Weir And Cockerham Fst from PLINK(bed/bim/fam) input files

Question

[kautilya]

I want to calculate markerwise Weir And Cockerham Fst for certain differentiating markers. I have my data in PLINK(bed/bim/fam) format. I have scourged the internet for packages to calculate the fst but very few seem to support native plink format directly as an input.

There exists snpstats (bioconductor R package) but the method of calculation seems to be a non standard one.

Can anyone suggest an R package/software that can calculate Fst(preferebly Weir And Cockerham) from PLINK format data as input?

[zx8754]

Geneland seems to calculates Weir And Cockerham Fstat. Though you will have to convert plink files into matrix: "Diploid codominant genotype data. A matrix with one line per individual and 2 columns per locus"

[kautilya]

Thanks for the reply. Geneland seems to be useful and i will try to give a shot to coverting the data into required format. Though it will be preferrable to have a tool which directly uses PLINK files since it seems to be pretty prevalent standard. I am pretty surprised by the scarcity of such tools.

Answer 1

[chrchang523]

How are your subpopulations defined? (I may just go ahead and add this to PLINK, but I first want to make sure that your subpopulations are defined in a PLINK-friendly manner.)

[zx8754]

This is a comment, not an answer.

[kautilya]

The subpopulations are defined by the values 0(unaffected) and 1(affected) in the phenotype column of the bim/ped file. No sure if this is what you had asked for. But it would be great if PLINK itself natively supports this calculation as this is a pretty frequently used statistic and will be a significant addition to the already great PLNIK 1.9

[chrchang523]

Okay, I'll try to add a --weir-fst flag next week (analogous to VCFtools --weir-fst-pop) which uses the set of cases as the subpopulation, and can be combined with --loop-assoc to operate on many subpopulations at once.

[kautilya]

Great. Thanks. Eagerly Looking forward to this release

[chrchang523]

Now implemented as --fst (or "--fst case-control" if your subpopulations are defined by case/control status).

[kautilya]

Thanks a lot. Tried --fst case-control works really well. Great ease of use.

[arslanzaidipsu]

Hi, Didn't mean to hijack this post but I was wondering if there is a way to use --fst to calculate pairwise fst values for multiple populations? I tried --loop--assoc but it didn't work. Thanks!

[chrchang523]

Use --within to load a file defining your subpopulations.

[arslanzaidipsu]

I tried that. It gives one Fst value per SNP, so it's presumably calculating it over all subpopulations and not for all pairwise comparison (example headings given below):

SNP | Pop1 | Pop2 | Fst

Thank you.

[chrchang523]

Oops, missed the "pairwise" part of your remark. You will need to write a brief script with a double-for-loop for this; PLINK does not have that built-in.

[Myosotis1979]

Please how are the data formatted in subpopulation file ? Thanks...

[Simo]

Hi, it's a three column file, first two of the fam file plus a third one with the clusters (you can repeat the first column here or put whatever you need).