Doesn't k-mer imply that the sequence has to be an exact match?
There are many motif-finding tools. Usually they compare a set of sequences against a set of motifs and give you the top occurring motifs. However, I have a specific motif in mind and I would like to find the positions where it occurs in the genome (account for mismatches, of course). Is there a tool that will do that?
2 answers
Recent thread that may work: Finding specific k-mer in human genome
fuzznuc, EMBOSS program referenced in the thread allows you to specify search patterns (with mismatches): http://emboss.sourceforge.net/apps/cvs/emboss/apps/fuzznuc.html I assume you know the sequence of the motif you want to search. You can specify search patterns based on that.
The command-line version of UCSC BLAT can be used locally with -minMatch and -minIdentity options to account for mismatches. It exports a PSL file, which contains position information and can be converted into other formats for operations.
I think blast also should work, doesn't it?
If not what might be the possible problems
I think BLAT will give you a bit more control over the number of allowed mismatches and other settings. It also directly outputs positional information, which I do not believe blast does without extra work.
I think BLAT needs a lot more modifications. Setting -minIdentity=60 and -minMatch=1 fails to match a 20bp sequence with just 1 mismatch.
Log in to answer this question.