[Biopython] Trimming adaptors sequences

[Peter]

Hi all,

Brad's got an interesting blog post up on using Biopython for trimming
adaptors for next gen sequencing reads, using Bio.pairwise2 for
pairwise alignments between the adaptor and the reads:

http://bcbio.wordpress.com/2009/08/09/trimming-adaptors-from-short-read-sequences/

The basic idea is similar to what Giles Weaver was describing last
month, although Giles was using EMBOSS needle to do a global pairwise
alignment via BioPerl:
http://lists.open-bio.org/pipermail/biopython/2009-July/005338.html

We already had a simple FASTQ "primer trimming" example in the
tutorial, which I have just extended to add a more general FASTQ
"adaptor trimming" example. For this I am deliberately only looking
for exact matches. This is faster of course, but it also makes the
example much more easily understood as well - something important for
an introductory example.

A full cookbook example of how to use pairwise alignments would seem
like a great idea for a cookbook entry on the wiki. It would be
interesting to see which is faster - using EMBOSS needle/water or
Bio.pairwise2. Both are written in C, but using EMBOSS we'd have the
overhead of parsing the output file.

Brad - why are you using a local alignment and not a global alignment?
Shouldn't we be looking for the entire adaptor sequence? It looks like
you don't consider the the unaligned parts of the adaptor when you
count the mismatches - is this a bug? I wonder if it would be simpler
(and faster) to take a score based threshold.

Regards,

Peter