yes indeed, could be better idea in fact to also check LD between loci. I think a ref here tells where to look for: http://www.nature.com/nrg/journal/v4/n12/full/nrg1226.html#B36. ciao

Hi Anurag, well why not just getting the VCF format into Genepop (PGDSpider can do this I think, or at least for many standard SNP table formats) and use Bayescan/FDist2(Lositan) instead ? 

Thanks for the suggestion Greg, but due to high number of variants these are not feasible approaches. Alternatively, I would like to try sliding window analysis.

yes indeed, could be better idea in fact to also check LD between loci. I think a ref here tells where to look for: http://www.nature.com/nrg/journal/v4/n12/full/nrg1226.html#B36. ciao

Thanks for the pointer Greg, but the issue is how to do it programatically. 

Oh yes Clarissa, depending upon which type of sequencing protocol you have used. If you have whole genomes a good start will be pFST function of VCFlib (https://github.com/vcflib/vcflib)