dN/dS ratio (omega)

Question

[dav]

So, I have a group of genes (proteins) - that arent necesarely related and I am interested in dn/ds ratio, or rather the conservation. They are all from the same species (A.thaliana) and I have data on around 600 individuals (mostly from 1001genomes).

What software should I stick to?

I was thinking of PAML (Phylogenetic Analysis by Maximum Likelihood), more specifically codeml. I would give it aligned sequences, so 600 sequences of the same protein and a tree that goes with it, but I saw people say it is best suited for more distant species.

Is my approach wrong? What should I stick to for this?

[Whetting]

if they are not related (i.e. not homologous) you CANNOT use phylogenetic based methods

Answer 1

[natasha.sernova]

You are on the correct track, but PAML description is very hard.

This is the manual.

I would recommend to read all about codeml program (pp. from 28th to 38th.). You will understand what these parameters represent and all that is necessary to specify to run PAML.

I suggest using runmode=-2. What this mode means is described on p. 34.

Good luck!

Natasha

[dav]

Ah, so runmode=-2 gives me the exact lineage ratio I need!

Thank you a lot!

[dav]

But wait , if I run it pairwise , (runmode -2) , then it compares all sequences to the first one?

Or will it compare 3rd to 2nd , 4th to 3rd etc?

Should I leave runmode = 0 to get it globaly?

Answer 2

[natasha.sernova]

Dear dav,

I think the first url contains the answer to your question about Dn/Ds for a global alignment.

This file is about runmode-option.

In any case, PAML is a very complicated program as well as its manual.

Below there are a manual how to read PAML-description and other useful urls

• PAML is not an obvious program at all.
• PAML FAQs
• PAML How-to (BioPerl)
• A: How Can I Compare Rates Of Evolution For Two Sets Of Genes?
• How To Compute Dn/Ds Ratio And Get Descriptive Statistics On Thousands Of Sequences ?

Good luck!

Natasha