Single-cell variant calling — detecting mutations at cellular resolution
Single-Cell Genomic Variant Calling · Also known as: scVariant calling, single-cell SNV calling, scDNA-seq variant detection, single-cell somatic mutation calling
Single-cell variant calling is a bioinformatics pipeline that identifies DNA sequence variants — single-nucleotide variants (SNVs), small insertions and deletions, and copy-number alterations — within individual cells rather than across a bulk tissue mixture. By resolving the mutational landscape cell by cell, it reveals intra-tumoral heterogeneity, clonal architecture, and somatic mutation patterns that bulk sequencing obscures. The approach is central to cancer genomics, developmental biology, and any study where cell-to-cell genetic diversity is the primary question.
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 single-cell variant calling when the scientific question requires resolving genetic heterogeneity at cellular resolution: mapping clonal evolution in tumors, tracing somatic mutations during development, characterizing mosaic genetic disorders, or linking genotype to transcriptional state. It is the right tool when bulk sequencing would average away minority subclones or when per-cell mutation-to-phenotype associations are needed. Do not use it as a drop-in replacement for bulk variant calling when only population-level allele frequencies are needed — bulk sequencing is cheaper, provides far higher per-locus coverage, and has lower false-positive rates. Also avoid it when amplification quality is poor (coverage <5x per cell) or when the number of cells is too small to model dropout rates reliably.
Strengths & limitations
- Resolves intratumoral and intercellular genetic heterogeneity that is invisible to bulk sequencing.
- Enables direct reconstruction of clonal evolution and phylogenetic relationships among individual cells.
- Can link somatic mutation status to gene-expression phenotype when paired with scRNA-seq.
- Probabilistic models explicitly account for whole-genome amplification artefacts such as allelic dropout.
- Applicable across diverse contexts: cancer, development, aging, and mosaic genetic disease.
- Whole-genome amplification introduces allelic dropout and uneven coverage, leading to higher false-negative rates than bulk sequencing.
- Per-cell sequencing depth is typically shallow, limiting sensitivity for detecting low-frequency somatic variants.
- Computationally expensive: joint modeling across hundreds to thousands of cells requires substantial memory and CPU time.
- High cost per variant discovery compared to bulk approaches when only population-level allele frequencies are needed.
- scRNA-seq-based variant calling is confounded by RNA editing, splicing variation, and the absence of intronic sequence.
Frequently asked
Can I use GATK HaplotypeCaller for single-cell data?
GATK HaplotypeCaller was designed for bulk sequencing and does not model whole-genome amplification artefacts such as allelic dropout. Applied naively to single-cell data it produces high false-negative rates at heterozygous sites. It can be used as a preliminary caller on scRNA-seq data with careful downstream filtering, but dedicated single-cell tools such as Monovar, SCcaller, or Cellsnp-lite are strongly preferred for scDNA-seq.
What is allelic dropout and why does it matter?
Allelic dropout (ADO) occurs when one of the two chromosomal copies at a heterozygous locus fails to amplify during whole-genome amplification, so only one allele appears in the sequencing data. The cell then looks homozygous even though it is not. Single-cell variant callers model ADO rates explicitly — typically estimating them from the data — so that heterozygous sites are not systematically miscalled as homozygous reference or homozygous alternate.
How many cells are needed for reliable variant calling?
There is no hard minimum, but joint statistical models that share information across cells become more accurate as cell count increases. Studies typically work with dozens to hundreds of cells for targeted panels and thousands for droplet-based platforms. With fewer than about 20 cells, per-cell genotype uncertainty is high and clonal inference is unreliable; enriching for the cell population of interest (e.g., tumor cells) before sequencing improves effective sample size.
Can single-cell variant calling be applied to scRNA-seq data?
Yes, but with caveats. scRNA-seq covers only expressed regions, misses intronic and intergenic variants, and is affected by RNA editing and splicing variation. Tools such as Cellsnp-lite, GATK, and vartrix have been adapted for RNA-based variant calling. The approach is most useful for genotyping known variants (e.g., assigning cells to donors in multiplexed experiments) rather than de novo somatic mutation discovery.
How do I distinguish somatic mutations from germline variants in single-cell data?
Cross-reference all called variants against population germline databases (dbSNP, gnomAD). Variants present in every cell with high confidence are likely germline; variants restricted to a subset of cells and absent from databases are candidate somatic events. Matched normal tissue (blood or saliva) from the same individual provides the most reliable germline baseline for filtering.
Sources
- Zafar, H., Wang, Y., Nakhleh, L., Navin, N., & Chen, K. (2016). Monovar: single-nucleotide variant detection in single cells. Nature Methods, 13(6), 505–507. DOI: 10.1038/nmeth.3835 ↗
- Singer, J., Ruscheweyh, H. J., Doherr, M. G., Stadler, T., & Althaus, C. L. (2021). Single-nucleotide variant calling in single-cell sequencing data with piccolo. BMC Bioinformatics, 22(1), 333. link ↗
How to cite this page
ScholarGate. (2026, June 3). Single-Cell Genomic Variant Calling. ScholarGate. https://scholargate.app/en/bioinformatics/single-cell-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