Codeml Branch-Site Model
Hi
I want to predict the positively selected genes in four closely related genomes. I was using Branch-site model 'Test-2' for this purpose. Can I conclude that the sites with labeled branches are under positive selection, if my LRT [H0(wfix=1) and H1(wfix=0)] test comes significant i.e p-value < .001. Or should I also try free-ratio and one-ratio tests ???
Thanks in advance,
Rahul
• 585 views
•
link
written
by
Rahul Sharma
0 answers
No answers yet.
Log in to answer this question.
More posts like this
-
What Does Elevation Of Dn/Ds Ratio Indicate?
written by qiyunzhu
-
Interpreting Hyphy Testbranchdnds.Bf
written by Rvosa
It always depends on your alignment and the data (i.e., transcriptome, whole gene, etc.), which we know nothing about. What are you testing selection on and did you run a codon model also?
Thanks for your message. Actually I assembled 1 genome and did gene predictions without transcriptome data. Then OrthoMCL for ortho predictions and mafft generated phylip alignments. Now I am trying to figure out positively selected genes among four closely related genomes(>5000 ortho), and the sites in these genes. I have taken only those othrologs sets for positive selection analysis which has <10%(gene length) gaps in the alignments. I did not run codon models yet. What are your suggestions regarding this? Thanks in advance.
I'm sorry but I'm still unclear how you are doing the comparisons. Anyway, you could try translating your sequences and aligning them to the protein ortholog of some outgroup, then use your nucleotide sequences to generate a codon alignment. Programs like Pal2Nal will do this for you and remove those gaps, after this you could run codeml selecting a codon model to identify sites under selection. Unless, you are only interested in branches showing selection. Regardless, be very skeptical of any of these tests because they are influenced by your data and how you are making comparisons. Finding the appropriate method is not easy, sorry I can't offer anything more concrete without knowing the system/data.
Thank you so much for your suggestions. I did the protein alignments using mafft (ginsi algorithm) and later converted these alignments to codon alignments using pal2nal package. Now I am running codeml branch-site model test2. I have following questions regarding this:
1). I did LRT test (H0 vs H1), and it favored H1(p<<0.001) but some Omega values of foreground branch are very large >100. Is this usual or I am getting some false+ results.
2). I have four species, Lets say A, B, C, and D. I want to check the sites under selection in a particular species. I tried many different combinations (A#1,(B,C)D), (A,(B#1,C)D), (A#1,(B,C)D#1), (A#1,(B#1,C)D), (A,(B,C#1)D), (A,(B,C#1)D). How can we check (A#1,(B,C)D) and (A,(B#1,C)D), as np is 10 for both of the models (For H1). I would really appreciate your suggestions regarding this.