BSM analysis in parallel evolution

[张东]

Dear all, 

I want to study parallel evolution, but foreground species are interlaced with background species in the tree,  I have run codeml using two approaches, 

1. label all foreground branches in the tree, and then run BSM analysis at once, like this: [https://groups.google.com/g/pamlsoftware/c/aVj2opOg7PA]; 

2. extract all parallel clades, and then run BSM for each clade seperately, finally find intersection in the resulting positively selected sites of each clade, which is a method used in a PNAS paper: "Comparative genomics reveals convergent evolution between the bamboo-eating giant and red pandas". 

I tried both methods, but the results of two approaches were different, I am not sure what cause the discrepancy and which approach is the best strategy, can you help me with this, 

Thank you

Dong

[Yangkang Chen]

I'm also facing the same problem. Did you managed to solve that?