Subclonal architecture is the inferred evolutionary tree of genetically distinct tumor cell populations—called subclones—coexisting within a single patient. It is reconstructed by analyzing the variant allele frequency (VAF) distribution of somatic mutations across multiple genomic loci, where each subclone's cellular prevalence determines the expected fraction of sequencing reads bearing its defining mutations.
Glossary
Subclonal Architecture
The inferred population structure of genetically distinct tumor cell clusters within a patient, reconstructed from the distribution of variant allele frequencies across multiple mutations.
TUMOR HETEROGENEITY
What is Subclonal Architecture?
Subclonal architecture defines the population structure of genetically distinct tumor cell clusters within a single patient, reconstructed computationally from variant allele frequencies.
Reconstructing this architecture reveals the temporal order of mutational acquisition and identifies clonal mutations (present in all cancer cells) versus subclonal mutations (present in a subset). This distinction is critical for understanding drug resistance, as minor subclones harboring resistance-conferring alterations can expand under therapeutic selective pressure, driving relapse.
TUMOR HETEROGENEITY
Key Characteristics of Subclonal Architecture
Subclonal architecture defines the evolutionary history of a cancer, reconstructed from the distribution of variant allele frequencies. These characteristics are essential for understanding drug resistance and metastatic potential.
01
Cancer Cell Fraction (CCF)
The proportion of tumor cells within a sample that harbor a specific somatic mutation. Unlike raw Variant Allele Frequency (VAF), CCF corrects for tumor purity and local copy number alterations.
- Clonal mutations: CCF ≈ 100%, present in all cancer cells.
- Subclonal mutations: CCF < 100%, present in a subset.
- Calculation requires integrating absolute copy number and purity estimates.
02
PyClone-VI Dirichlet Clustering
A Bayesian nonparametric method that clusters mutations into subclones based on their co-varying cellular prevalence across multiple sequenced regions.
- Uses a beta-binomial emission model to account for variable read depth.
- Infers the number of clusters directly from the data without pre-specification.
- Outputs posterior densities of cellular prevalence for each mutation cluster.
03
Phylogenetic Tree Reconstruction
The process of inferring the ancestral relationships between subclones, typically rooted at the normal germline and branching out to distinct tumor populations.
- Branching evolution: Subclones diverge from a common ancestor.
- Linear evolution: Sequential acquisition of mutations.
- Tools like PhyloWGS integrate somatic SNVs and CNAs to build a unified tree.
04
Multi-Region Sequencing
The physical sampling and sequencing of distinct geographical areas of a primary tumor and its metastases to resolve spatial heterogeneity.
- Reveals private mutations unique to specific regions.
- Essential for distinguishing truncal (early, ubiquitous) from branch (late, localized) mutations.
- Prevents underestimation of subclonal diversity caused by single-biopsy analysis.
05
Mutation Multiplicity Estimation
Determining the number of chromosomal copies that carry a specific mutation. A mutation on two copies in a diploid region has a multiplicity of 2.
- Critical for mapping VAF to CCF.
- A VAF of 0.5 in a pure tumor could represent a clonal heterozygous mutation (multiplicity 1) or a subclonal homozygous mutation.
- Resolved by integrating allele-specific copy number calls.
06
Neutral Evolution Detection
Identifying tumors where subclonal mutations accumulate neutrally without strong selective sweeps, resulting in a characteristic power-law distribution of VAFs.
- A subclonal cluster with a high number of low-CCF mutations suggests neutral drift.
- Contrasts with positive selection, where a single clone expands rapidly.
- Measured using the cumulative distribution of mutation cancer cell fractions.
SUBCLONAL ARCHITECTURE
Frequently Asked Questions
Explore the foundational concepts of subclonal architecture, the computational reconstruction of genetically distinct tumor cell populations that drives understanding of tumor evolution, heterogeneity, and therapeutic resistance.
