[Bioperl-l] ortholog identification for many species

Elia Stupka elia@fugu-sg.org
Mon, 8 Jul 2002 13:41:15 +0800 (SGT)


> Some would argue, and I've heard it argued, that orthologs are those pairs
> of genes that have arisen from a single gene in the ancestral genome. What

We've had major debates over this definition already, so I'd rather focus
on practical aspects of it. I think right now some of the most promising
methods that would span multiple organisms would be related to
phylogenetic trees. What I think many of us including Jason and me are
hoping to achieve at some point is a complex and integrated enough
represenation of proteins, their phylogenetic trees and other properties
to try to build automated methods for orthologue finding based on the tree
topologies.

The other approach is somewhat more simplistic, but in no way should be
discarded, and that is to decide what you consider a good orthologue
finding method for a pair of organisms depending on your strictness (from
reciprocal best blast, to ensembl's seeds and pips, to Wolfe's
methodology) and apply to many organisms pairwise. This will definitely
prove to be more accurate as once you put in multiple genomes you are
bound to make arbitrary decisions about what is in what is out, what is
a paralogue,etc. As more genomes are completed and thus the average
evolutionary distance between sequenced genomes gets smaller this should
also be a feasible and clean approach.

Elia

********************************
* http://www.fugu-sg.org/~elia *
* tel:    +65 6874 1467        *
* mobile: +65 9030 7613        *
* fax:    +65 6777 0402        *
********************************