Concepts

snipe computes QC metrics for sequencing data — coverage, depth, error and mutation rates, contamination, and sex-chromosome signals — from a k-mer sketch, without aligning reads. These pages explain how it works. They build on each other — reading them in order gives you the whole model, but each stands alone if you just need one idea.

Start here

K-mers & sketching in 2 minutesThe minimum background: k-mers, canonical form, and why we sketch.

EdgemersThe k-mer pair (K1 centred in K2) that snipe uses to separate errors from mutations.

FracMinHash & scaleHow snipe samples the hash space so signatures stay comparable.

Hashing & canonical k-mersCanonical murmur3, seed 42 — strand-independent hashing.

Signatures & formats

Sample vs. reference signaturesThe two signature roles and which operations are meaningful between them.

The .snipesig formatParquet-encoded, full-fidelity; the same input produces the same output.

Versioning & compatibilityWhat each encoding can read and write.

sourmash interoperabilityCourtesy interop: round-trip a sketch to/from a sourmash .sig at the K1 level.

The science

How QC worksHow the edgemer pair separates errors from mutations — the reasoning behind the metrics.

Why snipeWhen the edgemer pair helps — and when it doesn't.

Looking for a definition? See the Glossary.