Subclonal architecture is the inferred population structure of genetically distinct tumor cell clusters within a single patient, reconstructed computationally from the distribution of variant allele frequencies (VAFs) across multiple somatic mutations. It is inferred by clustering mutations that share similar cancer cell fractions (CCFs) using Bayesian Dirichlet processes or variational inference models. The core principle is that mutations arising in a founder cell will be present in all its descendants (clonal), while mutations acquired later define subclonal populations. Algorithms like PyClone, PhyloWGS, and DPClust integrate VAFs, copy number alterations, and tumor purity estimates to deconvolve these clusters and reconstruct the phylogenetic tree of tumor evolution.
CONCEPTUAL DISTINCTIONS
Subclonal Architecture vs. Related Concepts
How subclonal architecture reconstruction differs from related genomic analysis concepts in liquid biopsy and tumor sequencing.
| Feature | Subclonal Architecture | Variant Allele Frequency | Clonal Hematopoiesis Filter | Mutational Signature |
|---|---|---|---|---|
Primary Objective | Reconstruct tumor population structure and evolutionary history | Quantify proportion of mutated DNA molecules at a locus | Exclude blood-derived somatic variants from tumor calls | Identify mutagenic processes from nucleotide substitution patterns |
Input Data Type | Multi-sample or multi-region VAF distributions across many mutations | Single-locus read counts for reference and alternate alleles | Matched buffy coat sequencing or population databases | Catalog of somatic mutations with trinucleotide contexts |
Key Output | Phylogenetic tree of tumor subclones with cancer cell fractions | A percentage value (e.g., 12.5% VAF) | Binary flag: variant is CHIP-derived or tumor-derived | Deconvolved signature activities and proposed etiologies |
Temporal Resolution | Reconstructs past evolutionary dynamics and branching order | Snapshot of current allele frequency only | Distinguishes hematopoietic past from solid tumor present | Integrates lifetime mutational history into a single profile |
Dependency on Tumor Purity | Requires estimation of tumor purity to convert VAF to cancer cell fraction | Confounded by tumor purity and copy number | Independent of tumor purity; relies on matched normal | Independent of tumor purity |
Primary Algorithm Class | Bayesian Dirichlet processes, PyClone, PhyloWGS | Simple binomial or beta-binomial models | Likelihood ratio tests, random forest classifiers | Non-negative matrix factorization, expectation-maximization |
Clinical Application | Predicting treatment resistance from minor subclones | Monitoring tumor burden and treatment response | Preventing false-positive liquid biopsy results | Identifying homologous recombination deficiency or smoking signature |
Requires Multi-Region Sampling |
Enabling Efficiency, Speed & Accuracy
Intelligent Analysis, Decision & Execution
We build AI systems for teams that need search across company data, workflow automation across tools, or AI features inside products and internal software.
Talk to Us
Search across company data
Give teams answers from docs, tickets, runbooks, and product data with sources and permissions.
Useful when people spend too long searching or get different answers from different systems.
Enterprise searchRAGPermissions
Read more
Automate internal workflows
Use AI to route work, draft outputs, trigger actions, and keep approvals and logs in place.
Useful when repetitive work moves across multiple tools and teams.
AI agentsWorkflow automationGovernance
Read more
Add AI to products and internal tools
Build assistants, guided actions, or decision support into the software your team or customers already use.
Useful when AI needs to be part of the product, not a separate tool.
AI integrationDecision supportModel routing
Read more
About the author
Prasad Kumkar
CEO & MD, Inference Systems
Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.
His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.
LinkedIn
Limited slotsGet a Free AI Consultation
Partnered with leading AI, data, and software stack.
How We Work
Custom AI workflows for your Business
One-fit-all AI don't work for modern businesses. At Inferensys, we aim to understand your business & custom requirements; which we use to define most efficient agentic workflows, the data, and the tools for your business.
01
Review the use case
We understand the task, the users, and where AI can actually help.
Read more02
Pick the right approach
We define what needs search, automation, or product integration.
Read more03
Build the first useful version
We implement the part that proves the value first.
Read more04
Improve from there
We add the checks and visibility needed to keep it useful.
Read moreThe first call is a practical review of your use case and the right next step.
Talk to Us