Ks Saturation After A Pairwise Comparison Analysis Using Codeml

Question

[Juliana Soto]

Dear all,

I´m doing a pairwise comparison analysis using runmode -2 in the codeml option from the PAML package, and I would like to filter the gene pairs that have a ks (synonymous substitutions) with saturation.

I´m looking in the literature for an upper threshold on values of Ks (to avoid saturation), but I have not been found any value or parameter.

I would like to know your opinion about it.

thanks

best regards

Answer 1

[Casey Bergman]

Ks > 1 implies that all silent sites have been substituted one average more than once. Past this point, actual sequence divergence is a highly dependent on the assumptions of your substitution model. I wouldn't trust anything with Ks>1, but real experts in this area do trust ML estimate of Ks up to 2. Also pay attention to the standard error on your estimate of Ks, if it is as high or higher than your point estimate you are pretty much guessing.

[Juliana Soto]

thanks a lot! I looked the SE for each dS and took this as the threshold for saturation as well we decided to take only ks values < 1. I think that this parameter is better than just to accept Ks up to 2.

best regards!