Conservation At Synonymous Sites

Question

[Sankalp Sharma]

Hi all, this is the first time that I am posting a query here. So please do not be offended if I have written something which is not the norm here.

Essentially, I have an alignment of protein coding genes and I am trying to find if there are certain regions in the alignment which evolve under dual constraints i.e. in addition to the constraint for protein coding, there is also some regulatory role for certain DNA elements. Such constraints are usually observed for elements like Exonic splicing enhancers.

One way to do that is to look at the conservation at synonymous sites. Since synonymous sites are free from constraints of protein coding, any conservation observed at such sites would imply an additional role for such subsequences, possibly regulatory. However, I do not know how do some math/stats to prove that the conservation observed at synonymous sites is statistically significant and more than what would be observed under a null model (where the null model states that the only constraint/conservation observed in a multiple sequence alignment is because of protein coding). Any help in this regard would be very very welcome.

Regards

Sankalp

Answer 1

[Casey Bergman]

There is some recent work by the Kellis lab on this issue that may point you in the right direction:

Locating protein-coding sequences under selection for additional, overlapping functions in 29 mammalian genomes. http://www.ncbi.nlm.nih.gov/pubmed/21994248

[Casey Bergman]

Why not email the authors to ask for their code?

[Casey Bergman]

They used MULTIZ as input which you can generate for your data. They are unspecific about how they estimate parameters, but in general use a dataset of coding regions to estimate a codon-based rate matrix, which you could do in PAML. I would not talk yourself out of contacting them for code and more information about their approach, especially the parameter estimation. You could save yourself a lot of time by doing so.

[Sankalp Sharma]

Thanks Casey, I have seen that paper. But I am not really a molecular evolution person. And I was looking for something that is already available rather than trying to figure out all the parameters for modelling sequence/codon evolution and then write code for the same. But thanks anyway for the prompt reply.

[Sankalp Sharma]

Things are not that simple. What these people did was specific for the 29 way vertebrate alignments that they were interested in. They were estimating the parameters which again were specific for those organisms. I have an altogether different dataset and hence cannot use the same.

[Sankalp Sharma]

Thanks Casey, I will get in touch with these guys.

Answer 2

[Manu Prestat]

Aren't you talking about the dN/dS ratio, also known as the Ka/Ks ratio ? A brief summary can be found in this review. The seqinR R library is an example of tool that can compute this.

[Philippe]

dN/dS assesses the extent of conservation of protein-coding genes (even if it can be used for other sequences). Here he wants something different since he wants to test if some synonymous - supposedly neutral - sites undergo more selective pressure than others.

[Steve Moss]

Can't you calculat dN/dS per codon? May be quite computationally expensive though?

[Steve Moss]

Can't you calculate dN/dS per codon? May be quite computationally expensive though?

Answer 3

[Philippe]

Hi,

I never really played with this kind of data but aren't PhyloP or PhastCons a solution to your problem? They provide a score of conservation for every base in the genome.

Their differences have been debated here in BioStar. Nonetheless, they might not be available for all species (but can be generated I guess).

[Sankalp Sharma]

Thanks Philippe, I need to check if I can use Phastcons for the alignments that I am interested in.