higher in neurons than other cell types, with only a few sites in glutamate receptors edited in almost all neurons [60]. Similar cell-type-specific editing patterns likely exist in cancer ecosystems and may illuminate functional subpopulations.

Integrated Multi-Omic Approaches for Clonal Evolution

Multi-Omic Frameworks for CK-AML Analysis

Single-cell multi-omics approaches are revolutionizing our ability to dissect clonal evolution in cancers with complex karyotypes. In acute myeloid leukemia with complex karyotype (CK-AML), integrated analysis combining structural variant discovery, nucleosome occupancy profiling, transcriptomics, and immunophenotyping has revealed dynamic clonal evolution patterns [12].

The scNOVA-CITE framework couples single-cell nucleosome occupancy and genetic variation analysis (scNOVA) with cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq) [12]. This integration enables simultaneous assessment of genotype and phenotype in individual cells, revealing three distinct clonal evolution patterns in CK-AML: monoclonal growth, linear growth, and branched polyclonal growth [12].

Table 3: Clonal Evolution Patterns in Complex Karyotype AML

Evolution Pattern	Prevalence	Characteristics	Clinical Implications
Monoclonal Growth	2/8 cases	Single dominant subclone with minor deviations	Possibly more stable genome
Linear Growth	3/8 cases	Step-wise acquisition of structural variants	Gradual evolution
Branched Polyclonal Growth	3/8 cases	Multiple subclones with ongoing karyotype remodeling	Highest heterogeneity, potential for rapid adaptation

The Scientist's Toolkit

Diagram 2: Integrated Approach to Clonal Evolution. This diagram illustrates how multi-omic integration of allele-specific expression, dropout patterns, and RNA editing detection contributes to comprehensive clonal evolution insights.

Table 4: Essential Research Reagent Solutions for scRNA-seq Challenges

Reagent/Tool	Function	Application Examples
Chromium GEM-X Assays	Single-cell partitioning and barcoding	High-throughput scRNA-seq with reduced cost [55]
Tapestri Single-cell Multiomics Solution	Combined genomic, proteomic, and clonotypic analysis	Tracking tumor evolution in blood cancers [55]
REDItools	Variant calling for RNA editing detection	A-to-I editing identification in single cells [60]
DAESC Software	Differential allele-specific expression testing	Identifying context-specific cis-regulatory effects [57]
STRAND-Sc ScTRIP	Structural variant detection in single cells	Mapping complex chromosomal rearrangements in CK-AML [12]

Overcoming the technical challenges of allelic imbalance, transcriptional drop-outs, and RNA editing detection in scRNA-seq requires specialized methodologies and integrated approaches. The solutions presented in this guide—including DAESC for allele-specific expression analysis, co-occurrence clustering for dropout pattern utilization, and matched DNA-RNA sequencing for RNA editing detection—provide researchers with powerful strategies to extract more meaningful biological insights from single-cell data.

In cancer research, particularly studies of clonal evolution, multi-omic integration emerges as a critical theme. Approaches that combine genetic, transcriptional, and epigenetic information from the same single cells offer unprecedented resolution for mapping subclonal architecture and understanding tumor dynamics. As these methodologies continue to evolve, they will undoubtedly yield deeper insights into cancer biology and opportunities for targeted therapeutic intervention.

Clonal evolution is the fundamental process by which cancers progress, adapt, and develop therapy resistance. This evolution generates intratumor heterogeneity (ITH), where distinct subpopulations of cells with different genetic alterations coexist within the same tumor [62]. Traditional bulk sequencing approaches average signals across thousands of cells, masking this critical heterogeneity and obscuring rare but clinically significant subclones. Single-cell sequencing technologies have revolutionized cancer research by enabling the dissection of this complexity at unprecedented resolution [63] [62]. However, these technologies generate vast, multidimensional datasets that present substantial computational challenges. This technical guide examines how machine learning (ML) and advanced computational frameworks are addressing these challenges, specifically in genotyping and clonal inference, to illuminate cancer evolution and inform therapeutic strategies.

Machine Learning for Single-Cell Genotyping

The Genotyping Challenge in Single-Cell Data

Single-cell genotyping involves identifying somatic mutations—including single nucleotide variants (SNVs), insertions/deletions (indels), and copy number alterations—from sequencing data of individual cells. This process is complicated by technical artifacts from whole-genome amplification, such as allelic dropout and amplification bias, which lead to false negatives and uneven coverage [63]. Signal-to-noise ratios are lower than in bulk sequencing, requiring sophisticated computational methods to distinguish true biological variants from technical artifacts.

Ensemble Machine Learning for Optimized Genotyping

Ensemble-based machine learning pipelines represent the cutting edge in addressing genotyping inaccuracies. These methods combine multiple classifiers or algorithms to improve prediction accuracy and robustness over single-algorithm approaches.

The GoT-Multi (Genotyping of Transcriptomes for multiple targets and sample types) platform exemplifies this approach. It is a high-throughput, single-cell multi-omics method that co-detects multiple somatic genotypes and whole transcriptomes, even from formalin-fixed paraffin-embedded (FFPE) samples [64]. Its integrated machine learning pipeline leverages an ensemble of models to optimize genotype calling accuracy, effectively mitigating technical noise and enabling reliable detection of multiple mutations per cell.

Table 1: Key Machine Learning Approaches for Single-Cell Genotyping

Method/Platform	Core ML Approach	Input Data	Key Capabilities	Application Context
GoT-Multi [64]	Ensemble-based ML	scRNA-seq + multiplexed genotyping	Optimized genotyping from fresh & FFPE samples; links genotype to cell state	Therapy-resistant lymphoma
SCOOP [65]	XGBoost	scATAC-seq + WGS	Predicts cell of origin by modeling mutation density in chromatin bins	Pan-cancer cell of origin prediction
Foundation Models (e.g., scGPT) [66]	Transformer-based pretraining	Large-scale scRNA-seq datasets	Zero-shot cell annotation, perturbation prediction, multi-omic integration	Generalizable cell analysis and annotation

The following diagram illustrates the ensemble-based ML workflow for genotyping within the GoT-Multi framework:

Foundation Models for Generalized Cellular Analysis

A paradigm shift is underway with the emergence of single-cell foundation models (scFMs). These models, pretrained on massive datasets comprising millions of cells, learn universal representations of cellular state [66]. For instance, scGPT is a generative pretrained transformer model trained on over 33 million cells that demonstrates exceptional capability in zero-shot cell type annotation and perturbation response prediction [66]. While not exclusively designed for genotyping, these models provide a powerful foundational representation that can enhance downstream genotyping accuracy and integrate genotypic information with transcriptional and epigenetic states.

Computational Frameworks for Clonal Inference

From Genotypes to Clonal Architectures

Clonal inference involves reconstructing the evolutionary history and phylogenetic relationships between cells based on their somatic mutation profiles. In cancers with extensive chromosomal instability, this requires interpreting complex patterns of structural variants (SVs) and copy-number alterations (CNAs) alongside point mutations [12].

Multi-Omics Integration for Clonal Lineage Tracing

Single-cell multi-omics technologies enable the simultaneous measurement of genotype and phenotype, providing a powerful basis for tracing clonal lineages. The scNOVA-CITE framework couples single-cell analysis of structural variants (via Strand-seq) with transcriptome and surface protein measurements (via CITE-seq) [12]. This multi-layered data reveals how genetic subclones differ in their transcriptional programs, epigenetic states, and surface marker expression, providing a comprehensive view of functional heterogeneity.

Table 2: Computational Methods for Clonal Inference and Analysis

Method/Platform	Primary Function	Data Input	Inference Output	Identified Evolution Patterns
scNOVA-CITE [12]	Clonal evolution tracing	Strand-seq + CITE-seq	Subclonal architecture with linked phenotypes	Monoclonal, linear, and branched polyclonal
SCOPer [67]	B-cell clonal assignment	B-cell receptor sequences	Clonal families from VDJ recombination	Affinity maturation lineages
mPTP [67]	Phylogenetic clonal delimitation	B-cell receptor phylogenetic tree	Clonal families without reference genome	Clonal diversification rates

Application of these methods in complex karyotype acute myeloid leukemia (CK-AML) has revealed distinct modes of clonal evolution. Research has identified three primary patterns: 1) monoclonal growth, where a single dominant subclone is present; 2) linear evolution, characterized by step-wise acquisition of mutations; and 3) branched polyclonal evolution, where multiple subclones diverge and coexist, frequently associated with extensive karyotype remodeling and therapy resistance [12].

The following diagram illustrates the multi-omics workflow for clonal inference:

Phylogenetic Methods for Clonal Delimitation

Clonal inference also draws inspiration from phylogenetic species delimitation methods. The mPTP (multi-rate Poisson Tree Processes) model, originally designed for species delimitation, has been adapted to identify B-cell clonal families from antibody sequence data [67]. This method uses a phylogenetic tree of B-cell receptor sequences and models VDJ-recombination as a speciation-like event and somatic hypermutation as a within-clone diversification process. Its performance is competitive with specialized immunoinformatics tools like SCOPer, particularly for non-model organisms lacking reference genomes [67].

Successful implementation of ML-driven genotyping and clonal inference requires both wet-lab reagents and computational resources.

Table 3: Key Research Reagents and Computational Tools

Category/Name	Function/Purpose	Key Features/Applications
GoT-Multi [64]	Single-cell multi-omics genotyping	Links multiplexed genotyping with scRNA-seq; compatible with FFPE samples
CITE-seq [12]	Cellular indexing of transcriptomes and epitopes	Simultaneous measurement of transcriptome and surface protein expression
Strand-seq [12]	Haplotype-aware structural variant detection	Resolves complex chromosomal rearrangements and SVs in single cells
scGPT [66]	Foundation model for single-cell biology	Zero-shot cell annotation, perturbation modeling, multi-omic integration
DISCO/CZ CELLxGENE [66]	Data repositories and analysis platforms	Aggregate millions of single-cell datasets for federated analysis
SCOPer [67]	B-cell clonal assignment	Groups B-cell sequences into clonal families based on VDJ usage and junction similarity
NUC-Seq [62]	High-coverage single-cell genome sequencing	Achieves >90% physical coverage of single mammalian cell genomes for mutation detection

Experimental Protocols for Key Applications

Protocol: Linking Clonal Genotypes to Transcriptional States Using GoT-Multi

Purpose: To reconstruct clonal architecture and associate genetic subclones with distinct transcriptional programs in therapy-resistant cancers.

Steps:

• Sample Preparation: Process fresh frozen or FFPE samples into single-cell suspensions [64].
• Library Preparation: Perform GoT-Multi library prep, which combines:
  • Multiplexed PCR-based genotyping for 27+ known driver mutations.
  • scRNA-seq for whole transcriptome analysis [64].
• Sequencing: Use Illumina platforms for high-throughput sequencing.
• Computational Analysis:
  • Apply the ensemble ML pipeline for optimized genotyping of each cell.
  • Perform clonal deconvolution based on mutation co-occurrence patterns.
  • Conduct differential expression analysis between genetically defined subclones [64].

Application: In Richter transformation (progression of chronic lymphocytic leukemia to aggressive lymphoma), this protocol revealed that distinct subclonal genotypes, including those with therapy-resistant mutations, converged on a shared inflammatory transcriptional state, while other subclones exhibited enhanced proliferation and MYC activity [64].

Protocol: Dissecting Clonal Evolution in Complex Karyotype AML

Purpose: To characterize patterns of clonal evolution and intratumor heterogeneity in cancers with extreme chromosomal instability.

