Hey Sergei,

First off we want to thank you for all your help in the BEB thread.  It was very useful and we managed to accomplish all that you set out for us.  It really helped us understand our results better.

Now we have one simple question:  When doing all the analyses that we have done using HYPHY, should we use the tree inferred using HYPHY and PHYLIP via ML?  I am asking because the tree returned from ML estimate does not have the correct speciation order, i.e. the  one that has been shown in literature.  When we try to get a ML estimate using the tree with the correct speciation order, we get a lower ML score than the inferred tree.
Thanks,
Matt

Dear Matt,

For selection the tree has a fairly small effect on rate estimates, unless it is really wrong; you can run both to see if it makes a difference.

I would use the 'correct' tree, unless there is reason to believe that your gene tree is different.

Cheers,
Sergei