I have performed a codon based-Z test (MEGA) on a protein coding gene alignment from 30 species to compare the dN-dS values used as a measure of selective pressure indicating neutral mutation rates (dN/dS = 1), positive (dN/dS > 1) and purifying selection (dN/dS < 1). My Z-test rejected neutral and purifying selection and indicated a positive selection with significant p-value (p less than 0.05). I was further interested to find, which codons are under positive selection, so I used HYPHY and various tools within it to find the sites. But HYPHY found no evidence of positive selection at the codon positions. I have also used BUSTED for gene-wide episodic diversifying selection finding no signs, and FUBAR on the alignment to identify sites which have experienced pervasive diversification over the entire phylogenetic tree. This also gave negative results. Then I run aBSREL to test for episodic diversification, this gave strong evidence that episodic diversifying selection was found on one characteristic branch of the tree. Could you give an a reason why the codon based Z tests are indicating positive selection, while other tests don't and instead pointing to episodic diversifying selection on one branch of the tree and possibly what this could mean?
I would not trust MEGA with any of this and only use the tools in HyPhy. Episodic diversification is positiv selection and if it really is only episodic, than this could explain why it is difficult to detect. But positiv selection is always more difficult than negative. I would run RELAX to test for branches which have relaxed selection (this will be mainly the negative) to back up your suspected diversification. But I am sceptical like Artur about the significance of any positive selection in this case. Just would like to add, that this 5% really depends. Some of this tests are more conservative, if you look in the original paper were they have done it with simulated data. May this would be an idea to check for the reliability of aBSREL
I've just seen the question, but did you use MEME? It is more sensitive to find episodic diversifying selection and it's much better than FUBAR.
- Badges
- Science topic
- Similar topics
- Biological Science
- Genetics