M7vsM8; LRT not significant; M8 found 9 sites

86 views
Skip to first unread message

LysM LRR

unread,
Apr 3, 2023, 11:51:37 PM4/3/23
to PAML discussion group
Dear PAML group,
Dear Prof. Ziheng Yang,

I did M7 vs M8 comparison in CodeML. The likelihood ratio test was not significant. However, M8 found nine sites with omega significantly above one. Please help me to understand the conclusions. The gene is under positive selection or not?

I should also mention that many of the sites are laking synonymous mutations. Can CodeML consider the lack of synonymous mutations?    

Thanking you,
Mohammad Tanbir Habib

Janet Young

unread,
Apr 5, 2023, 3:00:12 PM4/5/23
to PAML discussion group
hi Mohammad,

If the LRT test is not significant, we ignore M8's sitewise BEB analysis (the sites with omega>1).  Saying that another way, we need to see statistical support for the hypothesis that there IS a subset of sites under positive selection (the LRT) before we look at the analysis of which sites those might be (the BEB).

Codeml is indeed comparing the synonymous rate to the non-synonymous rate.  Not sure what you meant by that question.  You definitely want enough overall divergence (synonymous and non-synonymous) in the alignment to have the power to detect positive selection.  Ziheng or someone else may be able to point you towards discussion and publication on how much divergence is useful (too much is not good, too little is not good). We have had good luck with alignments across primates: others use other species sets.

good luck with your analysis,

Janet

LysM LRR

unread,
Apr 19, 2023, 2:59:38 AM4/19/23
to PAML discussion group
Dear Janet,

Thank you for your kind reply. The information was really helpful.

With greetings,
Tanbir

Ziheng

unread,
Jul 20, 2023, 6:01:48 AM7/20/23
to PAML discussion group
"Can CodeML consider the lack of synonymous mutations? "

i think the answer to this question should be yes.  more precisely if there are no synonymous sustitutions, the estimate of dN/dS is not very reliable (you will get infinity), but the LRT should be fine.
as an analogy, if you observe 2 heads out of 2 coin tosses, you may not reject the null of a fair coin by using LRT, but if you observe 20 heads out of 20 coin tosses, the LRT will reject the null.  the LRT considers the uncertainties related to the sample size.  in both cases, the estimate of the probabilty of heads is very extreme (100%).
best wishes, 
ziheng
Reply all
Reply to author
Forward
0 new messages