Idea Ripper // ArchiveCards: 1555 [Live] | Updated: 2026-10-14
Card #c1412IR // #c1412SynthesisThesis

In metagenomic read classification, spaced seeds (k-mers with wildcard positions) beat traditional contiguous k-mers — but with an explicit operating envelope: the advantage holds below ~0.9 sequence identity; for very high-quality alignments (identity ~0.95+) contiguous k-mers with hit-number scoring win. Jaccard-index scoring amplifies the spaced-seed advantage for small-weight seeds.

— In our words · Source: Spaced seeds improve k-mer-based metagenomic classification (Karel Brinda)

[Use when]

Every method has an operating envelope — spaced seeds win at low identity, contiguous k-mers at high identity; the honest answer is a switch, not a champion.

[Why it matters]

A crisp method claim with an explicit operating envelope — metagenomics pipelines (and any fuzzy-matching problem) should switch estimators by identity range rather than picking one winner.