high_confidence_calls_manuscript/tasks/merge.wdl

task merge {
	Array[File] family_vcf_gz
	Array[File] family_vcf_idx
	String sample
	String docker
	String cluster_config
	String disk_size
	
	command <<<

		rtg vcfmerge --force-merge-all -o ${sample}.merged.vcf.gz ${sep=" " family_vcf_gz}

		zcat ${sample}.merged.vcf.gz | grep -v '#' | cut -f1-2 | sed s'/\t/_/g' | sort | uniq -c | sed 's/\s\+/\t/g' | awk '{ if ($1 != 1) { print } }' | cut -f3 > ${sample}.vcf_dup.txt

	>>>

	runtime {
		docker:docker
		cluster: cluster_config
		systemDisk: "cloud_ssd 40"
		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
	}
	output {
		File merged_vcf_gz = "${sample}.merged.vcf.gz"
		File merged_vcf_idx = "${sample}.merged.vcf.gz.tbi"
		File vcf_dup = "${sample}.vcf_dup.txt"
	}
}