Expand description
MMseqs2 aligner (protein-protein).
Mirrors gapseq’s explicit 4-command pipeline from
src/gapseq_find.sh:603–615:
mmseqs createdb TARGET targetDB
mmseqs createdb QUERY queryDB
mmseqs search queryDB targetDB resultDB tmp --threads N -c 0.C
mmseqs convertalis queryDB targetDB resultDB out.tsv
--format-output "qheader,pident,evalue,bits,qcov,theader,tstart,tend"Rationale: mmseqs easy-search would collapse the four steps into one,
but it defaults to align --alignment-mode 3 (full-alignment identity)
whereas search reports the k-mer prefilter identity. Identity cutoffs
downstream (analyse_alignments.R) were calibrated against the latter —
so we match that semantics exactly. Native coverage is a 0–1 fraction;
we rescale to 0–100 via crate::tsv::parse_tsv.