Steps:

• Multi-Omics Profiling:
  • Perform Strand-seq on 855+ single cells for haplotype-resolved structural variant detection.
  • Perform CITE-seq for simultaneous transcriptome and surface protein quantification [12].
• Variant Calling:
  • Use scTRIP to identify chromosomal alterations (translocations, inversions, aneuploidies).
  • Calculate structural variant burden and intrapatient karyotype heterogeneity [12].
• Clonal Reconstruction:
  • Infer clonal evolution patterns (monoclonal, linear, branched) based on shared structural variants.
  • Map phenotypic heterogeneity (transcriptomic, immunophenotypic) onto genetic subclones [12].
• Functional Validation:
  • Establish patient-derived xenografts (PDXs) for in vivo validation.
  • Perform ex vivo drug sensitivity testing to identify subclone-specific vulnerabilities [12].

Key Findings: This approach identified BCL-xL inhibition as a potential therapeutic strategy for targeting disease-driving leukemic stem cell subpopulations in CK-AML [12].

Machine learning has become indispensable for interpreting the complex datasets generated by single-cell genomics, transforming our ability to genotype individual cells and reconstruct clonal evolutionary trajectories in cancer. The integration of ensemble methods, multi-omics data, and foundation models is providing unprecedented insights into how tumors evolve, adapt, and resist therapy.

Future advancements will likely focus on several key areas: (1) improved spatial resolution through integration with spatial transcriptomics and proteomics; (2) enhanced temporal resolution through lineage tracing and longitudinal sampling; and (3) more interpretable and explainable AI models that can generate testable biological hypotheses. As these computational solutions mature, they will increasingly bridge the gap between cancer genomics and clinical application, enabling clonal tracking for disease monitoring and personalized therapeutic targeting.

The analysis of cell-free DNA (cfDNA) from liquid biopsies has emerged as a transformative, non-invasive tool for cancer monitoring, enabling applications from minimal residual disease (MRD) detection to therapy response assessment [5]. A central challenge in this field is the ultra-low abundance of circulating tumor DNA (ctDNA), which often exists at frequencies below the error rate of conventional next-generation sequencing (NGS) platforms [68]. This limitation creates a fundamental signal-to-noise problem where true somatic variants become indistinguishable from sequencing artifacts.

Error-corrected sequencing technologies, particularly duplex sequencing, have revolutionized cfDNA analysis by enabling the detection of mutations with frequencies as low as 1 in 10⁻⁷ [68]. This technical advancement provides the sensitivity required to study clonal evolution in cancer patients through liquid biopsies, offering unprecedented insights into tumor dynamics and drug resistance mechanisms that were previously inaccessible without invasive tissue sampling [5] [69]. This guide details the experimental and computational frameworks for implementing duplex sequencing to optimize signal-to-noise in cfDNA detection, specifically within the context of clonal evolution research.

Principles of Duplex Sequencing

Duplex sequencing is an error-correction methodology that achieves exceptional accuracy by independently tagging and sequencing both strands of each original DNA molecule. True mutations are only called when the variant appears at the same position in both complementary strands; errors occurring during PCR amplification or sequencing that affect only one strand are computationally filtered out [68] [70].

The power of this approach is quantified by its dramatically reduced error rate. While conventional NGS exhibits error rates around 10⁻³ to 10⁻⁴, duplex sequencing can achieve error rates as low as 7.7×10⁻⁸ [68]. This reduction in background noise enables the confident identification of extremely rare variants in complex biological samples like cfDNA.

Table 1: Key Performance Metrics of Sequencing Modalities for cfDNA Analysis

Sequencing Modality	Typical Error Rate	Effective VAF Detection Limit	Key Applications in cfDNA
Conventional NGS	10⁻³ to 10⁻⁴	~1%	Tumor genotyping at high variant allele frequency (VAF)
Simplex Sequencing (with UMIs)	~10⁻⁵	~0.1%	ctDNA detection in advanced cancers
Duplex Sequencing	7.7×10⁻⁸ [68]	<0.0001%	MRD, relapse monitoring, clonal evolution studies

Experimental Protocol: Duplex Sequencing Workflow for cfDNA

Sample Preparation and Library Construction

The initial steps focus on preserving strand-origin information for subsequent error correction:

• cfDNA Extraction and Quantification: Extract cfDNA from plasma using validated circulating nucleic acid kits. Precisely quantify using fluorometric methods to input 10-100 ng of cfDNA into the library preparation protocol [5].
• Molecular Barcoding (Adapter Ligation): Ligate double-stranded adapters containing random unique molecular identifiers (UMIs) to both ends of each cfDNA fragment. These UMIs uniquely tag each original DNA molecule, enabling bioinformatic consensus building [68].
• Target Enrichment (Optional): For focused analyses, use patient-bespoke hybrid capture probes targeting specific genomic regions. Design 60-base-pair flanking sequences on either side of breakpoints for structural variants (SVs) or point mutations of interest [5]. Whole-genome approaches are also feasible with ultradeep sequencing [68].

Sequencing and Data Analysis

• High-Throughput Sequencing: Sequence the prepared libraries on a high-output platform. Ultima Genomics mnSBS is noted for its cost-effectiveness in achieving the deep coverage required (~120x for whole-genome duplex sequencing) [68].
• Bioinformatic Processing:
  • Consensus Sequence Generation: Group reads derived from the same original DNA molecule using their UMIs. Generate a single consensus sequence for each of the two complementary strands.
  • Duplex Consensus Calling: Identify true variants by requiring the same mutation to be present in the consensus sequences of both the original template strand and its complementary strand. Discard mutations appearing in only one strand as technical artifacts [68] [70].
  • Variant Calling and Annotation: Call high-confidence somatic variants and annotate them using standard tools. For clonal evolution studies, specific structural variants can serve as highly specific endogenous markers for tracking individual clones [5].

Diagram 1: Duplex sequencing workflow for cfDNA.

Application: Tracking Clonal Evolution in Cancer

Studying clonal evolution—how distinct subpopulations of cancer cells change over time under therapeutic pressure—is critical for understanding drug resistance. Duplex sequencing of cfDNA enables high-resolution, non-invasive tracking of these dynamics.

In high-grade serous ovarian cancer (HGSOC), the CloneSeq-SV method combines single-cell whole-genome sequencing of pretreatment tumor tissue with duplex sequencing of cfDNA to track clone-specific structural variants over time [5]. This approach revealed that drug-resistant clones frequently pre-exist at diagnosis and are selectively enriched by therapy, leading to reduced clonal complexity at relapse [5]. These clones often possess distinctive genomic features such as chromothripsis, whole-genome doubling, and amplifications of oncogenes like CCNE1 and MYC [5].

Clone-specific SVs offer a particular advantage for tracking because their unique breakpoint sequences are highly specific and resistant to sequencing errors, resulting in a superior signal-to-noise ratio compared to single nucleotide variants (SNVs) [5]. SVs can achieve error rates below 1×10⁻⁷ even in uncorrected sequencing, facilitating confident detection of rare clones from a single event [5].

Table 2: Advantages of Structural Variants (SVs) vs. Single Nucleotide Variants (SNVs) for Clonal Tracking in cfDNA

Characteristic	Structural Variants (SVs)	Single Nucleotide Variants (SNVs)
Specificity	Extremely high (unique breakpoint junctions)	Lower (must be distinguished from background SNVs)
Error Rate in cfDNA Assays	<1×10⁻⁷ (even uncorrected) [5]	~6.7×10⁻⁶ (with duplex sequencing) [5]
Typical Abundance per Clone	Few, but often clonal markers	Numerous
Per-Cell Copy Number	Can be high (e.g., in amplifications), enhancing signal	Typically one or two
Utility for Clone-Specific Tracking	Excellent (often clone-specific)	Good (requires phylogenetic deconvolution)

Diagram 2: Clonal evolution tracking with cfDNA.

The Scientist's Toolkit: Essential Reagents and Technologies

Table 3: Research Reagent Solutions for Duplex Sequencing of cfDNA

Reagent / Technology	Function	Example Application / Note
Molecular Barcoded Adapters	Uniquely tags each original DNA molecule for error correction	Essential for distinguishing PCR duplicates from original molecules [68]
Hybrid Capture Probes	Enriches for specific genomic regions of interest	Patient-bespoke panels targeting clone-specific SVs; design with 60-bp flanking breakpoints [5]
Ultima Genomics mnSBS Platform	Low-cost, high-throughput whole-genome sequencing	Enables deep sequencing (~120x) for genome-wide mutation integration [68]
cfDNA Extraction Kits	Isolation of high-quality circulating nucleic acids from plasma	Maximize yield and integrity of low-abundance cfDNA
In Vitro MicroFlow Kit	Flow cytometry-based assessment of cytotoxicity and micronucleus formation	Complementary cytogenetic endpoint analysis [71]
Metabolically Competent HepaRG Cells	Human-relevant in vitro model for mutagenicity assessment	Provides endogenous xenobiotic metabolism; useful for genotoxicity studies [71]

Regulatory Context and Future Directions

Error-corrected sequencing (ECS) technologies like duplex sequencing are gaining formal recognition for regulatory safety assessment. An International Workshops on Genotoxicity Testing (IWGT) expert workgroup has endorsed ECS for in vivo mutagenicity assessment and recommended its inclusion in future OECD test guidelines [70]. The working group confirmed that ECS results are concordant with validated transgenic rodent (TGR) assays and can be incorporated into standard 28-day repeat-dose toxicity studies, advancing the 3Rs (Replacement, Reduction, and Refinement) principles in toxicology [70].

The future of duplex sequencing in cancer research and drug development lies in its integration with other multimodal data. Combining clonal tracking via cfDNA with matched single-cell RNA sequencing data can reveal pre-existing clone-specific transcriptional states—such as upregulation of epithelial-to-mesenchymal transition or VEGF pathways—linked to drug resistance [5]. As these technologies become more accessible, they will enable evolution-informed adaptive treatment regimens to combat therapeutic resistance in cancer.

The progression of cancer is a dynamic process of Darwinian evolution, where tumors consist of multiple cellular populations with distinct genotypes, known as clones, that undergo phylogenetic diversification driven by selective pressures [4]. Complex karyotypes—characterized by intricate rearrangements such as translocations, chromothripsis, and aneuploidy—serve as structural fingerprints of these evolutionary processes. Resolving these complex chromosomal alterations at base-level resolution provides critical insights into tumorigenesis, therapeutic resistance, and metastatic progression [72]. Traditional karyotyping techniques, such as G-banding and fluorescent in situ hybridization (FISH), have limited resolution (~5 Mb), making them insufficient for accurately identifying complex structural variants (SVs) in derivative or marker chromosomes [72]. The emergence of long-read sequencing technologies and advanced computational frameworks now enables researchers to reconstruct cancer genome karyotypes with unprecedented resolution, revealing the complex patterns of clonal evolution that underlie cancer progression and relapse [73] [72].

Structural variants (SVs), defined as genomic rearrangements longer than 50 bp, include deletions, duplications, inversions, insertions, and translocations, and account for more varying base pairs in the human genome than any other class of sequence variants [73]. In cancer genomics, SVs are not acquired as independent events but rather manifest in specific patterns that reflect underlying mutational processes: chromothripsis (chromosomal "shattering" and random reassembly), chromoplexy (interchromosomal translocations), and breakage-fusion-bridge cycles [73] [72]. These complex rearrangement patterns drive tumor evolution by disrupting tumor suppressor genes, activating oncogenes, and generating homogeneously staining regions (HSRs) and double minutes (DMs) that facilitate oncogene amplification [72]. Accurate resolution of these SVs is therefore essential for understanding the clonal dynamics that shape cancer genomes.

