Hello,

I am using the weir-fst in vcftools to calculate the pairwise fst. I have also written my own script using the Weir and Cockerham 1984.

The puzzling thing is that I get different estimates, the difference is between 0.004 to 0.01 but this is still puzzling to me, if both calculations are based on the same formula and considering the deterministic nature of weir-fst, I should get exactly the same estimates using exactly the same data.

I was wondering whether it is possible to get access to vcftools pairwise fst source code. I would also appreciate if you could share your ideas about this in case you have similar experience.

Cheers, Lucia

good comment, I check it now

Yes, I am using the version VCFtools-v0.1.10.2, so, I think this should be fine.

Actually, there were a couple of bugs fixed in v0.1.11, so please upgrade. Thank you.

Just installed v0.1.11 and did the analysis again, the results are exactly the same as the previous analysis. To give you an idea about the difference, the number I get for my analysis for example for one pair of populations is 0.3692089 using my own script and 0.36488 using vcf tools weir fst.

The vcftools calculation has been checked against the geneclust R package (at least the version in the SVN), and gets the same results. If you compare to geneclust, do you get the same issue?