plink 1.9 FST across multiple populations

[Lynne E]

Hello,

When using plink 1.9's --fst across multiple populations (e.g. the 1000 Genomes superpopulations), how are the raw and weighted global means estimated using the WC method? Does plink calculate pairwise FST across all populations first, then give the average for a given locus? And what is the weight added?

Thank you in advance,

Lynne

[Christopher Chang]

0. plink 2.0 --fst is probably more useful than the plink 1.9 command; in particular, it reports pairwise FST estimates, which are actually absent from the plink 1.9 report.
1a. plink 1.9 --fst's global means are of the "average of ratios" form.  Per-locus results are computed and reported, then the two averages over all the loci are reported.

1b. See the "weighting we advocated previously for distances" on the 6th page of the Weir-Cockerham paper (https://www.researchgate.net/publication/200102299_Weir_BS_Cockerham_CC_Estimating_F-Statistics_for_the_Analysis_of_Population-Structure_Evolution_38_1358-1370 ).