Extract SNPs from VCF files

vcf-extract.sh

#!/bin/bash
#PBS -l walltime=24:00:00,nodes=1:ppn=1
#PBS -o output.file
#PBS -me
#PBS -S /bin/bash

set -e

module add apps/vcftools-0.1.12b
export PERL5LIB=/cm/shared/apps/VCFTOOLS-0.1.12b/vcftools_0.1.12b/perl/
module add apps/tabix-0.2.6

wd="/projects/UK10K/GOYA/imputed"
snplist="/panfs/panasas01/sscm/eprcr/GOYA/speliotes_snps.txt"
tempdir="${HOME}/tempdir"
out="${HOME}/output.vcf.gz"

#  Run the code

cd ${wd}
mkdir -p ${tempdir}


find . -maxdepth 1 -name '*.vcf.gz' | while read f; do

	# Output filenames - strip path and extensions
	fn=$(basename "${f}")
	fn="${fn%.*.*}"

	# Create copy of file with tab delimiter instead of space delimiter
	# Note - you may want to just do this to all the files because vcftools won't recognise them as they currently are
	zcat ${f} | tr ' ' '\t' | sed 's/exm-//g' | gzip -c > ${tempdir}/temp.vcf.gz

	# Extract SNPs
	vcftools --gzvcf ${tempdir}/temp.vcf.gz --snps ${snplist} --recode --keep-INFO-all --out ${tempdir}/${fn}

	# Prepare for merging
	bgzip ${tempdir}/${fn}.recode.vcf
	tabix ${tempdir}/${fn}.recode.vcf.gz
done
rm ${tempdir}/temp.vcf.gz

# Merge all output files
# note: maybe better to just select files that have an rs in them:
# zgrep rs tempdir/*.vcf.gz
vcf-merge ${tempdir}/*vcf.gz | bgzip -c > ${out}