Technical Foundations: From Short-Read to Long-Read Sequencing Technologies

The Limitations of Short-Read Sequencing for SV Detection

Next-generation sequencing (NGS) technologies, particularly short-read platforms (e.g., Illumina), have been instrumental in advancing cancer genomics but face significant limitations in resolving complex SVs [73]. While paired-end read strategies have improved the detection of some rearrangements, short read lengths (100-500 bp) make it challenging to map repetitive regions, segmental duplications, and complex rearrangement breakpoints [73] [74]. The fundamental issue lies in the mappability of short reads, which decreases dramatically in regions with repetitive elements, precisely where SVs are known to cluster [73]. This technological gap has led to the underdetection of complex SVs, leaving a significant portion of the cancer genome's mutational landscape unexplored.

Advancements in Long-Read Sequencing Platforms

Third-generation sequencing (TGS) technologies, including Oxford Nanopore Technologies (ONT) and Pacific Biosciences (PacBio), have emerged as powerful tools for SV detection due to their ability to generate reads spanning several kilobases [73] [74]. Nanopore sequencing, in particular, offers the unique advantage of sequencing "native" long DNA molecules of virtually unlimited length (typical range 1-100 Kb), enabling the traversal of repetitive regions and the resolution of complex SVs that were previously intractable [73]. The enhanced phasing capability of long reads allows researchers to determine haplotype-specific SVs, providing crucial information about allele-specific events in cancer genomes [72]. While long-read technologies have historically faced challenges with higher error rates compared to short-read platforms, continuous improvements in accuracy and read length have positioned them as indispensable tools for comprehensive karyotype resolution [74].

Table 1: Comparison of Sequencing Technologies for SV Detection and Karyotype Resolution

Technology	Read Length	Advantages for SV Detection	Limitations for SV Detection
Short-Read (Illumina)	100-500 bp	High base-level accuracy, low cost per base	Limited ability to resolve repetitive regions, incomplete SV breakpoint resolution
Oxford Nanopore	1-100+ Kb	Ultra-long reads, direct detection of base modifications	Higher error rate, requires more DNA input
PacBio	10-100 Kb	High accuracy in circular consensus mode, excellent for phasing	Lower throughput, higher cost per sample
Linked-Reads (10X Genomics)	100-500 bp but with long-range information	Phasing information, detects large SVs	Limited complex SV resolution compared to true long-read technologies

Computational Frameworks for Integrative Karyotype Reconstruction

Graph-Based Approaches for Genome Reconstruction

The computational reconstruction of complex karyotypes requires sophisticated algorithms that can integrate multiple types of genomic evidence. InfoGenomeR represents a cutting-edge graph-based framework that reconstructs individual SVs into karyotypes by integrating SV calls, total copy number alterations, allele-specific copy numbers, and haplotype information based on whole-genome sequencing data [72]. This method constructs a breakpoint graph composed of nodes and segment edges, reference edges, and SV edges, which undergoes iterative refinement through three-step iterations that refine local genomic segments, estimate integer copy numbers using purity and ploidy, and determine edge multiplicities through integer programming [72]. The power of this approach lies in its ability to move beyond mere SV detection to actual karyotype reconstruction, enabling the identification of derivative chromosomes, homogeneously staining regions for oncogenes like CCND1 and ERBB2, and double minutes in glioblastoma and ovarian cancer samples [72].

Multi-Modal Data Integration Strategies

Effective karyotype resolution requires the integration of diverse data types to overcome the limitations of any single approach. The InfoGenomeR framework begins by evaluating all reads in WGS data sets and generating initial SV calls using multiple tools (DELLY, Manta, and novoBreak), then performs initial copy number segmentation using BIC-seq2 [72]. Crucially, at an intermediate step between first and second-round iterations, discordant or unmapped reads that do not pair properly are remapped to sequences of candidate adjacencies from unbalanced nodes, enabling the discovery of additional SVs [72]. The integration of allele-specific copy number information further enhances reconstruction accuracy by employing negative binomial models for different depths of heterozygous SNPs and using an expectation-maximization algorithm for parameter estimation [72]. This multi-modal approach demonstrates significantly improved performance compared to individual SV calling tools, achieving precision of 0.987 and recall of 0.825 for total SV calling at 15X haplotype coverage [72].

Table 2: Key Computational Tools for SV Detection and Karyotype Resolution

Tool	Primary Function	Data Inputs	Strengths
InfoGenomeR	Genome karyotype reconstruction	SVs, CNAs, allele-specific CNs, haplotype information	Reconstructs linear and circular karyotypic topologies
DELLY	SV calling	Paired-end, split-reads	Comprehensive SV type detection
Manta	SV and indel calling	Paired-end WGS	Rapid discovery of SVs
novoBreak	SV detection from WGS	Breakpoint evidence from read alignment	Sensitive for novel SV discovery
scClone	Clonal evolution from scRNA-seq	Single-cell transcriptomes	Links genotype and phenotype at single-cell level

Single-Cell Approaches for Resolving Clonal Heterogeneity

Technical Challenges in Single-Cell Genomics

While bulk sequencing approaches provide an averaged view of the tumor genome, single-cell technologies are essential for resolving the intricate clonal architecture that characterizes tumor evolution [4]. However, single-cell DNA sequencing (scDNA-seq) faces significant technical challenges, including ultralow DNA input per cell, amplification-induced artifacts, and high cost per cell, which limit its widespread adoption [4]. Single-cell RNA sequencing (scRNA-seq) has emerged as a powerful alternative for inferring clonal structure, but mutation detection from scRNA-seq data is complicated by factors such as differential gene expression, allelic imbalance, RNA editing, limited sequencing coverage, and technical artifacts [4]. Despite these challenges, the ability to link cellular genotypes with transcriptional phenotypes at single-cell resolution provides unprecedented opportunities for understanding clonal evolution in cancer.

scClone: Integrating Variant Detection and Genotype Inference

The scClone computational toolkit addresses key limitations in single-cell clonal analysis by integrating variant detection and genotype inference for scRNA-seq and spatial transcriptomic data [4]. This approach processes raw sequencing reads to detect somatic mutations, impute drop-outs, and visualize clonal structures and evolutionary relationships, effectively leveraging single-cell transcriptomic annotations and bulk sequencing-derived mutational signatures [4]. A critical innovation in scClone is the implementation of a support vector machine (SVM) filtering step that significantly improves mutation calling quality—increasing the proportion of mutation sites with read depth >20 from 29.5% to 73.3% while reducing mutations with depth of 1 from 48.3% to 7.5% [4]. This filtering also reduces T>C and C>T substitutions primarily attributed to RNA editing, resulting in mutational signatures that more closely resemble those derived from bulk WES data (cosine similarity increased from 0.47 to 0.79) [4]. Applied to spatial transcriptomics, scClone enables the delineation of clonal structures within histological sections, providing spatial context to tumor evolution [4].

Experimental Protocols for Comprehensive Karyotype Resolution

Integrated Workflow for Bulk Sample Analysis

A robust protocol for resolving complex karyotypes from bulk tissue samples involves multiple integrated steps:

• Sample Preparation and Sequencing: Extract high-molecular-weight DNA from tumor and matched normal tissue. For long-read sequencing, use the Oxford Nanopore LSK-114 ligation sequencing kit with library preparation optimized for ultra-long reads (protocol: shearing to 50-100 Kb fragments, end-repair, adapter ligation, and purification). Sequence on a PromethION flow cell with 48-hour run time to achieve >30X coverage [73].

• Multi-Tool SV Calling: Process raw sequencing data through multiple SV callers to generate a comprehensive set of candidate SVs. For short-read data: use DELLY2 (command: delly call -g reference.fa -o sv.bam -x human.hg19.excl.tsv tumor.bam control.bam), Manta (command: configManta.py --tumorBam tumor.bam --normalBam normal.bam --referenceFasta reference.fa --runDir manta_analysis), and novoBreak simultaneously [72]. For long-read data: use Sniffles2 (command: sniffles -i input.bam -v output.vcf --tandem-repeats tandem_repeats.bed) and CuteSV [74].

• Copy Number and Ploidy Estimation: Perform copy number segmentation using BIC-seq2 with 10-Kb bins, then estimate tumor purity and ploidy using ABSOLUTE with default parameters [72].

• Integrative Graph Construction: Implement the InfoGenomeR framework to construct an initial breakpoint graph using SV and copy number breakpoints, followed by iterative refinement through local segment refinement, integer copy number estimation, and edge multiplicity determination via integer programming of the copy number balance condition [72].

• Haplotype Phasing: Divide integer copy numbers into allele-specific copy numbers using negative binomial models for heterozygous SNP depths, phase balanced heterozygous SNPs using BEAGLE, and construct the final haplotype breakpoint graph [72].

• Karyotype Reconstruction: Enumerate Eulerian paths to obtain candidate genomes by pairing breakpoint graph edges using a multiway tree structure with minimum-entropy search, generating candidate karyotypes of cancer cells at the haplotype level [72].

Single-Cell Multi-Omic Protocol for Clonal Evolution

For resolving clonal evolution at single-cell resolution:

• Single-Cell Sequencing: Prepare single-cell suspensions from fresh tumor tissue using the 10X Genomics Chromium Next GEM Single Cell Multiome ATAC + Gene Expression kit according to manufacturer's protocol, targeting recovery of 10,000 cells per sample [21] [4].

• Mutation Detection from scRNA-seq: Process raw sequencing reads through the scClone pipeline, which includes read alignment using STAR with default parameters, mutation calling using a custom pileup approach, and SVM-based filtering to remove technical artifacts and RNA editing events [4].

• Genotype Imputation and Clonal Inference: Impute missing genotypes due to expression drop-outs using a k-nearest neighbors approach (k=15) based on transcriptional similarity, then perform hierarchical clustering of mutation profiles to infer clonal populations [4].

• Integration with Spatial Transcriptomics: For Visium spatial transcriptomics data, overlay clonal assignments with spatial coordinates to map clonal distribution within tissue architecture, using the Seurat R package for integration and visualization [4].

• Evolutionary Analysis: Construct phylogenetic trees of clonal relationships using the neighbor-joining method with Jaccard distance based on shared mutations, then map transcriptional phenotypes to clonal identities to investigate genotype-phenotype relationships [4].

Visualization and Data Interpretation

Workflow Diagram for Complex Karyotype Resolution

The following diagram illustrates the integrated computational workflow for resolving complex karyotypes from multi-modal sequencing data:

Diagram Title: Integrative Workflow for Karyotype Resolution

Table 3: Essential Research Reagents and Computational Resources for Karyotype Resolution Studies

Category	Specific Resource	Application/Purpose
Sequencing Kits	Oxford Nanopore LSK-114 Ligation Sequencing Kit	Long-read WGS for SV detection
Target Capture	IDT xGen Hybridization Capture	Target enrichment for specific genomic regions
Single-Cell Platforms	10X Genomics Chromium Single Cell Multiome	Simultaneous gene expression and chromatin accessibility
Reference Materials	Genome in a Bottle (GIAB) reference standards	Benchmarking SV detection accuracy
Bioinformatics Tools	InfoGenomeR package	Graph-based karyotype reconstruction
Variant Callers	DELLY2, Manta, novoBreak	Comprehensive SV detection
Visualization Software	Integrative Genomics Viewer (IGV)	Visualization of SVs and read alignments
Data Resources	TCGA, ICGC	Access to cancer genomics datasets for comparison

Clinical Implications and Future Directions

The resolution of complex karyotypes has profound implications for understanding clonal evolution in cancer and advancing precision medicine. In mantle cell lymphoma, for example, multi-omic studies integrating single-cell RNA sequencing and whole-genome sequencing have revealed significant intratumor heterogeneity already present at diagnosis, with minor clones acquiring different mutations and copy-number variations during disease progression [21]. The ability to distinguish private and shared SVs between primary and metastatic cancer sites provides critical insights into tumor evolution and the development of therapeutic resistance [72]. As these technologies mature, clinical implementation will require standardized frameworks to ensure accuracy and reproducibility in SV detection, including the use of reference materials, validated bioinformatics pipelines, and reporting standards [74].

