One card per tool, best Fmax first: what it compares, whether it will work for your
proteins, where it is strong and weak, what it costs, and the exact command that was run.
Figure 1. Each tool's best Fmax at each of the 15 relatives, one bar per
relative from the nearest (chimpanzee, 6 Mya) to the farthest
(E. coli, 2,000 Mya): Pfam domains, raw calls, all proteins, best
over the tool's settings. Every card uses the same scale, 0 to 0.45. Hover a bar for its relative and number.
aligns amino-acid letters: phmmer, MMseqs2builds a profile of a protein family: jackhmmer, MMseqs2, three rounds, HH-suite (hhsearch)compares 3-D shape letters (3Di): Foldseek, ProstT5 + Foldseekcounts shared k-mers in a reduced alphabet: kmerseekscans curated Pfam family models: hmmscan with Pfam
0.333 best Fmax$0.0186 per 1,000 queriesPfam, raw calls, zebrafish; default options
Amino-acid letters, one query protein against each protein of the relative, scored with a substitution matrix the way BLAST does but with HMMER's model of insertions and deletions.
0.327 best Fmax$0.00113 per 1,000 queriesPfam, raw calls, zebrafish; sensitivity 7
Amino-acid letters: it first keeps target proteins that share similar short words with the query, then aligns only those, which makes it much faster than aligning every pair.
0.290 best Fmax$0.109 per 1,000 queriesPfam, raw calls, zebrafish; 3 rounds
A profile of a protein family: it searches with one protein, builds a profile from the hits, and searches again with that profile, three rounds in all.
0.223 best Fmax$0.00486 per 1,000 queriesPfam, raw calls, zebrafish; shape letters from sequence
Shape letters (3Di) predicted from the amino-acid sequence by the ProstT5 protein language model, then searched with Foldseek, so no AlphaFold model is needed.