I understand that my end result will be a LRT statistic for me to ponder, and I understand that I need to use a branch model. However, I currently don't know how to partition my sequence groups in my codeML control file. I want to see if a certain group (four sequences) is under significantly greater/lesser selective pressure than the remaining sequences in my multiple alignment (5 other sequences). I am a noob when it comes to PAML and so any help would be greatly appreciated.