Variant Allele Frequency (VAF) Calculator
Variant Allele Frequency (VAF) Calculator
Variant reads as a percentage of total reads at a position — and what a VAF away from 50% or 100% is telling you, from mosaicism and copy-number change to clonal haematopoiesis in a blood sample used as a germline reference.
Variant Allele Frequency (VAF)
Variant reads, depth → VAF %47 variant reads out of 100 total reads at the position
Formula
- variant reads
- reads supporting the alternate allele at that position, after duplicate removal and quality filtering. Duplicates are copies of one original molecule, so counting them twice creates confidence that is not there
- total reads
- the depth at that specific position, reference-supporting plus variant-supporting. Not the mean depth across the target: coverage is uneven, and the depth at one position can be a small fraction of the mean
- ≈ 50%
- the expectation for a germline heterozygous variant, because half the molecules at the position carry it. Real heterozygous sites scatter around 50% through sampling alone, by several percentage points at typical depths
- ≈ 100%
- the expectation for a homozygous germline variant, a hemizygous variant on the X chromosome in a male sample, or a mitochondrial variant at or near homoplasmy
- why depth sets precision
- THE POINT OF THE PAGE: VAF is a ratio of counts, so its precision comes entirely from the denominator. Two reads of four is 50% and means almost nothing; two hundred of four hundred is the same 50% and carries real weight. A VAF reported without its depth is uninterpretable
Worked example
47 variant reads out of 100 total reads at the position
VAF = 47 ÷ 100 × 100 = 47.0%
That sits close to the 50% expected of a germline heterozygous variant, and the departure from exactly 50% is unremarkable — sampling 100 molecules from a 50:50 mixture lands near 47 routinely
Now hold the ratio and drop the depth. 2 variant reads of 4 total is also 50.0%, and it is worth almost nothing: one read either way would have given 25% or 75%
Hold the ratio and raise the depth. 200 of 400 is again 50.0%, and now one read either way moves the figure by a quarter of a percentage point
The same 47% at a depth of 8 (roughly 4 of 8, 50.0%) and at a depth of 1,000 (470 of 1,000, 47.0%) are utterly different pieces of evidence, which is why a VAF must never be reported without the depth it came from
And 47% would mean something quite different in a tumour sample with 40% tumour content, where a fully clonal heterozygous somatic variant would be expected nearer 20%
What a VAF away from 50% or 100% suggests
| VAF pattern | Candidate explanations |
|---|---|
| Near 50% | Germline heterozygous variant. Also a fully clonal heterozygous somatic variant in a near-pure tumour sample |
| Near 100% | Germline homozygous variant; a hemizygous variant on the X chromosome in a male sample; a near-homoplasmic mitochondrial variant; or a heterozygous variant where the other allele has been deleted |
| Clearly between 50% and 100% | Loss of the wild-type allele — deletion or copy-neutral loss of heterozygosity — or allelic imbalance from a copy-number gain. Check the copy-number data for the region |
| Clearly below 50% | A somatic variant diluted by normal cells; post-zygotic mosaicism; circulating tumour DNA in plasma; a subclone; or allele dropout from a variant under a primer or probe |
| Low VAF in a blood sample used as a germline reference | Clonal haematopoiesis. An expanded blood cell clone carrying a somatic variant appears at low VAF in a sample being treated as germline, and in tumour-normal comparison it can be mistaken for a germline finding or make a genuine tumour variant look germline |
| Low VAF, no biological candidate | Contamination from another sample, index hopping in a pooled run, or a library preparation artefact — oxidative damage, or deamination in formalin-fixed tissue |
The same 50% at four depths
| Variant reads / depth | VAF | One read either way moves it by | Worth |
|---|---|---|---|
| 2 / 4 | 50.0% | 25 percentage points | Almost nothing — the position is barely covered |
| 10 / 20 | 50.0% | 5 percentage points | Enough to suggest a heterozygous pattern, not to quantify it |
| 50 / 100 | 50.0% | 1 percentage point | A usable allele fraction |
| 200 / 400 | 50.0% | 0.25 percentage points | A precise allele fraction |
A ratio of counts, and what it is a ratio of
Variant allele frequency is the simplest calculation in sequence analysis: the reads supporting the variant, divided by the total reads at that position. Two details in that sentence do real work. The reads must be counted after duplicate removal, because duplicates are copies of a single original molecule and counting them as independent observations manufactures confidence. And the denominator must be the depth at that specific position, not the mean depth across the target — coverage is uneven, and the depth at one base can be a small fraction of the average.
