What does k (minimum supporting reads) mean?

The smallest number of alt-supporting reads your caller needs to make a call. Higher k is more stringent (fewer false positives) but needs more depth for the same power.

Does this account for sequencing errors?

No. It models sampling of true alt molecules only. Real callers also weigh base quality and error rates, which matter most when k is small and VAF is near the error floor.

Variant detection power

Compute the probability of seeing at least k alternate reads for a variant at a given allele fraction and depth — or the minimum depth to reach a target power.

How it works

Formula

Alt-read count is Binomial(depth, VAF). Detection power = P(X ≥ k) = 1 − P(X ≤ k−1). The inverse search returns the smallest depth whose power meets the target.

Worked example

At 10× depth, VAF 0.5 and k = 3: P(X ≤ 2) = (1 + 10 + 45)/1024 ≈ 0.055, so power ≈ 94.5%. Raising depth raises power.

When to use it

To plan depth for calling low-frequency variants — somatic mutations, mosaicism, cfDNA — where you need enough alt reads to call confidently, and to justify a sequencing depth target.

Sensible defaults

Defaults ask for the power to see a 5% VAF variant with ≥3 alt reads at 100×. Switch the mode to get the minimum depth for a target power instead.

FAQ

What does k (minimum supporting reads) mean?: The smallest number of alt-supporting reads your caller needs to make a call. Higher k is more stringent (fewer false positives) but needs more depth for the same power.
Does this account for sequencing errors?: No. It models sampling of true alt molecules only. Real callers also weigh base quality and error rates, which matter most when k is small and VAF is near the error floor.