Differential Variant Calling — Comparative Somatic Variant Detection
Differential Variant Calling in Genomics · Also known as: somatic variant calling, comparative variant analysis, tumor-normal variant calling, differential SNV/indel calling
Differential variant calling is a bioinformatics pipeline that identifies genetic variants — single nucleotide variants (SNVs), small insertions/deletions (indels), and structural variants — that are present in one biological sample or condition but absent (or significantly enriched) in a paired reference sample. The canonical application is tumor-versus-normal cancer genomics, where somatic mutations unique to the tumor are distinguished from germline variants shared with normal tissue. The same logic applies to comparing treated vs. untreated cell lines, evolved vs. ancestral strains, or case vs. control cohorts in population genomics.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use differential variant calling whenever you need to identify variants that are specific to one condition relative to a matched control — most commonly somatic mutations in tumor-normal paired cancer sequencing, but also evolved mutations in experimental evolution, de novo mutations in parent-offspring trios, or treatment-induced variants in cell-line experiments. A matched reference sample from the same individual is strongly preferred; without it, false-positive germline variants will dominate the output and genuine differential signals will be obscured. Do NOT use differential variant calling as a substitute for standard germline variant calling when the goal is to characterise an individual's inherited genetic variation. Also avoid it when sequencing depth is below approximately 30x for normal tissue and 60x for tumor, since low coverage severely reduces sensitivity for subclonal variants with low allele frequency.
Strengths & limitations
- Dramatically reduces false-positive rate compared with single-sample calling by using the matched reference to subtract germline and technical artefacts.
- Detects subclonal variants at low allele frequencies (as low as 1–5% VAF in high-coverage data) that would be filtered out in germline pipelines.
- Enables mutational signature analysis and driver-gene identification directly from the somatic call set.
- Well-supported by mature, benchmarked tools (MuTect2, Strelka2, VarScan 2) with extensive community validation.
- Applicable across variant types — SNVs, indels, copy number alterations, and structural variants — within a unified paired framework.
- Requires a high-quality matched reference sample; unpaired analyses are significantly less specific and require larger panels of normals.
- Sensitivity for low-VAF subclonal variants depends on sequencing depth; standard exome depth (50–100x tumor) may miss variants below 10% VAF.
- Tumor heterogeneity, contamination, and copy number alterations complicate allele frequency interpretation and require specialised downstream tools.
- Computational resource demands are substantial — joint analysis of 100GB+ BAM file pairs requires significant CPU and memory infrastructure.
- Does not directly provide functional validation; experimental confirmation (e.g., Sanger sequencing, functional assays) remains necessary for high-priority candidates.
Frequently asked
What is the difference between somatic variant calling and differential variant calling?
Somatic variant calling is the most common application of differential variant calling — it finds mutations present in tumor but not in matched normal tissue. Differential variant calling is the broader concept: comparing any case sample against any reference sample to find condition-specific variants. All somatic calling is differential calling, but differential calling also includes de novo mutation calling in trios, evolution experiments, and case-control comparisons.
Can I do differential variant calling without a matched normal sample?
Yes, but with substantially reduced specificity. Without a matched normal, you must rely on a panel of normals (PON) — a collection of unrelated normal samples — to identify and remove common germline variants and technical artefacts. Tools like MuTect2 support PON-based unpaired calling, but the false-positive rate is higher and sensitivity for low-VAF variants may be reduced. A matched normal from the same individual is always preferable.
What sequencing depth do I need?
For reliable somatic SNV calling, a common guideline is at least 30x mean coverage for the normal sample and 60–100x for the tumor. For detecting subclonal variants at VAF below 10%, ultra-deep sequencing (500x or more) may be needed — particularly for liquid biopsy applications. Coverage requirements also depend on tumor purity: a 30% purity tumor effectively dilutes variant signals threefold.
Which tool should I use — MuTect2, Strelka2, or VarScan 2?
All three are well-validated; choice depends on context. MuTect2 (within GATK) is the current community standard for human tumor-normal pairs and is actively maintained. Strelka2 offers fast, memory-efficient calling well-suited to large cohorts. VarScan 2 remains useful when sample purity and copy number estimates are already available for direct input. For maximum sensitivity, running two callers and taking the intersection or union is common in clinical research settings.
How do I interpret variant allele frequency (VAF) in differential calling?
VAF is the fraction of sequencing reads carrying the alternate allele at a given position. In differential calling, the somatic VAF reflects both tumor purity and clonality: a clonal mutation in a 50% purity tumor will appear at ~25% VAF. To convert VAF to cancer cell fraction (CCF) you need estimates of tumor purity and local copy number. Tools such as PyClone-VI or ABSOLUTE perform this conversion, enabling distinction of clonal driver mutations from subclonal passenger events.
Sources
- Koboldt, D.C., Zhang, Q., Larson, D.E., Shen, D., McLellan, M.D., Lin, L., Miller, C.A., Mardis, E.R., Ding, L., & Wilson, R.K. (2012). VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Research, 22(3), 568–576. DOI: 10.1101/gr.129684.111 ↗
- Cibulskis, K., Lawrence, M.S., Carter, S.L., Sivachenko, A., Jaffe, D., Sougnez, C., Gabriel, S., Meyerson, M., Lander, E.S., & Getz, G. (2013). Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nature Biotechnology, 31(3), 213–219. DOI: 10.1038/nbt.2514 ↗
How to cite this page
ScholarGate. (2026, June 3). Differential Variant Calling in Genomics. ScholarGate. https://scholargate.app/en/bioinformatics/differential-variant-calling
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Copy Number Variation AnalysisBioinformatics↔ compare
- RNA-seq Differential ExpressionBioinformatics↔ compare