Dear authors of HYPHY,

The difference between two relatively new options in the positive selection analyses are not clear to me.

The two options are "compare selection pressure.bf" and "compare selection pressureVL.bf" (across a branch). Are these two analyses the ones that were used in your recent paper in PLOS Computational Biology using HIV-1 data sets, with the former being the fixed-effects site-analyses within populations and the latter the fixed-effects branch-site-like analyses?

Perhaps my confusion simply arises from the fact that I  could not find  the documentatiion of these two analyses in the HYPHY documentation or  that there is no need to specify a branch of interest in the second (branch-site-like) type of analysis as in PAML.

Thanks, Yer

Dear Yer,


wrote on Jun 11^th, 2006 at 2:32am:

Dear authors of HYPHY, The two options are "compare selection pressure.bf" and "compare selection pressureVL.bf" (across a branch). Are these two analyses the ones that were used in your recent paper in PLOS Computational Biology using HIV-1 data sets, with the former being the fixed-effects site-analyses within populations and the latter the fixed-effects branch-site-like analyses? Perhaps my confusion simply arises from the fact that I  could not find  the documentatiion of these two analyses in the HYPHY documentation or  that there is no need to specify a branch of interest in the second (branch-site-like) type of analysis as in PAML.

CompareSelectivePressureIVL.bf implements the Differential Selection test from the PLoS paper (selective differences along internal branches of the tree), whereas CompareSelectivePressure.bf includes all branches when comparing selection among sites in two different populations. One doesn't need to specify 'selected' branches, because the test always tests for differential selection along all internal branches in the tree.

There is no documentation for these analyses in the distribution (sadly). We have been invited to contribute a book chapter on methods for detecting positive selection for a book (coming out sometime next year). We'll make the chapter available online probably by October, and this should address most of the 'missing documentation' issues, at least for the analyses of selection. In the meantime, thanks for your patience! I'll try to promptly answer any additional questions you may have on these boards.

Cheers,
Sergei