SNPs with high population differentiation from 1k Genome dataset

Question

I am trying to reproduce the results from this paper "Human genomic regions with exceptionally high levels of population differentiation identified from 911 whole-genome sequences".

Specifically I want to extract high differentiated ( or less) SNPs from 1k Genome data set representing high proportion of the population present.

The paper mentioned using Vcffixup from Vcflib (https://github.com/vcflib/vcflib) but there might still be other things I am missing.

• Update 2 (MWE)

So this is my code snippet.

## to dowload datafile, from linux terminal, use curl or wget
curl -O https://1000genomes.s3.amazonaws.com/release/20110521/ALL.chr22.phase1_release_v3.20101123.snps_indels_svs.genotypes.vcf.gz
### unzip the file
gunzip dowloaded file
### take the first few lines from the vcf file.
### since actual file is huge, the name of sfile for testing is 
### myfile.vcf
head -200 downloadedfile > myfile.vcf 

Now the next section was done in python environment (jupyter notebook).

## using the library pysam
from pysam import VariantFile
import pandas as pd 
with VariantFile(test_datafile) as vcf_reader:
for record in vcf_reader:
    sample_file = record.samples.keys() 
    alleles = [record.samples[x].allele_indices for x in record.samples]
    break ## just to make sure my data is as intended

Now to the VCF tools part - vcffixup was used in the paper.

## Install vcflib
   sudo apt-get install libvcflib-tools libvcflib-dev
### export path
   export PATH=$PATH:/usr/lib/vcflib/bin 
### refresh enviroment variables
   source ~/.bashrc 
## run vcffixup
  vcffixup myfile.vcf

A sample of vcffixup output is hown below:

So the questions questions I have is:

1. how to compute derived allele frequency differences at all variants for pairs of continent and pairs of population within continent

Thanks!