About Fst calculation in Plink.

[Bernabé Ignacio Bustos Becerra]

Dear Dr. Chang,

I'm working in a population genetics project using SNPs called from whole-genome sequences. One of my task is to calculate genetic differentiation with other populations (i.e 1000 genomes project) so I'm using Plink's Fst to do this. After doing pairwise calculations and getting Fst values for each SNP, I understand that in order to select variants with statistically significant Fst values I need to do a permutation process to create an empirical distribution and then use methods like Bonferroni-corrected permutation test P-value [1]. I found a helpful topic on Biostars (https://www.biostars.org/p/115453/)  where the user Zev.Kronenberg pointed out the necessary steps to follow. That's why I would like to know if this analysis or an alternative could be implemented in PLINK's Fst calculations?.

I hope you can hep me, I'll look forward to your comments.

Thanks in advance,

Best regards, 

Bernabé Bustos

1.  http://www.nature.com/ejhg/journal/vaop/ncurrent/full/ejhg2014233a.html

[Christopher Chang]

I will look into adding pFst to PLINK in the future.  For now, though, you are probably best off using the GPAT++ tool.

[Bernabé Ignacio Bustos Becerra]

Thakns for your answer.
I have tried GPAT++ but pFST actually computes a chi-square statistic comparing allele frequency between 2 populations , it doesn't use heterozygosity so is not a real WC Fst estimator. I'll keep looking forward to se a way to boostrap plink Fst output to get a null distribution and then calculate Pvalues in order to detect significantly divergent SNPs. It would be nice to see this implemented in PLINK as standar procedure in evolutionary genetics.

Thanks again, have a great weekend.

Cheers,

Bernabé