[Biopython] Problems parsing with PSIBlastParser

[Peter]

CC'd back to mailing list

On Thu, Oct 15, 2009 at 4:51 PM, Miguel Ortiz Lombardia
<ibdeno at gmail.com> wrote:
> Le 15 oct. 09 à 17:39, Peter a écrit :
>
>>> For me the XML is a no issue, since the NCBIXML parser does not really
>>> support PSI-BLAST searches:
>>> it can't get information on the rounds, convergence... If you have a look
>>> to NCBIXML.py you see a lot of XXX TODO PSI...
>>
>> There may well be some things missing in our parser, but last time I
>> checked, the XML file itself was missing lots of information found in
>> the plain text output.
>>
>> Peter
>
> I am sending to you an xml file from a PSI-Blast run that converged. You see
> there for example info about iteration number and convergence, for example.
> It's just 59 Kb, I can upload it to the bug 2927, but I suspect you prefer
> not, since this is a new issue (XML parser).
>
> IMHO, the NCBIXML parser should behave (concerning PSI-BLAST) just like
> NCBIStandalone.PSIBlastparser. Perhaps a NCBIXML.PSIBlasparser should be
> created ? Just an idea...

Michiel also thinks the PSI BLAST XML parser could be better, see:

http://lists.open-bio.org/pipermail/biopython-dev/2009-September/006780.html
...
http://lists.open-bio.org/pipermail/biopython-dev/2009-September/006791.html
http://lists.open-bio.org/pipermail/biopython-dev/2009-September/006802.html

Can you file a new bug about PSI-BLAST XML parsing (and attach that example)
please? I'd have to look over the new PSI-BLAST XML files before having an
informed opinion.

Peter