Morning (here) Sergei,
I am familiar with the SelectionLRT analysis and found it very useful in another study, which I had published in MBE last year. I used HyPhy alot in the paper (thanks!), if you are interested: Mol Biol Evol. 2010 Aug;27(8):1886-1902.
However, for my current work, this analysis is not as helpful, since I am looking at duplicated genes with the topology: ((Paralog1_Species1,Paralog1_Species4),(Paralog1_Species2,Paralog1_Species3)),(P
aralog2_Species1,(Paralog2_Species4,(Paralog2_Species2,Paralog2_Species3))),Undu
plicated_Outgroups)
I am interested in comparing (in a single analysis preferably since I am considering setting up a high-throughput analysis with 100s of genes) gene wide constraints for 5 compartments i.e. Paralog 1 orthologs, the branch leading to Paralog 1, Paralog 2 orthologs, the branch leading to Paralog 2 and Outgroup orthologs. The same LRT approach as in SelectionLRT but with these 5 comparments (or a subset based on specific hypotheses) would be perfect.
With Selection LRT I could only infer selection for one paralog clade VS its branch VS all other seperating branches (including the other paralog clade plus more basal unduplicated outgroups).
Thoughts?
All the best
Dan
ps, the HyPhy wikis are an excellent resource- you should have many PAML converts soon!