Add option to define identity as including terminal gaps about vsearch HOT 3 CLOSED

Seems like USEARCH used to have an option --iddef for defining identity, but it has been removed. It was present at least in version 4.1. See pages 42-44 in the Usearch 4.1 User Guide. It allows 5 different definitions of identity to be used (0-4). I think the current defintion used in USEARCH is --iddef 2 (internal diffs). I assume that a similar definition including terminal gaps is --iddef 1 (all-diffs).

from vsearch.

As we only do global alignment, the different definitions listed in usearch v4.1 documentation could be simplified a bit. Right now, we use --iddef 2 in vsearch. One could argue that this definition is more suitable for local alignments. The definition used by CD-HIT (--iddef 0) is unusual. The definition used by the Marine Biology Lab (--iddef 3) is interesting as it assumes long gaps to be single evolutionary events. For a global alignment, --iddef 1 is indeed the most usual definition (--iddef 4 is equivalent).

from vsearch.

Fixed in VSEARCH 0.1.0. Seems like --iddef 1 considerably improves both recall and especially precision on the Rfam classification test.

from vsearch.