[Vcftools-help] Fst and non-called samples

[Travis C. <tra...@gm...>]

Ugh...
weir-fst is pretty broken when working from a vcf with non-called (./.)
genotypes.
If any sample (not just the ones being included in the weir-fst-pops) is
uncalled, then it skips position.  That is very broken IMO.

One can somewhat work around it by using --keep to filter for just the
samples included in the weir-fst-pops, but that leads to the still
'not-very-nice' behaviour of skipping positions where any sample in those
pops is uncalled.

It really should be treating uncalled genotypes / missing data properly.  I
thought one of the big points of Weir and Cockerham's Fst estimator was
handling situations where the sample sizes are unequal and/or variable.

I don't have the time to work out a proper fix at the moment, but really
wanted folks to be aware of the problem.