[Bioperl-l] Re: Determining strand orientation from bl2seq report

[Boris Lenhard]

> Dave - a little bit of detective work.  This stems from the fact that
> bl2seq reports don't include information about which blast program was
> used (blastn, blastp, etc) and that determines whether or not we set
> stranded information or set it to 0 (for proteins).  Surprised this was
> not caught before, I've added tests and testfiles to make sure this
> doesn't happen again.
> 
> I've committed changes to CVS allow you to specify the program used and
> added tests for blastn bl2seq reports.  You can see how it works by
> looking at t/BPbl2seq.t .
> 
> Basically when you initialize a BPbl2seq object you add in the flag
> 
> -report_type => 'blastn'
> 

After this, one should perhaps also change the in
Bio::Tools::Run::StandAloneBlast _runblast (line 650) to pass the
object's p (program) parameter as -report_type to the
Bio::Tools::BPbl2seq it calls. It is a god occasion for me to ask for a
full cvs account, I guess :)

In principle, a bl2seq report contains enough info to deduce indirectly
which BLAST program produced it - from Strand or Frame lines (currently
not used by the parser),  and  the end-start/hsp_length  ratio (3 or 1),
so using that info instead of explicitly passing around the report type
might be cleaner as a long-term solution. 

Boris



-- 
==========================================
 Boris Lenhard, Ph.D. 
 Group Leader, Applied Genome Informatics 
 Center for Genomics and Bioinformatics 
 Karolinska Institutet 
 Berzelius väg 35, B326b 
 171 77 Stockholm, SWEDEN 
 Phone: +46 (0)8 728 6391 
 FAX: +46 (0)8 32 48 26 
 E-mail: Boris.Lenhard at cgb.ki.se
==========================================