[Bioperl-l] BPlite and frames

[Hilmar Lapp]

Hilmar Lapp wrote:
> 
> Thanks to Roger's recent fixes BPlite can now handle/decode the
> frame information properly. However, there's now a problem with
> the other BLAST flavours, because BPlite::Sbjct and BPlite::HSP
> are now expecting the parent object to know about a BLAST_TYPE
> member variable, which it doesn't for bl2seq reports and
> apparently neither for PSIBlast/PHIBlast.
> 
> It appears that at least bl2seq reports do not contain any
> information about the "program" that was used (you have the same
> palette to choose from: blastn, blastx, etc). Does anyone know
> whether this was changed in the latest versions (I have 2.1.3
> installed)?
> 
> Also, does anyone already know whether it is safe to assume that
> for PSIBlast you always have a protein query (and DB)? I.e. frame
> does not apply?
> 
> The AlignIO test spits out warnings related to this same problem,
> too. If I have time I'll investigate.
> 

I changed and rearranged the report type querying and
interpretation such that it will be UNKNOWN if not set or not
available, and the default for frame is to pertain to query. The
Blast types are all recognized and dealt with accordingly, if it
is set. Note that this means that a bl2seq run with e.g. tblastn
as the program (i.e., you aligned a protein against a dna seq)
will not be handled correctly as far as frame assignment is
concerned. Same holds true for tblastx.

Any suggestions for alleviating this problem? The bl2seq/tblastx
case could be solved by looking for the double frame indication,
without caring for the report type set. But how about
bl2seq/tblastn?

Note also that bl2seq runs with blastn, blastx, and blastp
(dna-dna, dna-prt, prt-prt) should be fine, as these fit the
default (either no frame or frame pertains to query).

BTW the AlignIO test shuts up again. Not sure whether this means
everything's okay on that front.

	-hilmar
-- 
-----------------------------------------------------------------
Hilmar Lapp                              email: hilmarl@yahoo.com
A-1120 Vienna
-----------------------------------------------------------------