Future developments in cancer karyotype resolution will likely focus on the integration of artificial intelligence approaches to improve SV calling accuracy, the widespread adoption of single-cell multi-omics to resolve fine-scale clonal architecture, and the incorporation of spatial transcriptomics to map clonal distributions within tissue context [74] [4]. The ongoing development of long-read sequencing technologies with improved accuracy and throughput will further enhance our ability to resolve complex karyotypes, ultimately advancing our understanding of clonal evolution in cancer and opening new avenues for targeted therapeutic interventions.

Benchmarking and Clinical Correlations: Validating Single-Cell Findings

The delineation of clonal evolution in cancer is critical for understanding therapeutic resistance and disease progression. This complex process, characterized by the emergence of genetically distinct subpopulations, requires a multi-faceted genomic approach for comprehensive characterization. This technical guide details the methodology for cross-platform validation integrating Whole Genome Sequencing (WGS), Fluorescence In Situ Hybridization (FISH), and Optical Genome Mapping (OGM) to reconstruct accurate clonal architectures. We provide experimental protocols, analytical frameworks, and validation metrics that leverage the complementary strengths of each technology, enabling researchers to achieve unprecedented resolution in tracking tumor heterogeneity and evolution in the era of single-cell cancer analysis.

Cancer progression follows Darwinian evolutionary principles, where tumors consist of cellular populations with distinct genotypes that dynamically evolve over time and during treatment, a process known as clonal evolution. This diversity drives tumor heterogeneity, leading to differential growth advantages, metastatic potential, and therapeutic responsiveness [4]. Advanced cancers often develop resistance to multiple therapies partly as a result of this diversity, complicating treatment strategies [9].

Traditional bulk sequencing approaches suffer from a fundamental limitation: they infer clonal architectures through variant allele frequency (VAF)-based clustering, but the essence of tumor clones is the clustering of cell lineages. This inherent flaw introduces inaccuracies and deviates from true clonal structure [4]. The integration of multiple orthogonal technologies provides a solution to this challenge, allowing researchers to overcome the limitations of individual platforms.

Whole Genome Sequencing (WGS) offers base-pair resolution across the entire genome, enabling detection of single nucleotide variants, small insertions/deletions, and structural variants. Fluorescence In Situ Hybridization (FISH) provides spatial context and validation of structural variants within tissue architecture and at single-cell resolution. Optical Genome Mapping (OGM) delivers long-range genomic information with high sensitivity for large structural variants, serving as a bridge between cytogenetic and sequencing approaches. The convergence of these technologies creates a powerful framework for validating clonal populations and their evolutionary trajectories.

Technology Comparisons and Complementary Strengths

Table 1: Technical Specifications and Performance Metrics of Genomic Technologies

Parameter	Optical Genome Mapping	Whole Genome Sequencing	FISH
Resolution	~500 bp for SVs, >30 kbp for CNVs	Base-pair for SNVs, >50 bp for SVs	>50 kbp
Variant Types Detected	SVs, CNVs, aneuploidy	SNVs, indels, SVs, CNVs	Targeted SVs, aneuploidy
Throughput	High (genome-wide)	High (genome-wide)	Low (targeted)
Turnaround Time	5-7 days	7-10 days	2-3 days
Sample Requirements	High molecular weight DNA (>150 kbp)	Standard DNA (>1 μg)	Intact cells/tissues
Key Strengths	Genome-wide SV detection, no amplification bias	Comprehensive variant detection, base resolution	Single-cell resolution, spatial context
Limitations	Limited small variant detection	Short reads miss complex SVs	Targeted approach, low resolution

Table 2: Clinical Validation Performance of OGM Versus Standard Methods in AML

Performance Metric	OGM Result	Standard Methods Result
Concordance for SVs/CNVs	100% (when clone >5%)	Reference standard
Additional Clinically Relevant Findings	13% of cases	Not detected
Cryptic Translocations in Normal Karyotypes	3 cases identified	Reported as normal
Cases with Altered Clinical Management	4%	N/A
Cases Eligible for Trials Based on OGM	Additional 8%	N/A

The data in Table 2 comes from a multicenter evaluation of OGM in 100 AML cases, which demonstrated that OGM not only recovers all clinically relevant SVs and CNVs found by standard cytogenetic methods but also reveals additional structural variants not previously reported [75]. This enhanced detection capability directly impacts clinical decision-making, as evidenced by the percentage of cases where management would have been altered.

Comparative studies between OGM and long-read sequencing platforms (PacBio, ONT) show that approximately 99% of translocations and 80% of deletions identified by OGM were confirmed by both PacBio and ONT, while 10x Genomics in combination with PacBio and/or ONT confirmed approximately 70% [76]. Interestingly, long deletions (>100 kbp) were detected only by 10x Genomics, while inversions and duplications detected by OM were not detected by WGS platforms, highlighting the complementary nature of these technologies [76].

Experimental Protocols for Cross-Platform Validation

Optical Genome Mapping Workflow

Sample Preparation and DNA Extraction:

• Isolate high molecular weight (HMW) DNA from fresh or frozen tissue using specialized isolation kits (Bionano Prep SP Blood and Cell DNA Isolation Kit).
• Assess DNA quality using pulsed-field gel electrophoresis or the Agilent Femto Pulse system, ensuring DNA fragments >150 kbp with minimal degradation.
• Label DNA using a direct enzyme approach (Bionano Prep DLS Labeling Kit) with a fluorescent label at a specific 6-base sequence pattern (CTTAAG).
• Perform DNA backbone staining with a counterstain to visualize the full DNA molecule.

Data Collection and Analysis:

• Load labeled DNA into the Saphyr chip for linearized imaging through nanochannel arrays.
• Image hundreds of thousands of molecules per sample using the Saphyr instrument to achieve approximately 400X effective coverage.
• Assemble genomes de novo or align to a reference genome using Bionano Access software.
• Call structural variants (>500 bp) and copy number variants (>30 kbp) using built-in algorithms with manual review in Access software.

WGS Library Preparation and Sequencing

Library Preparation:

• Fragment DNA to 350-400 bp using acoustic shearing (Covaris).
• Perform library preparation using PCR-free methods (Illumina DNA PCR-Free Prep) to minimize bias, particularly important for preserving authentic structural variant detection.
• Assess library quality using fragment analyzers (Agilent 5400) to ensure fragment size distribution peaks overwhelmingly above 3000 bp for optimal WGS [77].

Sequencing and Analysis:

• Sequence on Illumina NovaSeq X Plus to target coverage of 30-60X with paired-end 150 bp reads.
• Perform quality control using FastQC v.0.12.1 and adapter trimming with AdapterRemoval v.2.3.3 [77].
• Map to reference genome using BWA v.2.2.1, sort with SAMtools v.1.19.2, and remove optical duplicates using picard v.3.2.0 [77].
• Call structural variants using Manta, Delly, or Lumpy; call SNVs/indels using GATK Mutect2 for somatic variants.

FISH Validation Protocol

Probe Design and Hybridization:

• Design break-apart or fusion probes targeting specific structural variants identified by OGM and WGS.
• Prepare metaphase chromosomes or interphase nuclei from patient samples.
• Denature probe and target DNA simultaneously at 75°C for 5 minutes.
• Hybridize overnight at 37°C in a humidified chamber.

Detection and Analysis:

• Wash stringently to remove non-specifically bound probe.
• Counterstain with DAPI and image using a fluorescence microscope with appropriate filter sets.
• Score a minimum of 100 interphase cells or 20 metaphase spreads for signal patterns.
• Validate clonal nature of variants by confirming presence in significant subset of cells.

Integrated Analytical Framework for Clonal Evolution

The workflow begins with parallel data generation from the three complementary technologies. OGM provides comprehensive structural variant calls, WGS delivers base-resolution mutation data, and FISH offers spatial validation at single-cell resolution. The integration point occurs during SV concordance analysis, where variants are categorized based on confirmation across platforms.

High-confidence variants from the concordance analysis feed into clonal structure reconstruction. At this stage, computational tools such as MEDICC2 for single-cell phylogenetics or scClone for mutation detection from single-cell transcriptomics can be employed to infer clonal relationships [5] [4]. These tools enable the construction of phylogenetic trees based on allele-specific copy-number alterations or detected somatic mutations.

The final stage involves evolutionary trajectory inference, where temporal relationships between clones are reconstructed and potential drivers of clonal expansion are identified. This integrated approach allows researchers to distinguish truncal events present in all clones from subclonal mutations that define branching evolution, providing critical insights into resistance mechanisms and disease progression.

Research Reagent Solutions and Computational Tools

Table 3: Essential Research Reagents and Computational Tools for Cross-Platform Validation

Category	Product/Tool	Specific Application	Key Features
DNA Extraction	QIAGEN DNeasy Blood & Tissue Kit	DNA extraction from swabs, fin clips, tissues	Includes RNase treatment, Proteinase K option [77]
HMW DNA Isolation	Bionano Prep SP Blood and Cell DNA Isolation	OGM-compatible DNA extraction	Preserves long DNA fragments >150 kbp [75]
DNA Labeling	Bionano Prep DLS Labeling Kit	Fluorescent labeling for OGM	Specific 6-base sequence recognition [75]
Library Prep	Illumina DNA PCR-Free Prep	WGS library preparation	Minimizes amplification bias [77]
FISH Probes	Region-specific break-apart/fusion probes	Validation of specific SVs	Custom design for target regions
SV Analysis	AnnotSV	SV annotation and comparison	Facilitates OGM-WGS comparisons [76]
Clonal Analysis	MEDICC2	Single-cell phylogenetics	Phylogenetic trees from copy-number data [5]
Mutation Calling	scClone	Mutation detection from scRNA-seq	Integrates variant detection and genotype inference [4]

Case Studies in Cancer Clonal Evolution

High-Grade Serous Ovarian Cancer (HGSOC)

The CloneSeq-SV approach exemplifies successful technology integration by combining single-cell whole-genome sequencing with targeted deep sequencing of clone-specific genomic structural variants in time-series cell-free DNA [5]. This method exploits tumor clone-specific SVs as highly sensitive endogenous cell-free DNA markers, enabling relative abundance measurements and evolutionary analysis of co-existing clonal populations over the therapeutic time course.

In a study of 18 HGSOC patients followed from diagnosis to recurrence, researchers demonstrated that drug resistance typically arose from selective expansion of a single or small subset of clones present at diagnosis [5]. CloneSeq-SV provided several advantages: SVs showed error rates orders of magnitude lower than SNVs, enabled confident detection of tumor DNA even from single events without requiring error correction, and their frequent association with high-level amplifications resulted in high per-cell copy numbers that enhanced detection sensitivity.

Acute Myeloid Leukemia (AML) Evaluation

A multicenter evaluation of OGM in 100 AML cases demonstrated its significant value in clinical assessment [75]. The study showed that OGM identified all clinically relevant SVs and CNVs reported by standard cytogenetic methods when representative clones were present in >5% allelic fraction. Importantly, OGM revealed additional clinically relevant information in 13% of cases that had been missed by routine methods, including three cases with normal karyotypes that were shown to have cryptic translocations involving gene fusions.

The study further quantified the clinical impact: findings from OGM would have altered recommended clinical management in 4% of cases and rendered an additional 8% potentially eligible for clinical trials [75]. This demonstrates how advanced genomic technologies can directly influence therapeutic decision-making in clonal disorders.

The integration of WGS, FISH, and OGM provides a powerful framework for resolving clonal evolution in cancer with unprecedented resolution. This cross-platform validation approach leverages the complementary strengths of each technology: WGS for base-pair resolution, OGM for comprehensive structural variant detection, and FISH for spatial validation at single-cell resolution.

