Here's a summary of replies I received from my query about how to calculate pairwise FSTs when you only have allelic frequency data not raw genotypes. For my purposes using the POPTREE2 program (#7) was the simplest solution.
1. from memory, Swofford and Selander's BIOSYS does this.
2. if you have sample size (the number of individuals from which your allele frequencies were obtained), I guess you can calculate pairwise FST by using directly the Weir&Cockerham(1984) method. In their equations, you need to know allele frequencies, sample size, number of populations plus a few other variables such the heterozygosities and mean allele frequencies that you should be able to compute from allele frequencies. I remember I had implemented a Matlab script to do it and it worked pretty well.
3. Attached please find a script that calculates pairwise Fst using simuPOP (simupop.sourceforge.net). My previous script used real genotype but I have modified it to assign random genotype using specified frequencies so you should be able to make it work for your data. You can check the simupop user's guide if you are interested in understanding how this script works. NOTE that the results will be slightly different each time you run this script because of the random initialization of genotype, and I do not know if this is the correct way to calculate Fst from frequency data.
(Script is from Bo Peng bpeng@mdanderson.org)
4. Yes, you can use just the allele frequencies. In fact, all you need are allele frequencies because genotype frequencies are not used for Fst (only for Fis). But you are going to have to write an analysis program yourself. I have some simple spreadsheet simulators associated with my textbook that do it for Rst and Fst (see chapter 4). You could write a program in Matlab really easily. Alternatively, you could code up a mock data set with the correct allele frequencies but composed entirely of homozygotes.
5. There is a bit of software written by Richard Nichols called Nichols Fst. Unfortunately, you have to run each population pair separately. You can download it from his website. It's in the FstLikelihoods directory:
http://webspace.qmul.ac.uk/ranichols/programs/
You might also consider using a measure of information theory to estimate pairwise difference, e.g. Shannon's index (see Sherwin et al
2006 Mol Ecol) and D (Jost 2008 Mol Ecol). Emerging studies seem to indicate that FST is not a reliable measure of differentiation.
6. I have used Arlequin for data from genotype frequencies and haplotype frequencies only but, it seems to me, that you could name the alleles consecutively from allele 1 of the first locus to the last allele (allele X) of the 6th locus and the pairwise comparison analyses between populations would work. I would be curious to know what works.
7. We have developed a computer program called POPTREE2 that can calculate the pairwise FST between populations from allele frequency data. It can be used through a Windows-interface easily. POPTREE2 is available at the following webpage.
http://www.med.kagawa-u.ac.jp/~genomelb/takezaki/poptree2/index.html
8. Why don't you calculate it by hand/using a spreadsheet or statistical package?
9. You've probably received an answer to your question already but my suggestion would be to fudge the appropriate input file using your allele frequency data. Fst's for msats are based on frequency and not genetic distance so you can assign an arbitrary fragment size to each allele. Obviously this would only work if you know which alleles are shared between the different populations. As an aside, another issue to consider is the impact of diversity on fixation indices. See Hedrick 2005, Meirmans 2006, and Jost 2009 for a description of the statistical issues and remedies associated with using fixation indices to estimate population differentiation.
Thanks for all the suggestions,
Hilary
Dr Hilary Miller
School of Biological Sciences
Victoria University of Wellington
PO Box 600, Wellington
New Zealand
Ph: +64 4 463 7432
Fax: +64 4 463 5331
email: hilary.miller@vuw.ac.nz
Hilary.Miller@vuw.ac.nz