extract_individuals.py

from subprocess import call, run
from multiprocessing import Pool

command = 'wget -r -l 1 ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/release/20130502/'
run(command, shell=True)

path = '/storage3/1000genomes/ftp.1000genomes.ebi.ac.uk/vol1/ftp/release/20130502'

ind_file = open('{}/individuals.txt'.format(path))

individuals  = []
for line in ind_file:
	individuals.append(line.strip())

lista = list(range(1,23))
vcfs = {}
for i in lista:
    # print(i)
    vcf = '{}/ALL.chr{}.phase3_shapeit2_mvncall_integrated_v5a.20130502.genotypes.vcf.gz'.format(path, i)
    vcfs[i] = vcf

vcfs['X'] = '{}/ALL.chrX.phase3_shapeit2_mvncall_integrated_v1b.20130502.genotypes.vcf.gz'.format(path)
vcfs['Y'] = '{}/ALL.chrY.phase3_integrated_v2a.20130502.genotypes.vcf.gz'.format(path)
vcfs['MT'] = '{}/ALL.chrMT.phase3_callmom-v0_4.20130502.genotypes.vcf.gz'.format(path)

commands = []

for individual in individuals:
	for key in vcfs:
		command = 'bcftools view -s {} -O z -o samples/{}.{}.vcf.gz {}'.format(individual, individual, key, vcfs[key])
		commands.append(command)

def extract_samples(command):
	run(command, shell=True)

with Pool(8) as p:
	print(p.map(extract_samples, commands))

def concat_sample_vcfs(command):
	print(command)