modules/hmmer_hmmrank
R script that scores output from multiple runs of hmmer/hmmsearch
Input
Optional per-domain hit tables from hmmsearch (–domtblout). When supplied, the ranked table gains the target and profile lengths and, for each of the three coordinate sets, the outer bounds of the match, the size of the union of the pair’s domains, and the number of separate stretches that union spans. Every domain hmmsearch reported is counted, whatever its own significance: a hit can clear the full-sequence threshold while each of its domains is weak in isolation, and dropping those would leave exactly such hits without coordinates. Domains are grouped by query name as well as by file, so an HMM file holding several models searched in one go still yields one row per (sequence, model).
*.domtbl.gzOutput
TSV file with ranked hmmer results: one row per (accno, profile, profile_desc) combination – profile_desc holds the query name, i.e. the model, which one file may hold several of – with profile, accno, profile_desc, evalue, score and rank. When domtblouts are supplied it also carries tlen and qlen, plus from, to, len and n_islands for each of the hmm, ali and env coordinate sets. n_islands is 1 when the domains merge into a single continuous stretch and rises as they scatter; coordinates are inclusive, so domains that abut with no gap between them count as one. All fourteen are NA where a hit has no domain records behind it: hmmsearch reports a sequence on the per-sequence threshold, while a domain also has to clear the per-domain one, so the two tables need not agree on every hit.
*.hmmrank.tsv.gzincluded in
maintainer
