[adegenet-forum] different results for pegas and hierfstat Fst via weir and cockerham

[Thibaut Jombart]

Hi Gary,

it is not the first time we see discrepancies in F stats across different
packages and estimators. As these are related to pegas and hierfstat, maybe
you'll get better feedback posting it as an issue on the github project
pages? I think Jérome and Emmanuel are on the adegenet forum, but they may
miss this.

Best
Thibaut


--
Dr Thibaut Jombart
Lecturer, Department of Infectious Disease Epidemiology, Imperial College
London
Head of RECON: repidemicsconsortium.org
WHO Consultant - outbreak analysis
https://thibautjombart.netlify.com
Twitter: @TeebzR
+44(0)20 7594 3658

On 13 April 2018 at 00:09, Gary Longo <garycharleslongo at gmail.com> wrote:

> Hi Thibaut and company!
>
> # I've noticed differences with Fst and Fis calculations in Pegas vs
> Hierfstat while using Adegenet
>
> > data("nancycats")
>
> # using weir and cockerham in pegas
> > nancycats_Fstat_as_loci <- Fst(as.loci(nancycats))
> > nancycats_Fstat_as_loci
>
>             Fit        Fst         Fis
> fca8  0.2447420 0.10146648 0.159454807
> fca23 0.1646295 0.06746762 0.104191391
> fca43 0.1514487 0.06893755 0.088620458
> fca45 0.1010807 0.09792456 0.003498722
> fca77 0.2790495 0.10036588 0.198618075
> fca78 0.1842490 0.07025915 0.122603911
> fca90 0.2098744 0.09168833 0.130116240
> fca96 0.2034755 0.10744024 0.107595351
> fca37 0.2604033 0.06985321 0.204860244
>
> # also using weir & cockerham in hierfstat
> > nancycats_hfstat <- genind2hierfstat(nancycats)
> > nancycats_wc <- wc(nancycats_hfstat)
> > nancycats_wc_loci_stats <- as.data.frame(nancycats_wc[["per.loc"]])
> > nancycats_wc_loci_stats
>
>          FST          FIS
> 1 0.10150515  0.148673460
> 2 0.06746762  0.104191391
> 3 0.06893755  0.088620458
> 4 0.07652596 -0.001451681
> 5 0.10036588  0.198618075
> 6 0.07025915  0.122603911
> 7 0.09168833  0.130116240
> 8 0.10981110  0.094857474
> 9 0.06985321  0.204860244
>
>
> > colSums(is.na(nancycats_hfstat))
>
>   pop  fca8 fca23 fca43 fca45 fca77 fca78 fca90 fca96 fca37
>     0    20     0     0    21     0     0     0     9     0
>
>
> The results are identical in loci that are not missing data but are
> different in loci with missing data.
>
> Two questions:
> 1) How are they handling missing data differently since they are both
> using Weir and Cockerham 1984? Follow up: which is better suited for
> calculating Fst and Fis values when there are missing data? I'm analyzing a
> SNP dataset based on ~2100 RADseq loci in over 500 individuals, which of
> course has some missing data at most loci. My results from calculating
> these values in Pegas vs Hierfstat are very different. Specifically Fst and
> Fis values are generally much higher and I don't get any negative Fis
> values when calculated in pegas.
>
> 2) Why does the conversion to hierfstat result in the loss of loci name?
> This would be very useful to retain for downstream comparisons.
>
> Cheers,
> Gary
>
>
> --
> Gary Charles Longo
> NRC Research Associate
> NOAA, National Marine Fisheries Service
> 2725 Montlake Blvd E
> Seattle, WA  98112
> garycharleslongo.wordpress.com
> (831) 247-3056
>
> _______________________________________________
> adegenet-forum mailing list
> adegenet-forum at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/
> listinfo/adegenet-forum
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/adegenet-forum/attachments/20180419/fa2c8c72/attachment-0001.html>