How to know if the third codon position is evolving too fast

Question

[qiyunzhu]

Dear all, I would like to know if there is any statistical test / program that allows me check whether the third codon position in my alignment is evolving too fast (too divergent) that it should not be included in a phylogenetic analysis. I googled around and didn't get any hits. I appreciate if you can share what you know on this topic.

[Gabriel R.]

I don't get it, if the third codon is highly divergent that is information per se, why would you throw it out ?

[qiyunzhu]

Because I fear (well, someone fears) that this could lead to bias in phylogenetic reconstruction.

[5heikki]

Why are you studying nt seqs instead of aa seqs?

[qiyunzhu]

DNA provides more information, you know.

[5heikki]

If you're studying really close lineages, like say strains of E. coli, then you're right. Otherwise, protein alignments almost certainly provide more information since things like saturation of the 3rd codon position can't happen there. For reference, see e.g. How To Tell If The Third Codon Position Is Saturated?.

[Gabriel R.]

You cannot just have "bias". Bias needs to be accompanied with a word like "it will bias towards [insert slant here]"

[Biomonika (Noolean)]

You may want to take a look at this post: How To Tell If The Third Codon Position Is Saturated?

Answer 1

[David W]

There are a couple of tests for this, including Xia (2003, http://dx.doi.org/10.1016/S1055-7903(02)00326-3), which I gather is implemented in DAMBE.

Probably the most common way to asses this though, is to plot the number of 3rd codon substitutions in a sequence against some corrected distance . If the curve "plateaus" you know you are hitting saturation.

[qiyunzhu]

Thanks! I tried DAMBE. It works great!