Hi

This may be a silly question, but I’ve been using the GABranch method to look at variation in dN/dS among lineages for different protein domains of a gene. I’d now like to compare the trees to each other to see if there are significant differences among domains. Is there an easy way to do this through Datamonkey or do I need to switch to the HyPhy interface?

Thanks for the help!
Daryn

Hi Daryn, if by trees you mean topologies then the KHTest.bf in ./TemplateBatchFiles in the HyPhy distribution will do this. If by trees you mean the distribution of selection classes among lineages then the answer is not trivial. There is no statistically robust way to do this, but some hackish measures can be attempted. For instance you could do a branch by branch comparison of estimated dNdS values for a full protein alignment versus that of each domain. Clearly, the single result output from GABranch is not sufficient for this and we would need to process the credible set of GABranch models. If you're keen to pursue the hackish approach then please send your job upload id's so i can access the full output files. cheers ./w

Sorry, I meant the distribution of selected classes among domains. We’re basically trying to test a long standing hypothesis that there is variation in the strength of purifying selection among domains at this particular protein. I’m definitely up for a hackish approach – thanks! The upload ID for the entire gene is 769413173912666.1 and the upload IDs for the domains are 476578907412493.1; 930161873912609.1; and 99377543463329.1.

In the mean time, I started using HyPhy to compare likelihoods where each domain evolves independently to likelihoods where I constrain the nonsynonymous rate in one domain to that in another domain (doing all pairwise comparisons for both nonsynonymous and synonymous rates). Is that similar to the hackish approach you were thinking of? I know the dN and dS estimates aren’t exactly the same as those provided by GABranch, but this method was all I could think of :)

Thanks for the help – I really appreciate it!
Daryn