As single-cell technologies continue to advance, the framework described here will enable researchers to address fundamental questions in cancer evolution, including the dynamics of therapeutic resistance, the identification of mutationally cooperative clones, and the spatial organization of clonal populations within tumors. The computational integration of multi-platform data represents the next frontier in cancer genomics, promising to transform our understanding of clonal evolution and ultimately improve patient outcomes through more precise diagnostic and therapeutic approaches.

The progression of cancer is a dynamic evolutionary process driven by the accumulation of somatic mutations, resulting in distinct cellular clones that compete within the tumor ecosystem [78] [79]. Clonal evolution describes the process through which tumor clones undergo phylogenetic diversification, driven by selective pressures exerted by the tumor microenvironment (TME) or therapeutic interventions, ultimately shaping the tumor's evolutionary trajectory [4]. While genetic drivers have long been the focus of cancer research, non-genetic mechanisms can modulate cellular states and enhance adaptive flexibility if their diversification can persist and form an epigenetic memory [14]. Single-cell technologies have revolutionized our ability to study this complexity by enabling joint profiling of both mutational and transcriptomic landscapes within the same cells, revealing intricate genotype-phenotype relationships that are obscured in bulk analyses [78] [80] [79]. This technical guide examines current methodologies and insights into correlating genetic mutations with transcriptional states, providing a framework for researchers investigating clonal evolution in cancer.

Theoretical Framework: Genetic and Non-Genetic Evolutionary Mechanisms

Genetic Evolution and Clonal Selection

Cancer evolution is traditionally characterized by branching phylogenies, where subclones with unique genetic profiles emerge at different locations and time points [9]. Propagation of clonal regulatory programs contributes to cancer development through driver mutations that provide selective advantages [14]. The clonal composition of a tumor changes over time, and this evolution is one of the mechanisms by which new characteristics can be acquired during cancer progression, including clinically significant phenotypical changes such as metastasis or drug resistance [78].

Non-Genetic Evolution and Epigenetic Memory

Non-genetic mechanisms enable rapid adaptation and diversification in the context of a dynamic stromal environment, immune interactions, or following treatment [14]. The ability of cells to maintain their molecular identity through mitotic cell divisions is essential for establishing functionally coherent and stable clonal cell populations. DNA methylation is the best-studied epigenetic mechanism for stable memory formation, and the ability of cells to copy their methylation makeup to daughter cells is well established [14]. Beyond methylation, theoretical and experimental models demonstrate commitment and memory through specific gene network architecture that can generate clonally stable transcriptional phenotypes in mammalian systems [14].

Integrated Evolutionary Model

An integrated model of cancer evolution recognizes that genetic and non-genetic mechanisms operate in parallel to shape tumor progression. Extrachromosomal DNA (ecDNA) has recently emerged as a crucial player in driving the evolution of about 20% of all tumors, contributing to genomic instability and treatment resistance [81]. Meanwhile, germline genetic variation influences somatic evolution in tissues, shaping tissue-specific mutational fitness and impacting the risk of progression to hematologic malignancies [82]. This complex interplay creates a degenerated relationship between mutational and transcriptional states, where clones can converge on similar transcriptional fates through different mechanisms [78] [14].

Methodological Approaches: Experimental and Computational Frameworks

Single-Cell Multi-Omic Profiling Technologies

Comprehensive multiomic analysis of single cells addresses the challenges associated with traditional cancer profiling methods by offering a holistic view of clonal heterogeneity [79]. This approach provides a high-resolution and integrated understanding of cancer biology by simultaneously analyzing multiple molecular modalities—such as DNA, RNA, and proteins—within individual cells [79].

Table 1: Single-Cell Technologies for Correlating Genetic and Transcriptional States

Technology	Molecular Modality	Key Applications	Limitations
Full-length scRNA-seq (SMART-seq2)	Transcriptome + inferred mutations	Mutation calling from RNA reads, CNV inference	Limited genomic coverage, RNA editing artifacts
scClone computational toolkit [4]	scRNA-seq + spatial transcriptomics	Variant detection, genotype inference, clonal visualization	Expression drop-out, allelic imbalance
Single-cell multiome (Mission Bio) [79]	DNA + protein (simultaneous)	Clonal architecture, surface protein expression	Requires specialized platform, cost
Luria-Delbrück design [14]	Longitudinal transcriptome + epigenome	Distinguishing stable vs. transient expression	In vitro model system
CloneSeq-SV [5]	scWGS + cfDNA tracking	Clonal tracking via structural variants in plasma	Complex workflow, analysis pipeline

Computational Framework for Joint Analysis

The Canvolution computational framework provides a standardized approach for joint characterization of the mutational and transcriptional landscapes from full-length scRNA-seq data, consisting of five integrated steps [78]:

• Preprocessing: Single-nucleotide variants (SNVs) and short indels are identified using CTAT in combination with a method based on the STAR aligner and GATK-best practice variant calling pipeline for inferring SNVs from full-length scRNA-seq protocols [78].

• Clonal identification and tree inference: Based on the mutations, clones are inferred using the DENDRO algorithm, and an evolutionary tree is generated by RobustClone [78].

• Clonal enrichment characterization: For each path through the evolutionary tree, clonal enrichment is characterized. A gene signature score (Ms) is defined as the intersection between a set of pre-defined genes with the mutated genes in a clone [78].

• Transcriptional state identification: Clustering of cancer cells by gene expression is done by standard Louvain clustering using the Seurat package [78].

• Integrated scoring: Calculation of gene signature scores for mutation, transcription, mutated-gene expression, and mutated ligand-receptor pairs in each clone-cluster combination [78].

Specialized Methods for Evolutionary Tracking

CloneSeq-SV combines single-cell whole-genome sequencing with targeted deep sequencing of clone-specific genomic structural variants in time-series cell-free DNA [5]. This approach exploits tumor clone-specific structural variants as highly sensitive endogenous cell-free DNA markers, enabling relative abundance measurements and evolutionary analysis of co-existing clonal populations over the therapeutic time course [5].

The Luria-Delbrück experimental design distinguishes clonally stable epigenetic memory from transient transcriptional fluctuations by comparing single-cell transcriptional and epigenetic distributions to the distributions of mean gene expression and methylation across clones originating from the same cell populations [14]. This design enables researchers to determine whether transcriptional heterogeneity represents stable, heritable programs or transient cellular states.

Key Analytical Metrics and Quantitative Frameworks

Signature Scores for Evolutionary Analysis

The evolutionary path score quantifies how gene signatures change as clones evolve by calculating the correlation coefficient between a gene signature score (Ms) and the tree depth [78]. Similarly, the clonal selection score identifies mutated gene sets associated with increasing clone sizes by correlating Ms with clone size [78].

Table 2: Quantitative Metrics for Evolutionary Analysis

Metric	Calculation	Biological Interpretation	Application Example
Evolutionary Path Score	Correlation between gene signature score (Ms) and clonal tree depth	Identifies features associated with disease progression	Mutations affecting angiogenesis genes increasing with tree depth [78]
Clonal Selection Score	Correlation between Ms and clone size	Identifies gene sets associated with clonal expansion	Drug resistance mutations enriched in larger clones [78]
Transcriptional Signature Score (Ts)	AddModuleScore in Seurat package	Characterizes transcriptional states associated with clonal age or size	EMT signature correlated with metastatic potential [78]
Mutated LR Score (Mi)	Overlap between mutated genes and ligand-receptor pairs	Quantifies altered tumor microenvironment interactions	Mutations in ligand genes affecting immune cell crosstalk [78]
CNV Score	Extent of copy number variations per cell	Measures genomic instability	Higher in metastatic vs. primary tumors [80]

Quantitative Insights from Cancer Studies

Application of these frameworks to human cancers has yielded quantitative insights into the relationship between genetic and non-genetic evolution:

• In lung cancer and chronic myeloid leukemia, analyses reveal high clonal and transcriptional diversity with little evidence for clonal sweeps, suggesting selection based solely on growth rate is unlikely to be the dominating driving force during cancer evolution [78].

• In ER+ breast cancer, metastatic tumors demonstrate higher CNV scores compared to primary tumors, indicating increased genomic instability in advanced disease [80].

• Across multiple cancer types, each clone is associated with a preferred transcriptional state, demonstrating a degenerated relationship between mutational and transcriptional landscapes [78].

• For metastasis and drug resistance, the number of mutations affecting related genes increases as the clone evolves, while changes in gene expression profiles are limited [78].

• Mutations affecting ligand-receptor interactions with the tumor microenvironment frequently emerge as clones acquire drug resistance [78].

Research Reagent Solutions and Experimental Tools

Table 3: Essential Research Reagents and Computational Tools

Reagent/Tool	Function	Application Context
Canvolution framework [78]	Computational pipeline for joint mutation/transcriptome analysis	Evolutionary path and clonal selection analysis from scRNA-seq
scClone toolkit [4]	Mutation detection and clonal evolution from scRNA-seq	Genotype-phenotype association in single-cell and spatial transcriptomes
DENDRO algorithm [78]	Clone inference from mutation data	Defining clonal architecture from single-cell data
RobustClone [78]	Evolutionary tree generation	Phylogenetic reconstruction from clonal data
InferCNV [80]	Copy number variation inference from scRNA-seq	Identifying malignant cells and genomic instability
CellChat [78]	Cell-cell communication inference	Analysis of ligand-receptor interactions in TME
SVM-based mutation filtering [4]	Artifact reduction in mutation calling	Improving mutation detection accuracy in scRNA-seq
Luria-Delbrück framework [14]	Distinguishing stable vs. transient heterogeneity	Epigenetic memory detection in cell populations

Signaling Pathways and Biological Processes in Clonal Evolution

Analysis of clonal evolution has identified several key signaling pathways and biological processes that connect genetic alterations to transcriptional states:

The epithelial-to-mesenchymal transition (EMT) spectrum represents a key transcriptional axis in clonal evolution. In colon cancer cells, longitudinal transcriptional and genetic analysis reveals a slowly drifting spectrum of epithelial-to-mesenchymal transcriptional identities that is seemingly independent of genetic variation [14]. DNA methylation landscapes correlate with these identities but also reflect an independent clock-like methylation loss process [14].

For drug resistance, interpretable and distinctive genomic features emerge in resistant clones, including chromothripsis, whole-genome doubling, and high-level amplifications of oncogenes such as CCNE1, RAB25, MYC, and NOTCH3 [5]. Phenotypic analysis of matched single-cell RNA sequencing data indicates pre-existing and clone-specific transcriptional states such as upregulation of EMT and VEGF pathways, linked to drug resistance [5].

The JAK-STAT pathway and RAS signaling represent key connectors between germline genetic variation and somatic evolution, with germline variants in genes like MPL and PTPN11 shaping the fitness landscape for clonal expansions [82].

Integrating genetic and transcriptional data at single-cell resolution provides unprecedented insights into the parallel evolutionary mechanisms driving cancer progression. The correlative frameworks and methodological approaches outlined in this technical guide enable researchers to dissect the complex interplay between stable genetic alterations and more plastic transcriptional and epigenetic states. As single-cell multi-omic technologies continue to advance and computational frameworks become more sophisticated, we anticipate increasingly detailed maps of clonal evolutionary trajectories that will inform more effective, personalized treatment protocols and evolutionary-informed therapeutic strategies [9]. Future research directions should focus on longitudinal tracking of clonal dynamics in patient samples, developing more sophisticated computational models of evolutionary trajectories, and translating insights from clonal architecture into improved clinical stratification and patient-specific therapeutic approaches.

