Identifying sites under positive selection using the BEB method

Question

[ta2007]

Hello! I ran codeml branch-site test (null and alternate) on 2118 genes and found that 42 of these are significant and thus, contain sites under positive selection. Now I need to find these sites using the Bayes Empirical Bayes (BEB) method. I read the paper already but I don't understand which tool to use to find these sites under selection. Does anyone know how to identify specific sites under positive selection using the PAML branch-site model output? Thanks!

Answer 1

[Brice Sarver]

Sites under selection are listed after the analysis for that given model. As an example, this is in the codeml output for one of my analyses after testing model M8:

Bayes Empirical Bayes (BEB) analysis (Yang, Wong & Nielsen 2005. Mol. Biol. Evol. 22:1107-1118)
Positively selected sites (*: P>95%; **: P>99%)
(amino acids refer to 1st sequence: S03amoenu)

            Pr(w>1)     post mean +- SE for w

  2043 N      0.759         1.245 +- 0.469
  2119 L      0.559         1.007 +- 0.579
  2221 T      0.588         1.043 +- 0.569
  2310 N      0.744         1.229 +- 0.480
  2321 I      0.633         1.099 +- 0.549
  2606 P      0.655         1.125 +- 0.538
  3434 V      0.865         1.373 +- 0.330
  3451 N      0.756         1.242 +- 0.472
  3454 I      0.717         1.198 +- 0.498
  3518 I      0.631         1.096 +- 0.550
  3575 T      0.607         1.067 +- 0.561

Given that model M8 is selected as the best-fit model, the BEB analysis suggests that those sites (codons) are under positive selection.

Answer 2

[ta2007]

Hi Brice! Thanks so much! I was just about to reply to my own question because I learned this info right after putting up the post this morning. Thanks for your reply!

Taruna