IUPAC module generated regexs will not match Ns in the target about bioperl-live HOT 11 CLOSED

@passdan That is by design; the bases in those positions for the 'probe' is a defined sequence. In order to get approximate matches you will need to use a module like Text::Fuzzy or String::Approx.

from bioperl-live.

@passdan Try Bio::Regexp

from bioperl-live.

@tsibley interesting, that may be a good internal replacement for IUPAC or SeqPattern

from bioperl-live.

Yeah, I've been wondering about such things. It looks quite nice.

from bioperl-live.

Thanks for the advice. As soon as I got the first reply I realised how obviously wrong it would have been to match in that direction!

I'm considering the best way of incorporating the fuzzy or approx matching. At the moment I take an input IUPAC-compliant 'probe' sequence, generate a regex and perform a match:

    if ( $seq->seq =~ m/^(.*)($regexp)(.*)/i ){

but the fuzzy matching tools won't accept a regular expression. Would the best method be to create an array of the possible sequences then loop through them and fuzzy match each? I imagine that this could increase the compute time significantly, but is there a better way?

Thanks

from bioperl-live.

@passdan Did you look at Bio::Regexp? It'll do fuzzy matching from an IUPAC-compliant string or your (constrained) regular expression.

my $probe   = Bio::Regexp->new->dna->add( $IUPAC_string );
my @matches = $probe->match( $seq->seq );

from bioperl-live.

I had a look at that, but I can't see the fuzzy matching reference. From what I see it will only match a defined sequence or regex and not allow any substitutions.

I've been looking at the Text::Approx and it seems appropriate, where Ix, Dx and Sx allow x number of insertion, deletion or substitutions:

                my $index = aindex($probe, ["I0","D0","S1"], $ref);

Unfortunately, 0 everything seems to work but once I apply S1 it matches much more indiscriminatingly than a single base sub.

Sorry, I know this is supposed to be a bug report channel and I'm over running it.

from bioperl-live.

Ah, you're right. Apologies for not reading more closely.

You might look at Bio::Grep with the Vmatch backend?

from bioperl-live.

That does look good. I'll look into it, thanks.

from bioperl-live.

@passdan Out of curiosity, I turned your musing of "generating all possible sequences and matching on those" on its head and inverted the match sense while generating all possible probes. The idea is that probes are shorter and much more likely to have fewer degenerate bases. This was mostly just an exercise in curiosity, but here it is: https://gist.github.com/tsibley/15edab0b97683c3ce9fe

There are some obvious improvements that could be made in the name of memory efficiency, but they slightly complicate the clarity of the approach.

from bioperl-live.

@tsibley If we could only +1 a response 👍

Marking as closed.

from bioperl-live.