BUG: -p (palindromic matches) and -e (edit distance) features have strange relationship.

[alexlenail]

It seems the -p (palindromic matches) and -e (edit distance) features are incompatible.

$ vmatch -l 15 -d index     ## length 15, direct matches
--> 5 matches
$ vmatch -l 15 -d -e 5 index    ## length 15, matches up to edit distance of 5
--> 661435 matches

$ vmatch -l 15 -p index     ## length 15, palindromic (i.e. reverse complement) matches
--> 3 matches
$ vmatch -l 15 -p -e 5 index    ## length 15, palindromic matches up to edit distance of 5
--> 0 matches

Am I making a mistake with the syntax?

[alexlenail]

I've narrowed in on this rather peculiar issue:

vmatch -l 15 -p -e 2 index
--> 765 matches

vmatch -l 15 -p -e 3 index
--> 0 matches

vmatch -l 10 -p -e 1 index
--> 765 matches

vmatch -l 10 -p -e 2 index
--> 0 matches

vmatch -l 20 -p -e 3 index
--> 62 matches

vmatch -l 20 -p -e 4 index
--> 0 matches

There is some relationship between -l and -p -- only some ranges are searchable.

[alexlenail]

There seems to be a relationship between whether this search is allowed, and the prefixlength (-pl) set during mkvtree: decreasing the prefixlength from 5 to 3 allowed me to get results for the query:

vmatch -l 18 -p -e 5 index