PsiBlast->Align (was Re: [Bioperl-l] Newbie bioperl questions)

[Arne Elofsson]

Peter Schattner <schattner@alum.mit.edu> writes:

> Arne Elofsson wrote:
> 
> > 2) I will need to obtain multiple sequence alignments (and profiles)
> > from psiblast. I have perl code that does this. I would like to modify
> > BPlite (and related modules) to include this so that you can call
> > $sbjct->msa and obtain a \@seq_array
> >
> 
> Hi Arne
> 
> This sounds great.  I would suggest looking at modifying Bppsilite.pm rather
> than Bplite since it deals directly with parsing psiblast reports.
> Bppsilite doe not currently produce an alignment (object) from the psiblast
> report.  If bppsilite were modified so that it also could produce an
> alignment I think this would be an significant enhancement.  I would
> recommend that the alignment be created as a bioperl "SimpleAlign" object.
> If this is a task you might be interested in taking on, I (and I am pretty
> sure others in bioperl) would be glad to support you in doing it.
> 

Ok. I am done. $last_iteration->Align should return a simplealign
object.  (Well the code might need a little bit of
cleanup/testing). 

My main question is what do I do with the patches. Send them to the
maintainer, check them in by CVS or ???

The easy way to fix this was to use the flag "-m 6" when you run
psiblast, however this made it necessary to add some more fixes so
that ->nextSbjct works. Unfortunately using "-m 6" breaks ->nextHSP
(It could be fixed but it involves some work). 

My next small project will be to create a PDB object. I think it will
actually be based on the SimpleAlign object. Each PDB structure will
have two sequences, one from the header and one from the C-alpha
trace. In addition there might be extra information such as, C-alpha
coordinates, secondary structure etc. This will not be a full object
for handling protein structures but rather an object for obtaining
structural information from a sequence. If you have comments about
this please send them to me.



yours

arne