Performing many pairwise alignments

[fbarreto]

Hi all,

I've stumbled into an issue in my analysis for which I haven't been able to think of a single solution.
I have ~4,000 distinct pairs of orthologous sequences, and I would like to have global alignments for each pair. I am not interested in a multiple sequence alignment, but simply distinct pairwise alignments. In papers with similar-sized data sets, authors have briefly mentioned that they used standard alignment programs, like ClustalW, MUSCLE, or T-Coffee. However, short of inputing each pair of sequences at a time into these programs (i.e. thousands of inputs!), I haven't been able to understand how that can be done in those programs. As far as I can tell, the set of sequences one supplies as input will undergo a multiple sequence alignment.

Am I missing something in how to use those programs for this problem? Or is this an issue that must be dealt with Perl programming?

Any insights would be greatly appreciated!

Thanks!

[simonandrews]

This sounds like a situation where you really need a script. You don't say how your data is currently structured, but you'll probably only need a very simple script to automate the submission of 4000 jobs to any alignment program.

Of course your next problem will be what to do with 4000 pairwise alignments, so you'll probably need a script for that too eventually...

[maubp]

Try needleall from EMBOSS 6.2.0 or later for many-to-many Needleman-Wunsch global alignment:

http://lists.open-bio.org/pipermail/emboss/2010-January/003824.html

http://emboss.sourceforge.net/apps/release/6.2/emboss/apps/needleall.html

[fbarreto]

Thanks! That helps a lot! I tried EMBOSS's needleall, and it does wonderfully in aligning pairs, but I still need to do some scripting to parse that output. Eventually, I want to have each alignment in Phylip format for PAML.

Cheers!

[maubp]

The EMBOSS tools support a range of output files, but you may still need some scripting...