The longitudinal tracking of cancer clones from diagnosis through minimal residual disease (MRD) to relapse is a cornerstone of modern oncological research, providing an unparalleled window into the dynamic process of clonal evolution. This evolution is the primary driver of therapeutic resistance and disease recurrence. The comparison of samples across these critical timepoints reveals how tumor populations, under the selective pressure of treatment, undergo dynamic changes in their genetic architecture and cellular composition. Understanding these patterns is not merely an academic exercise; it is essential for developing more effective, evolution-informed treatment strategies that can preempt resistance and improve patient outcomes. This technical guide details the methodologies and analytical frameworks enabling researchers to decode this evolutionary narrative, framing the discussion within the broader thesis that cancer is a disease of constant Darwinian adaptation, the traces of which can be tracked in real time.

Methodologies for Longitudinal Sample Analysis

A diverse array of technologies is employed to capture the complex clonal dynamics across the cancer treatment timeline. The choice of method profoundly influences the resolution, sensitivity, and type of evolutionary insight that can be gained.

Genomic and Transcriptomic Profiling Techniques

Table 1: Core Methodologies for Longitudinal Clonal Tracking

Methodology	Core Principle	Applications in Longitudinal Tracking	Key Advantages	Inherent Limitations
Single-Cell DNA Sequencing (scDNA-seq) [83] [5] [84]	Sequences the genome of individual cells to resolve co-mutation patterns and phylogeny.	Tracking subclonal architecture and emergence of resistant clones from Dx to relapse.	Unambiguous determination of clonal structure and phylogenetic relationships.	Limited by input cell numbers; may miss very rare subclones.
Single-Cell Multi-Omics (DNA+Protein) [85] [84]	Simultaneously profiles mutations and surface protein expression in single cells.	Correlating genotype with immunophenotype in MRD; distinguishing MRD from CHIP.	Enhances MRD detection specificity by combining genotypic and phenotypic data.	Technically complex and costly; requires specialized platforms.
Tumor-Informed ctDNA Tracking [86] [5] [87]	Uses patient-specific mutations identified in tumor tissue to design a bespoke panel for detecting ctDNA in plasma.	Ultrasensitive monitoring of MRD and early relapse via liquid biopsy.	Highly sensitive (up to 10-6), non-invasive, allows frequent monitoring.	Requires a high-quality baseline tumor sample; turnaround time for panel design.
Single-Cell RNA Sequencing (scRNA-seq) [88] [89]	Profiles the transcriptome of individual cells to identify cellular states and expression programs.	Identifying therapy-resistant cellular states (e.g., quiescent stem cells) and transcriptomic reprogramming.	Reveals functional cellular heterogeneity and plasticity in response to therapy.	Does not directly sequence somatic mutations; inference of clonal relationships is indirect.
Structural Variant (SV) Tracking (CloneSeq-SV) [5]	Uses somatic structural variants as highly specific endogenous markers to track clones in cfDNA.	Evolutionary tracking of co-existing clonal populations over a therapeutic time course.	Extremely low error rates, high specificity, effective in SV-rich cancers like HGSOC.	Less effective in cancers with low SV burden; identification requires deep sequencing.

Key Experimental Protocols

Protocol 1: Single-Cell Multi-Omic MRD Analysis [85] This protocol is designed to characterize residual disease in hematologic malignancies with single-cell resolution, co-assaying DNA mutations and protein expression.

• Sample Preparation: Obtain bone marrow or peripheral blood samples at diagnosis and post-treatment remission timepoints. Isolate mononuclear cells (MNCs) via density gradient centrifugation.
• Cell Enrichment: To enrich for potential leukemic blasts, use magnetic-activated cell sorting (MACS) with antibodies against CD34 and CD117. This step increases the likelihood of capturing rare MRD cells.
• Library Preparation & Sequencing: Process the enriched cell population using a commercial platform (e.g., Mission Bio Tapestri). The workflow involves:
  • Single-Cell Partitioning: Cells are co-encapsulated with beads in oil droplets.
  • DNA Barcoding: A DNA oligonucleotide barcode unique to each droplet tags all DNA from a single cell.
  • Multiplexed PCR Amplification: Targeted amplification of specific genomic regions (e.g., SNVs/indels from ELN guidelines) is performed.
  • Protein Barcoding: Oligonucleotide-tagged antibodies against surface markers (e.g., CD45, CD34, CD33) are used to label cells, allowing protein data to be captured in the same sequencing library.
• Data Analysis: After sequencing, bioinformatic pipelines demultiplex cells, align sequences, and call mutations and antibody-derived tags (ADTs) for each cell. Clones are identified based on co-mutation patterns, and their immunophenotypes are characterized.

Protocol 2: CloneSeq-SV for Tracking Clonal Evolution in cfDNA [5] This innovative protocol leverages structural variants to track clones non-invasively in solid tumors.

• Baseline Tumor Profiling:
  • Collect fresh tumor tissue at diagnosis (e.g., during debulking surgery).
  • Perform single-cell whole-genome sequencing (scWGS) on thousands of tumor cells using a high-throughput, tagmentation-based approach (e.g., DLP+).
  • Infer clonal composition and phylogeny from allele-specific copy number alterations.
  • Identify high-confidence, clone-specific SVs (e.g., from chromothripsis, breakage-fusion-bridge cycles) from the pseudobulk data of each clone.
• Bespoke Probe Design: For each patient, design hybrid-capture probes targeting the breakpoint junctions of ~50-100 truncal and clone-specific SVs (and optionally, key SNVs like TP53).
• Longitudinal Plasma Collection and Processing: Serially collect blood at multiple timepoints (pre-treatment, during therapy, post-surgery, at suspected relapse). Isolate cell-free DNA (cfDNA) from plasma.
• Library Prep and Deep Sequencing: Construct cfDNA libraries and perform duplex sequencing (error-corrected sequencing of both strands of the original DNA molecule) using the patient-specific probe panel.
• Variant Calling and Clonal Quantification: Identify supporting reads for each SV in the cfDNA. The variant allele frequency (VAF) of clone-specific SVs is used to calculate the relative abundance of each clone over time, reconstructing the evolutionary trajectory.

Diagram 1: CloneSeq-SV Workflow for Clonal Tracking in cfDNA.

The Scientist's Toolkit: Essential Research Reagents and Platforms

Table 2: Key Reagent Solutions for Longitudinal Single-Cell Studies

Reagent / Platform	Function	Application in Workflow
CD34/CD117 Magnetic Beads [85]	Cell Surface Marker-Based Enrichment	Isolates progenitor cells from bulk bone marrow or blood, enriching for leukemic blasts prior to single-cell MRD analysis.
Multiplexed scDNA-seq Panels [85] [84]	Targeted Gene Amplification	Enables focused, cost-effective sequencing of key mutational hotspots (e.g., ELN guideline genes in AML) in thousands of single cells.
Oligonucleotide-Tagged Antibodies [85] [84]	Multiplexed Protein Detection	Allows for simultaneous quantification of dozens of surface protein markers (immunophenotype) alongside genomic data in single-cell multi-omic assays.
Unique Molecular Identifiers (UMIs) [87]	Error Correction in NGS	Tags individual DNA molecules before PCR amplification, enabling bioinformatic correction of sequencing errors and more accurate variant calling in ctDNA assays.
Patient-Specific Hybrid Capture Panels [5] [87]	Ultrasensitive ctDNA Detection	Custom-designed probes that target a patient's unique set of somatic variants, enabling highly sensitive (10-5 - 10-6) tracking of MRD in plasma.

Key Findings and Clinical Implications from Longitudinal Studies

Longitudinal studies have consistently revealed patterns that challenge traditional, static views of cancer.

Patterns of Relapse and Resistance

• Pre-Existence of Resistant Clones: A pivotal finding across cancer types is that clones driving relapse often pre-exist at diagnosis, rather than universally arising de novo during therapy. In High-Grade Serous Ovarian Cancer (HGSOC), CloneSeq-SV analysis demonstrated that drug resistance typically arose from the selective expansion of a single or a small subset of clones present at diagnosis, leading to reduced clonal complexity at relapse [5]. Similarly, in AML, single-cell DNA sequencing has revealed that primary resistance can be mediated by pre-existing clones lacking the drug target (e.g., a FLT3-wildtype clone in a predominantly FLT3-mutant AML), while secondary resistance can emerge from rare, newly arisen clones with resistance mutations (e.g., NRAS) that expand under therapy [84].
• Clonal Architecture as a Prognostic Indicator: The complexity of the clonal architecture at diagnosis holds prognostic value. In AML, single-cell multi-omic analysis has shown that cases that eventually relapsed possessed a more branched, heterogeneous clonal architecture at diagnosis (averaging 4.6 clones/patient) compared to those that did not relapse (averaging 1.3 clones/patient) [85]. This greater genetic diversity provides a larger reservoir for selection under therapeutic pressure.
• Cellular Plasticity and Transcriptomic Reprogramming: Resistance is not solely genetic. scRNA-seq of relapsed/refractory AML bone marrow has identified a subpopulation of quiescent stem-like cells (QSCs) characterized by high expression of CD52 and LGALS1 (Galectin-1). Longitudinal analysis showed that proliferating stem/progenitor-like cells (PSPs) can be reprogrammed to a QSC-like state during chemotherapy, contributing to a dormant, therapy-resistant reservoir [89]. This demonstrates a non-genetic, cellular state mechanism of resistance.

Resolving Diagnostic Dilemmas with Single-Cell MRD

Single-cell multi-omics is particularly powerful for resolving discordant results from traditional MRD methods.

• A study analyzing discrepant AML cases found that in patients who were flow cytometry-positive for MRD but did not relapse, single-cell analysis sometimes confirmed the presence of cells with an abnormal immunophenotype but a non-malignant genotype, potentially representing benign, regenerating progenitors [85].
• Conversely, in patients who were flow cytometry-negative but later relapsed, single-cell analysis successfully identified residual leukemic cells that had acquired a different immunophenotype, evading detection by standard flow panels [85]. This genotype-phenotype discordance underscores the limitation of relying on a single analyte and the superior resolution of multi-omic approaches.

Diagram 2: Evolutionary Paths from Diagnosis to Relapse.

Longitudinal tracking of diagnostic, MRD, and relapse samples has fundamentally advanced our understanding of cancer as a dynamically evolving system. The integration of single-cell multi-omics and sensitive liquid biopsy technologies provides a high-resolution lens to view the Darwinian drama of clonal selection, revealing the genetic and non-genetic strategies tumors use to survive therapy. The consistent finding that resistant clones are often present early in the disease course mandates a shift in clinical strategy from reactive to pre-emptive. The future lies in evolution-informed adaptive therapy: using longitudinal monitoring to detect the first signs of resistant clone expansion and swiftly modifying treatment to suppress them, much like antimicrobial stewardship. As these technologies become more standardized and accessible, they promise to transform cancer from a lethal, relentless foe into a manageable, chronic condition by continuously anticipating and outmaneuvering its next evolutionary move.

Functional precision medicine represents a paradigm shift in oncology, moving beyond static genomic features to directly assess tumor behavior through dynamic, functional testing of live patient-derived cells [90]. This approach addresses a critical limitation of genomics-only strategies, which have shown modest clinical benefit, with overall response rates in large trials like NCI-MATCH often below 5% in an intention-to-treat analysis [90]. Patient-derived xenograft (PDX) models have emerged as a powerful platform for functional validation because they maintain key biological characteristics of original tumors, including intratumoral heterogeneity and tissue architecture, which are essential for predicting therapeutic response [90]. When integrated with single-cell analysis technologies, PDX models provide unprecedented resolution for tracking clonal evolution and understanding the dynamics of drug resistance development in cancer populations.

