[Bioperl-l] Reciprocal best blast hits / Orthology

Christopher Larsen clarsen at vecna.com
Mon Mar 10 15:56:55 UTC 2008


Matt, Dave,

Regarding reciprocal best blast hit, yes -- its beyond the list and heres
how/ where to go. It seems what you are looking for is actually an
Ortholog search. If so there is more to it than reciprocity and
ranking--other groups are using phylo trees and bootstrap values etc.

Perhaps check out the perl written up by David Roos and Chris Stockerts
work: OrthoMCL. Their group is quite helpful as well.

http://orthomcl.cbil.upenn.edu/cgi-bin/OrthoMclWeb.cgi

The perl install goes local and will help you to identify a homolog that
should have the same enzyme function or cellular role. Importantly it
tends to the idea of paralogy and pseudogenes as well so you dont step
into a pit. The site explains more.

The point is, you're on the right track, but theres a group that's been
through what you are doing and can supply you with a working
implementation thats very robust and uses BioPerl modules already, so you
don't have to scratch up some code. Also you can check out INPARANOID for
the same reasons. Having just been through this, I'm just trying to lead
to you where we went. Right now we point OrthoMCL at a whole folder of
proteomes (*.faa) and it groups them accordingly.

Brian O: Don't know if this folds well into your MCL wiki page or not,
apologies.

If the group wants some post-processing code that shows the
presence/absence of proteins in any one group perhaps we can help too as
there are a few things written that take the raw output directly.

Cheers,

Chris L

=========================
Message: 5

Dear experts,

I want to do a best reciprocal blastp of a fasta protein dataset against the
protein models of various species also in fasta format. The aim is o have an
output showing presence/not presence. I think this is possible to do using
perl, but i'm very much a beginner so any help in this would be greatly
appreciated.

Thanks

Matt

-- 
Christopher Larsen, Ph.D.
Senior Scientist
Research Grants Manager
Vecna Technologies
5004 Lehigh Ave
College Park, MD 20740
240-737-1625



More information about the Bioperl-l mailing list