Tool for counting mismatches in hundreds of files?

[pepperoni]

Hello, I need to align 2 sequences in hundreds of files and then write a list with the count of the mismatches in each file.

I thought I could use muscle for the alignments as it can run in command line and write a script to run all my files but then how do I count the mismatches of each alignment?

since they are just two protein sequences in each file I cannot use bowtie or similar. I also thought that maybe there is a way to convert a phylip or clustal format into sam so I can use some snp calling tools.

Does any one know how can I align and count the mismatches of all my files?

[rflrob]

This snippet of python might do the trick for you, for the simplest possible definition of "counting mismatches", where that's any column where it's not all identical. It assumes you have BioPython installed, and that you've done your alignments in Clustal format (MUSCLE claims to be able to output in this format, and I have no reason to disbelieve it).

from Bio import AlignIO
from glob import glob

for file in glob('*.aln'):
    a = AlignIO.read(file, 'clustal')
    print file, len([1 for i in range(a.get_alignment_length())
                     if len(set(a[:,i])) != 1])
    # Iterate over the alignment in a list comprehension, and only keep 
    # elements where there's more than one value.

[Richard Finney]

"blat" by Jim Kent of UCSC is a good tool.

There are various methods for structuring a command batch file to process (input data) files.

blat (PSL format) output contains a count of the number of mis matches.

[pepperoni]

Many thanks to both of you for your quick replies. I have just tried rflrob's python script, and it works great

This snippet of python might do the trick for you, for the simplest possible definition of "counting mismatches", where that's any column where it's not all identical. It assumes you have BioPython installed, and that you've done your alignments in Clustal format (MUSCLE claims to be able to output in this format, and I have no reason to disbelieve it).

from Bio import AlignIO
from glob import glob

for file in glob('*.aln'):
    a = AlignIO.read(file, 'clustal')
    print file, len([1 for i in range(a.get_alignment_length())
                     if len(set(a[:,i])) != 1])
    # Iterate over the alignment in a list comprehension, and only keep 
    # elements where there's more than one value.