Problems With The Xml Output Of Psi-Blast Via Qblast When Using Biopython

Question

[eric.lubeck]

I'm trying to use Biopython to do a psi-blast search via the internet for homologous proteins. Unfortunately I'm having several problems with this that I could use some help with.

1.

The returned XML files seem to be abruptly truncated. When Biopython attempts to parse these files it gets XML reading errors. When I examine these files by hand I see that most of the files do not have all the XML tags closed. Most files end at a </hsp> tag without closing off the iteration tags. 2.

I can't find an argument to specify the # of iterations psi-blast should perform in the Biopython qblast function.

I'd appreciate any help. I imagine my description might be too vague. If so, I'd be happy to post a more detailed description once i get out of the microscope room.

I'm using version 1.62 on the newest OSX.

Thanks

[Istvan Albert]

show the parts of the program that saves the XML files.

[Venky]

Are you using remote blast? I do get some similar problems when doing a remote blast, it seems to be solved when you make a local copy of the DB or blast against a local DB.

Answer 1

[Peter]

Five years ago the NCBI said their QBLAST web-API does not support PSI-BLAST, so you would be better off using the standalone BLAST+ binaries at the command line (optionally via the Biopython wrappers). See:

• http://lists.open-bio.org/pipermail/biopython/2008-May/004231.html
• ...
• http://lists.open-bio.org/pipermail/biopython/2008-May/004234.html
• http://bugzilla.open-bio.org/show_bug.cgi?id=2496
• https://redmine.open-bio.org/issues/2496

Someone asked something similar on Reddit recently http://www.reddit.com/r/bioinformatics/comments/1qmpbm/problems_with_psiblast_results_from_biopython/ which was flagged on the Biopython mailing list http://lists.open-bio.org/pipermail/biopython-dev/2013-November/010976.html

If you posted your code, it might help us modify Biopython to give a clearer error message?