Interpretation starts from two expectations. A germline heterozygous variant sits near 50%, because half the molecules at that position carry it; a germline homozygous variant sits near 100%. A VAF far from both is therefore a finding in itself, and the list of explanations is short enough to work through. A somatic variant appears at a fraction set by how much of the sample is tumour and how clonal the variant is. Post-zygotic mosaicism gives a fraction set by the proportion of cells in that tissue carrying the variant. A copy-number change shifts the ratio by changing how many copies of each allele are present, and loss of the wild-type allele pushes a heterozygous variant towards 100%. Contamination and index hopping in pooled runs generate low-fraction reads that are real sequence from the wrong sample.
One explanation deserves particular attention because it is routinely missed. Blood is the usual source of a germline reference, and blood is not a germline tissue in the way the assumption requires: with age, expanded haematopoietic clones carrying somatic variants become common. Clonal haematopoiesis presents as a low-VAF variant in a sample being treated as germline. In tumour-normal comparison it can make a genuine somatic tumour variant look germline, because it is also present in the normal, or it can be reported as a germline predisposition variant it is not. The genes involved overlap heavily with those on myeloid and hereditary cancer panels, so this is not a rare curiosity.
The final point is about precision, and it is the one most often lost. VAF is a ratio of counts, so how much it can be trusted depends entirely on the denominator. Two variant reads out of four is 50%, and it means almost nothing: one read either way would have made it 25% or 75%. Two hundred of four hundred is the same 50% and carries real weight, because a single read moves it by a quarter of a percentage point. A VAF quoted without its depth cannot be interpreted at all, and a difference in VAF between two samples or two time points is only a difference if the depths behind both can support it. Depth also sets the floor: below a few per cent, a variant call competes with the background error rate of the chemistry, and distinguishing a real low-fraction variant from an artefact needs both very high depth and an assay validated to a stated limit of detection.
Frequently asked questions
How do you calculate variant allele frequency?
Divide the number of reads supporting the variant by the total number of reads at that position and multiply by 100. Forty-seven variant reads out of a depth of 100 gives a VAF of 47.0%. Use read counts after duplicate removal, and the depth at that specific position rather than the mean depth.
What VAF is expected for a germline variant?
About 50% for a heterozygous variant, since half the molecules at the position carry it, and about 100% for a homozygous one — or for a hemizygous variant on the X chromosome in a male sample. Sampling scatters real heterozygous sites around 50% by several percentage points.
What does a VAF of 30% mean?
That the variant is present in a minority of molecules, and the explanation depends on the sample. Candidates are a somatic variant diluted by normal cells, post-zygotic mosaicism, a subclone, allelic imbalance from a copy-number change, or allele dropout. In a blood sample sent as a germline reference, clonal haematopoiesis should be considered.
Can clonal haematopoiesis affect a germline result?
Yes. Expanded blood cell clones carrying somatic variants become common with age, so a low-VAF variant in a blood sample being used as a germline reference may be clonal haematopoiesis rather than germline. In tumour-normal comparison it can make a somatic variant look germline, and the genes involved overlap with myeloid and hereditary cancer panels.
Why does depth matter for VAF?
Because VAF is a ratio of counts and its precision comes entirely from the denominator. Two of four reads is 50% and means almost nothing — one read either way gives 25% or 75%. Two hundred of four hundred is the same 50% but moves only a quarter of a percentage point per read. Always report depth alongside VAF.
Related calculators
References
- Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17(5):405–424.
- Steensma DP, Bejar R, Jaiswal S, et al. Clonal hematopoiesis of indeterminate potential and its distinction from myelodysplastic syndromes. Blood. 2015;126(1):9–16.
- Jaiswal S, Fontanillas P, Flannick J, et al. Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371(26):2488–2498.
- Li MM, Datto M, Duncavage EJ, et al. Standards and guidelines for the interpretation and reporting of sequence variants in cancer. J Mol Diagn. 2017;19(1):4–23.
Medical Disclaimer: The tools and content provided here are for educational and reference purposes only. They are not intended to substitute for professional medical advice, diagnosis, or treatment. Clinical decisions should always be based on the comprehensive assessment of a qualified healthcare professional.