germline_variants.sh

conda activate primary-env

function extract_gts() {
    i=${1}
    patient=$(echo ${i} | egrep -o "LTX[0-9]+" | head -n 1)
    sample=$(echo ${i} | cut -f 5 -d "/")
    vcf2tsv -g ${i} | \
    awk -v patient=${patient} -v fname=${i} -v OFS="\t" -v sample=${sample} 'NR == 1 { print "patient", "sample", "fname", $0 } NR > 1 { print patient, sample, fname, $0 }' | \
    awk '!arr[$0]++' > gl_genotypes/${patient}.tsv
}

export -f extract_gts

# Get a list of germline files
ls release_*/**/exome/Platypus/*GL/Analysis_SNPs/*raw_GL_calls.vcf > gl_files.txt

# Parallel extract genotypes
parallel --verbose -j 32 extract_gts :::: gl_files.txt


# Combine genotype files
tut stack *.tsv | pigz -p 32 > gl_genotypes.tsv.gz