run_quickstat

#!/bin/bash

DEFVCFDIR=data/variant_calling
DEFTYPE=snv
VCFDIR=${VCFDIR:-$DEFVCFDIR}
TYPE=${TYPE:-$DEFTYPE}

for TYPE in snv indel
do
    for exp in wgs wxs
    do
        echo "---------------------------------------------------------------------------"
        echo "Total donors with $exp $TYPE data: " $( grep ${exp} ${VCFDIR}/union*.${TYPE}.all | awk '{print $2}'|sort -u|wc -l )

        declare -A private gnomad
        
        for caller in sanger mutect2 sanger,mutect2
        do
        private[$caller]=$( grep $exp ${VCFDIR}/union.${TYPE}.all | awk -v awkvar="$caller" '$9==awkvar'|wc -l)
        gnomad[$caller]=$( grep $exp ${VCFDIR}/union.${TYPE}.all | awk '$11!="."'|awk -v awkvar="$caller" '$9==awkvar'|wc -l)
        done
        total=$( cat ${VCFDIR}/union.${TYPE}.all | grep -c ${exp} )
        gnomad_all=$( grep $exp ${VCFDIR}/union.${TYPE}.all | awk '$11!="."'| wc -l)
        echo -e "Total\t${private[sanger]} / ${gnomad[sanger]}\t${private[mutect2]} / ${gnomad[mutect2]}\t${private[sanger,mutect2]} / ${gnomad[sanger,mutect2]}\t$total\t$gnomad_all"

        for region in cds exon utr3 utr5 protein_coding_promoter protein_coding_splice_site lncRNA_promoter lncRNA lncRNA_splice_site smallRNA smallRNA_promoter open_access
        do
            declare -A private gnomad
            
            for caller in sanger mutect2 sanger,mutect2
            do
            private[$caller]=$( grep $exp ${VCFDIR}/union_${region}.${TYPE}.all | awk -v awkvar="$caller" '$9==awkvar'|wc -l)
            gnomad[$caller]=$( grep $exp ${VCFDIR}/union_${region}.${TYPE}.all | awk '$11!="."'|awk -v awkvar="$caller" '$9==awkvar'|wc -l)
            done
            total=$( cat ${VCFDIR}/union_${region}.${TYPE}.all | grep -c ${exp} ) 
            gnomad_all=$( grep $exp ${VCFDIR}/union_${region}.${TYPE}.all | awk '$11!="."'| wc -l)
            echo -e "$region\t${private[sanger]} / ${gnomad[sanger]}\t${private[mutect2]} / ${gnomad[mutect2]}\t${private[sanger,mutect2]} / ${gnomad[sanger,mutect2]}\t$total\t$gnomad_all"
        done
    done
done