I have a SNP dataset that contains several cryptic species. I can separate the individuals into groups using STRUCTURE and PCA, but I'd like to give them a nice number so we can say that yes, they are actually substantially differentiated.
For my last project we were just using single sequences, so I could calculate plain old sequence divergence. Is there anything like this for SNP data?
Alternatively, I do have the sequences the SNPs were retrieved from. Is there a way to calculate divergence using many sequences?
Emma,
I think you can use nucleotide-based Fst. I have been using it in my SNP datasets and I think the results are fine. I can suggest the R package PopGenome, as it is very flexible and allows you to estimate several other diversity statustics.
Hope this can help you
Hi Emma,
If you want to calculate the genetic diversity of a single population, there are several estimates that you can use such as the allelic diversity, proportion of polymorphic loci, observed and expected heterozygosity, and gene diversity of Nei (pretty much the same that you use for sequence data). A very friendly software that estimates all these parameters is GenAlEx (http://biology-assets.anu.edu.au/GenAlEx/Welcome.html).
Now, if you want to quantify the level of population subdivision, as Manolo suggested, F-statistics are more appropriate. PopGenome package is one of the best alternatives in this regard, although there are also other free available softwares such as Genepop (http://genepop.curtin.edu.au/) that are also pretty easy to use.
Cheers,
Hi Emma,
Yes, you can calculate genetic divergence using SNP data. Are your sequence data nuclear or mitochondrial?, Whatever it is, you can align your sequences and use MEGA6.06 to calculate Mean diversity of the entire population, within sub-populations, inter-population diversity and coefficient of differentiation. Good luck with your analyses.
Best,
Hi Emma,
I suggest you to improve your PCA analyses using DAPC (Lombart et al. 2010), which in my opinion will provide you a better graph and with optim.a.score function you can infer the admixture pattern across your populations and with compoplot function you can plot (this plot will provide to you a better knowlegment about structure of your data/populations) too.
Cheers,
- Badges
- Science topic