The convergence of functional testing with single-cell technologies creates a powerful framework for studying clonal evolution in cancer. As tumors evolve under therapeutic pressure, minor subclones with resistant phenotypes can expand and drive disease recurrence [5] [21]. Single-cell analysis of PDX models enables researchers to map these evolutionary trajectories and identify pre-existing resistant clones that may not be detectable through bulk sequencing approaches. This integration provides critical insights for developing evolution-informed adaptive treatment regimens aimed at circumventing or delaying the emergence of drug resistance [5].

PDX Models: Biological Fidelity and Technical Considerations

Model Establishment and Characterization

The generation of PDX models begins with the implantation of patient tumor tissue—obtained either through surgical resection or biopsy—into immunocompromised mice. This process preserves crucial aspects of the original tumor's biology. A comprehensive analysis of over 500 PDX models across various cancer types has demonstrated that these models recapitulate human tumors with relatively high fidelity and exhibit treatment responses concordant with those observed in the patients from whom they were derived [90].

Successful model validation requires thorough characterization to confirm that key pathological and molecular features of the original tumor are maintained. This includes histological comparison, verification of cancer markers, and genomic profiling. For instance, in high-grade serous ovarian cancer (HGSOC), isolated micro-tumors have been shown to recapitulate markers such as PAX8 and WT1, with strong correlation of protein expression between original tumor tissue and matched isolated micro-tumors [91]. Additionally, the BRCA mutation status of original tumors is maintained in PDX models, with a high correlation observed between the original tumor and the derived micro-tumors [91].

Table 1: Key Characterization Metrics for PDX Model Validation

Characterization Aspect	Analytical Method	Validation Benchmark
Histopathology	H&E staining	Maintenance of tumor morphology and architecture
Tumor Marker Expression	Immunohistochemistry (IHC)	Consistent expression patterns (e.g., PAX8, WT1 for ovarian cancer)
Genomic Stability	Whole-genome sequencing	Preservation of driver mutations and copy number variations
BRCA Status	Targeted sequencing	High correlation with original tumor tissue
Tumor Microenvironment	Single-cell RNA sequencing	Presence of immune populations and stromal components

Advantages and Limitations of PDX Models

PDX models offer several significant advantages for drug sensitivity profiling. They maintain the original tumor's heterogeneity and stromal components better than conventional 2D cell cultures, providing a more physiologically relevant system for therapeutic testing [90]. This preservation of tumor biology extends to functional characteristics, with PDX models demonstrating treatment responses that correlate with clinical outcomes in patients [90]. Additionally, these models serve as a renewable resource for longitudinal studies and allow for the study of human tumor biology in an in vivo context.

However, PDX models also present notable limitations. The engraftment process typically requires 3-6 months, making it challenging to use these models for real-time clinical decision-making in many cases [90]. The necessity for immunocompromised host animals limits the study of immunotherapeutic approaches and immune-mediated mechanisms of response and resistance. There may also be selective pressure during engraftment that alters clonal representation compared to the original tumor. Furthermore, the technical expertise and resource requirements for maintaining PDX colonies present significant practical barriers to implementation.

Experimental Design and Methodologies

The REMIT Assay for Microtubule-Targeting Agents

The REplication MITosis (REMIT) assay represents an innovative approach for assessing sensitivity to microtubule-targeting agents such as paclitaxel and eribulin. This method addresses the challenge that conventional viability readouts (e.g., apoptosis assays) may not capture the primary cytotoxic effects of these agents during short-term ex vivo exposure [92]. Unlike drugs that induce immediate apoptosis, microtubule-targeting agents primarily arrest cells in mitosis, a effect that may not be detected by standard cell death assays within typical experimental timeframes [92].

The REMIT assay quantifies drug effect by calculating the ratio between replicating cells (measured by EdU incorporation) and cells in mitosis (identified by phospho-Histone H3 (pH3) staining). This EdU/pH3 ratio decreases when tumor cells are sensitive to treatment, indicating cell cycle blockade in mitosis [92]. The assay has demonstrated 90% concordance between ex vivo predictions and in vivo responses to paclitaxel treatment in breast cancer PDX models, with a reproducibility of 80% for paclitaxel and 83% for eribulin [92].

Table 2: REMIT Assay Protocol and Thresholds for Drug Sensitivity

Assay Parameter	Paclitaxel	Eribulin
Treatment Duration	3 days	3 days
Key Readout	EdU/pH3 ratio	EdU/pH3 ratio
Critical Concentration	10 nM	Information not available in search results
Sensitivity Threshold	Relative EdU/pH3 < 45%	Information not available in search results
Concordance with In Vivo Response	90%	Suitable assay identified (exact concordance not specified)
Reproducibility	80%	83%

3D Micro-Tumour Platform for HGSOC

For high-grade serous ovarian cancer (HGSOC), a 3D micro-tumour testing platform has been developed that enables direct ex vivo assessment of chemosensitivity using tumor cells isolated from malignant ascites. This approach preserves important aspects of the original tumor microenvironment and can generate results within two weeks of sample collection, aligning with clinical decision-making timelines [91].

The platform involves isolating micro-tumors from ascites, embedding them in a 3D matrix, and exposing them to standard-of-care chemotherapies. High-content 3D imaging captures morphological features that are used to generate sensitivity profiles. A linear regression model trained on these features has demonstrated a strong correlation (R = 0.77) between predicted and clinical CA125 decay rates [91]. Patients classified as having high ex vivo sensitivity to carboplatin/paclitaxel showed significantly increased progression-free survival and decreased tumor size compared to those with predicted resistance [91].

Diagram 1: 3D Micro-tumor Testing Workflow

CloneSeq-SV for Tracking Clonal Evolution

The CloneSeq-SV approach combines single-cell whole-genome sequencing (scWGS) with targeted deep sequencing of clone-specific genomic structural variants in cell-free DNA to monitor clonal dynamics during therapy [5]. This method leverages tumor clone-specific structural variants as highly sensitive endogenous markers, enabling relative abundance measurements and evolutionary analysis of co-existing clonal populations over the therapeutic time course [5].

In application to HGSOC, CloneSeq-SV has revealed that drug resistance typically arises from selective expansion of a single or small subset of clones present at diagnosis rather than from de novo mutation events [5]. Drug-resistant clones frequently show distinctive genomic features including chromothripsis, whole-genome doubling, and high-level amplifications of oncogenes such as CCNE1, RAB25, MYC, and NOTCH3 [5]. Phenotypic analysis of matched single-cell RNA sequencing data has indicated pre-existing and clone-specific transcriptional states—such as upregulation of epithelial-to-mesenchymal transition and VEGF pathways—linked to drug resistance [5].

The Scientist's Toolkit: Essential Research Reagents

Table 3: Key Research Reagent Solutions for PDX-Based Drug Sensitivity Profiling

Reagent/Resource	Function	Application Example
EdU (5-ethynyl-2′-deoxyuridine)	Labels replicating DNA for detection of cell proliferation	REMIT assay to identify replicating cells [92]
Anti-phospho-Histone H3 (pH3)	Detects cells in mitotic phase	REMIT assay to quantify mitotic arrest [92]
TUNEL Assay Reagents	Labels apoptotic cells with DNA fragmentation	Assessment of drug-induced apoptosis [92]
3D Extracellular Matrix	Provides physiological scaffold for micro-tumor growth	3D micro-tumor culture platform [91]
Patient-Derived Xenografts	In vivo model maintaining tumor heterogeneity	Drug sensitivity validation and clonal evolution studies [90] [92]
Single-Cell RNA Sequencing Kits	Transcriptomic profiling at single-cell resolution	Analysis of tumor heterogeneity and resistant subpopulations [21]
Clone-Specific Structural Variant Panels	Tracking clonal dynamics in cfDNA	CloneSeq-SV for monitoring tumor evolution [5]

Data Analysis and Interpretation

Correlation with Clinical Outcomes

Validating PDX-derived drug sensitivity data against clinical outcomes is essential for establishing predictive value. In HGSOC, the 3D micro-tumor platform demonstrated significant correlation between ex vivo predictions and multiple clinical endpoints [91]. Patients with predicted high ex vivo sensitivity to carboplatin/paclitaxel showed significantly increased progression-free survival compared to those with predicted resistance (median 18 months vs. 12 months) [91]. This difference was particularly pronounced in patients undergoing interval debulking surgery, highlighting how functional testing could potentially guide treatment sequencing decisions [91].

The REMIT assay for breast cancer has shown 90% concordance between ex vivo predictions and in vivo responses in PDX models [92]. This high concordance supports the biological relevance of the assay despite not directly measuring cell death. The reproducibility of 80% for paclitaxel and 83% for eribulin further validates the robustness of this approach for microtubule-targeting agents [92].

Integration with Genomic and Transcriptomic Data

Integrating functional drug sensitivity data with genomic and transcriptomic profiles provides a more comprehensive understanding of therapeutic response mechanisms. For example, in HGSOC, most patients with BRCA mutations demonstrate ex vivo responses to olaparib, as expected [91]. However, a subset of patients without BRCA mutations also responded to olaparib, potentially due to alternative aberrations in DNA repair pathways—highlighting how functional testing can identify patients who might benefit from targeted therapies beyond those predicted by genomic markers alone [91].

Single-cell RNA sequencing of mantle cell lymphoma has revealed that relapse is driven not only by genetic evolution but also by transcriptional heterogeneity and remodeling of the tumor microenvironment [21]. CD70-mediated signaling was identified as a potential contributor to disease progression and relapse, suggesting novel therapeutic targets that might not be evident from genetic analysis alone [21].

Diagram 2: Multi-modal Data Integration for Clonal Evolution Analysis

Functional validation using PDX models for ex vivo drug sensitivity profiling represents a powerful approach for bridging the gap between genomic discoveries and clinical application. The integration of these functional assays with single-cell technologies provides unprecedented resolution for mapping clonal evolution and understanding the dynamics of therapeutic response and resistance. As these methodologies continue to mature, they hold tremendous promise for guiding personalized treatment strategies and developing novel therapeutic approaches that account for tumor evolution.

The field is moving toward increasingly complex assay systems that better recapitulate the tumor microenvironment, including the incorporation of immune components to enable immunotherapy testing. Technological advances in automation and miniaturization are making functional screening more scalable and efficient, potentially reducing turnaround times for clinical application. The integration of functional data with multi-omic profiling using computational models represents the next frontier in precision oncology, offering the potential to predict evolutionary trajectories and design evolution-informed therapeutic strategies that preemptively target resistant clones before they dominate the tumor population.

Conclusion

Single-cell analysis has unequivocally demonstrated that cancers are complex ecosystems governed by Darwinian evolution, where pre-existing and dynamically evolving subclones drive therapeutic failure. The integration of genomic, transcriptomic, and epigenetic data at single-cell resolution is critical to dissect this heterogeneity, revealing not just the 'what' but the 'why' of resistance—from distinct genetic alterations to convergent transcriptional states like inflammatory programs or metabolic shifts. Future research must focus on the real-time clinical application of these insights through liquid biopsy monitoring and the development of evolution-informed therapeutic strategies, such as adaptive therapy or combination treatments targeting both genetic drivers and the resistant cell states they produce. Ultimately, defeating cancer requires moving beyond static genomic snapshots to dynamically intercept and control its evolutionary trajectory.

References

• 1. Cancer evolution: Darwin and beyond - PMC - PubMed Central [https://pmc.ncbi.nlm.nih.gov/articles/PMC8441388/]
• 2. from Darwin to the Extended Evolutionary Synthesis [https://www.sciencedirect.com/science/article/pii/S2405803325000019]
• 3. Single cell deciphering of progression trajectories of the ... [https://www.nature.com/articles/s41467-024-46912-6]
• 4. De Novo Detection of Clonal Structure and Evolution in ... [https://www.mdpi.com/1422-0067/26/23/11428]
• 5. Tracking clonal evolution during treatment in ovarian ... [https://www.nature.com/articles/s41586-025-09580-0]
• 6. Single-Cell RNA Sequencing of Ewing Sarcoma Tumors ... [https://pubmed.ncbi.nlm.nih.gov/40029262/]
• 7. Quantitative Framework for Bench-to-Bedside Cancer ... [https://www.mdpi.com/2072-6694/14/21/5254]
• 8. Revisiting Clonal Evolution Through the Light of ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12632430/]
• 9. Clonal evolution and therapy resistance in the era of ... [https://www.sciencedirect.com/science/article/pii/S1044579X25001257]
• 10. Single-cell multi-omics in cancer immunotherapy: from tumor ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12376342/]
• 11. Decoding leukemia at the single-cell level: clonal architecture ... [https://ehoonline.biomedcentral.com/articles/10.1186/s40164-024-00479-6]
• 12. Single-cell multiomics analysis reveals dynamic clonal ... [https://www.nature.com/articles/s41588-024-01999-x]
• 13. Distinguishing linear and branched evolution given single-cell ... [https://almob.biomedcentral.com/articles/10.1186/s13015-021-00194-5]
• 14. Single-cell analysis of clonal maintenance ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC7610382/]
• 15. Allelic Imbalance and Chromothripsis Lead to Diversity in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12485667/]
• 16. Ongoing genome doubling shapes evolvability and ... [https://www.nature.com/articles/s41586-025-09240-3]
• 17. Genome doubling as a dynamic driver of ovarian cancer ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12628590/]
• 18. Clone copy number diversity is linked to survival in lung ... [https://www.nature.com/articles/s41586-025-09398-w]
• 19. Chromothripsis drives the evolution of gene amplification in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC7933129/]
• 20. Whole-genome doubling drives oncogenic loss of ... [https://www.nature.com/articles/s41586-023-05794-2]
• 21. Tumor evolution and immune microenvironment dynamics in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12490240/]
• 22. Drug resistance and combating drug resistance in cancer [https://pmc.ncbi.nlm.nih.gov/articles/PMC8315569/]
• 23. Clonal interactions in cancer: Integrating quantitative ... [https://www.sciencedirect.com/science/article/abs/pii/S1044579X23000603]
• 24. Quantitative measurement of phenotype dynamics during ... [https://www.nature.com/articles/s41467-025-59479-7]
• 25. The role of single-cell sequencing in studying tumour ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC8170688/]
• 26. Clonal tracking in cancer and metastasis - PMC [https://pmc.ncbi.nlm.nih.gov/articles/PMC11500829/]
• 27. Characterizing the evolutionary dynamics of cancer ... [https://www.nature.com/articles/s41588-024-01989-z]
• 28. Single-Cell RNA Sequencing in Cancer: Lessons Learned ... [https://www.sciencedirect.com/science/article/pii/S1097276519303569]
• 29. Application and research progress of single cell ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC11390353/]
• 30. scGAL: unmask tumor clonal substructure by jointly analyzing ... [https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-024-10319-w]
• 31. To Study Treatment Resistance in High-Grade Serous ... [https://www.mskcc.org/news/to-study-treatment-resistance-in-high-grade-serous-ovarian-cancer-msk-researchers-develop-new-approach]
• 32. New Blood-Based Technology Tracks Tumor Evolution in ... [https://breakthroughcancer.org/2025/10/new-blood-based-technology-tracks-tumor-evolution-in-ovarian-cancer-patients/]
• 33. Blood Test Tracks Treatment Resistance in High-Grade ... [https://www.labmedica.com/immunology/articles/294806894/blood-test-tracks-treatment-resistance-in-high-grade-serous-ovarian-cancer.html]
• 34. New Blood Test Predicts Ovarian Cancer Recurrence [https://www.technologynetworks.com/tn/news/new-blood-test-predicts-ovarian-cancer-recurrence-405367]
• 35. Tracking clonal evolution during treatment in ovarian ... [https://pubmed.ncbi.nlm.nih.gov/41034582/]
• 36. How to map drug resistance in ovarian cancer using ... [https://www.linkedin.com/posts/sohrab-shah-74745489_tracking-clonal-evolution-during-treatment-activity-7379224406927233024-wQnr]
• 37. A Novel Approach to Study Drug Resistance in HGSOC [https://www.linkedin.com/posts/liquid-biopsy-insights_researchers-developed-cloneseq-sv-a-novel-activity-7379576409771409409-r-lN]
• 38. Tracking clonal evolution of drug resistance in ovarian cancer ... [https://pubmed.ncbi.nlm.nih.gov/39229105/]
• 39. Co-mapping clonal and transcriptional heterogeneity in ... [https://pubmed.ncbi.nlm.nih.gov/41075791/]
• 40. Leukemia Progression Mapped by New Multi - Omics Tool, GoT - Multi [https://www.genengnews.com/topics/cancer/leukemia-progression-mapped-by-new-multi-omics-tool-got-multi/]
• 41. Emerging technologies of single-cell multi-omics [https://haematologica.org/article/view/11928]
• 42. Integrated analysis of multimodal single-cell data - PubMed [https://pubmed.ncbi.nlm.nih.gov/34062119/]
• 43. SCClone: Accurate Clustering of Tumor Single-Cell DNA ... [https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2022.823941/full]
• 44. Single-cell multi-omics analysis reveals cancer regulatory ... [https://www.nature.com/articles/s41419-025-08060-7]
• 45. Single-cell multiomics reveals the interplay of clonal ... [https://www.nature.com/articles/s41467-024-47280-x]
• 46. Clonal evolution of acute myeloid leukemia revealed by ... [https://www.nature.com/articles/s41467-020-19119-8]
• 47. Single cell mutation analysis of clonal evolution in myeloid ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC7677169/]
• 48. Liquid biopsy in cancer: current status, challenges and ... [https://www.nature.com/articles/s41392-024-02021-w]
• 49. Liquid biopsy – a narrative review with an update on current ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12333414/]
• 50. Minimal Residual Disease Detection: Implications for ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12079024/]
• 51. testing in haematological and solid cancers | Leukemia [https://www.nature.com/articles/s41375-024-02252-4]
• 52. How MRD Is Being Used to Guide Treatment Decisions [https://www.targetedonc.com/view/how-mrd-is-being-used-to-guide-treatment-decisions]
• 53. Cell-free DNA Testing (Liquid Biopsy) for the Management of ... [https://guidelines.carelonmedicalbenefitsmanagement.com/cell-free-dna-testing-liquid-biopsy-for-the-management-of-cancer-2024-11-17-updated-2025-07-01/]
• 54. Challenges and proposed solutions to the adoption of cell ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12362566/]
• 55. Single Cell Sequencing Market Grows at 15.05% CAGR ... [https://www.towardshealthcare.com/insights/single-cell-sequencing-market-sizing]
• 56. Allele-specific expression analysis: pipelines, applications, ... [https://www.sciencedirect.com/science/article/pii/S0010482525012429]
• 57. Single-cell allele-specific expression analysis reveals ... [https://www.nature.com/articles/s41467-023-42016-9]
• 58. Embracing the dropouts in single-cell RNA-seq analysis [https://www.nature.com/articles/s41467-020-14976-9]
• 59. The impact of dropouts in scRNAseq dense neighborhood ... [https://www.sciencedirect.com/science/article/pii/S2001037025001023]
• 60. Single-cell transcriptomics reveals specific RNA editing ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC5435858/]
• 61. Identification and Interpretation of A-to-I RNA Editing ... [https://www.mdpi.com/1422-0067/24/24/17126]
• 62. Delineating cancer evolution with single cell sequencing [https://pmc.ncbi.nlm.nih.gov/articles/PMC4785805/]
• 63. Single-cell analysis technologies for cancer research [https://pmc.ncbi.nlm.nih.gov/articles/PMC10602688/]
• 64. Co-mapping clonal and transcriptional heterogeneity in ... [https://www.sciencedirect.com/science/article/pii/S2666979X25002927]
• 65. Learning the cellular origins across cancers using single ... [https://www.nature.com/articles/s41467-025-63957-3]
• 66. Transformative advances in single-cell omics - PubMed Central [https://pmc.ncbi.nlm.nih.gov/articles/PMC12560279/]
• 67. Evaluating methods for B-cell clonal family assignment - PMC [https://pmc.ncbi.nlm.nih.gov/articles/PMC11160721/]
• 68. Error-corrected flow-based sequencing at whole genome ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12077166/]
• 69. Clonal expansion of cancer driver gene mutants ... [https://www.sciencedirect.com/science/article/pii/S1383574224000279]
• 70. APPLICATION OF ERROR-CORRECTED SEQUENCING ... [https://www.sciencedirect.com/science/article/abs/pii/S027323002500217X]
• 71. Error-corrected next-generation sequencing mutagenicity ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12477553/]
• 72. Integrative reconstruction of cancer genome karyotypes ... [https://www.nature.com/articles/s41467-021-22671-6]
• 73. Resolving complex structural variants via nanopore ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC10479017/]
• 74. Towards accurate and reliable resolution of structural variants ... [https://genomebiology.biomedcentral.com/articles/10.1186/s13059-022-02636-8]
• 75. in Acute Myeloid Leukemia: A Multicenter... Optical Genome Mapping [https://www.bohrium.com/paper-details/optical-genome-mapping-in-acute-myeloid-leukemia-a-multicenter-evaluation/817367743601836032-6914]
• 76. of structural variants detected by Comparison with... optical mapping [https://pubmed.ncbi.nlm.nih.gov/33983367/]
• 77. of whole Comparison sequencing performance from genome ... fish [https://bmcresnotes.biomedcentral.com/articles/10.1186/s13104-024-07075-1]
• 78. Joint analysis of mutational and transcriptional landscapes ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC10921788/]
• 79. Uncovering Clonal Heterogeneity with Single-Cell ... [https://missionbio.com/resources/learning-center/beyond-bulk-sequencing-uncovering-clonal-heterogeneity-with-single-cell-multiomic-cancer-profiling/]
• 80. Single-cell RNA sequencing reveals different cellular ... [https://www.nature.com/articles/s41523-025-00808-w]
• 81. shaping the evolutionary dynamics of cancer [https://www.sciencedirect.com/science/article/pii/S2405803325001463]
• 82. Germline genetic variation impacts clonal hematopoiesis ... [https://www.nature.com/articles/s41588-025-02250-x]
• 83. Measuring Longitudinal Genome-wide Clonal Evolution of ... [https://pubmed.ncbi.nlm.nih.gov/40166290/]
• 84. - Single Sequencing Applications | Mission Bio cell [https://missionbio.com/resources/learning-center/single-cell-sequencing-applications-in-cancer/]
• 85. Single-Cell Multi-Omic Analysis of AML MRD Reveals ... [https://www.sciencedirect.com/science/article/abs/pii/S0006497124043155]
• 86. Plasma: A Game Changer in Oncology Research and... Longitudinal [https://blog.crownbio.com/longitudinal-plasma-a-game-changer-in-oncology-research-and-biomarker-discovery]
• 87. monitoring of circulating tumor DNA to detect Longitudinal ... relapse [https://link.springer.com/article/10.1007/s10549-024-07508-2]
• 88. Annotation-free discovery of disease-relevant cells in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12383263/]
• 89. Single-cell analysis reveals the chemotherapy-induced ... [https://www.nature.com/articles/s41375-022-01789-6]
• 90. Functional precision oncology: testing tumors with drugs to ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC8752507/]
• 91. Ex vivo 3D micro-tumour testing platform for predicting ... [https://www.nature.com/articles/s41698-025-01080-8]
• 92. Development and validation of a functional ex vivo ... [https://www.nature.com/articles/s41523-025-00734-x]