By using next generation sequencing data and pipelines, we can easily find SNPs for individuals.
Assuming SNP information has already been obtained for each individual, is there a published tool to calculate dN/dS?
You may find the commonly shared genes between all genomes (core genes) using orthologous prediction with orthomcl, blastclut or whatever tool you prefere. Than, you may align all of them with muscle, concatenate all alignments and them input it to DNAsp. DNAsp calculates Tajima's D and others.
Ps.: you need at least four organisms in order to calculate it.
Hi, you would have to first go from the SNP mapping to the alignments of the haplotypes and then you can use the standard tools (CodeML, etc)
If your individuals are from the same species perhaps rather than dN/dS you want to compute Pi-n/Pi-S which is a similar measure but with polymorphisms instead of fixed mutations:
Look at this:
http://www.plosgenetics.org/article/info%3Adoi%2F10.1371%2Fjournal.pgen.1000304
The dN/dS ratio is the rate of non-synonymous (dN) and synonymous (dS) substitution between species. You can't measure it with one species (but you don't need four, just two). Within species you can compute the Pn/Ps (relative abundance of non-synonymous and synonymous polymorphisms) that measure the direct effect of natural selection removing slighlty deleterious non-synonymous variants.
http://www.plospathogens.org/article/info%3Adoi%2F10.1371%2Fjournal.ppat.1002893
Rephrasing.
In order to use DNAsp, you need at least four datasets, as highlighted in its manual, here:
http://www.ub.edu/dnasp/DnaSPHelp.pdf
It is probably because DNAsp uses the "Confidence limit of D obtained by assuming the beta distribution" from the table 2 of the article:
"Statistical Method for Testing the Neutral Mutation Hypothesis by DNA Polymorphis"
And really, I would be highly skeptical of the statistical significance of a dataset composed of only 2 samples.
I agree with you if you use the Dn/Ds ratio to find gene/genomic under positive selection with two species that will be statistically very low.
But, in my opinion (of a small phD student), if you compute the McDonald-Kreitmann test for contrasting polymorphisms and divergence with dN/dS and Pn/Ps (e.g if Pn/Ps = dN/dS => neutral evolution and if Pn/Ps < dN/dS => positive selection) i think it works with both species.
But i thinkthe more powerful appoaches to explore and quantify genome evolution is to infer this with coalescence models:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2746150/
But that's another story
how about within one species? what is a good measure to look at genes under selection if Im looking at SNPs
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics