mixdata.sh

#!/usr/bin/env bash

coverage=${COVERAGE:-3000}
percentage=${PERCENTAGE:-1}
outputprefix=${OUTPUT_PREFIX:-mix}
numdatasets=$#

coverageMajority=$((coverage * (100 - (numdatasets - 1) * percentage) / 100))
coverageMinority=$((coverage * percentage / 100))

samtools view -H $1 > ${outputprefix}.sam

for i in ${@:2}
do
    samtools view -H $i | grep '^@RG' >> ${outputprefix}.sam
done

samtools view $1 | shuf | head -n $coverageMajority >> ${outputprefix}.sam
shift

while test $# -gt 0
do
    samtools view $1 | shuf | head -n $coverageMinority >> ${outputprefix}.sam
    shift
done

samtools view -@10 -bS ${outputprefix}.sam > ${outputprefix}.bam
rm ${outputprefix}.sam