mendelian_concordance/tasks/mendelian.wdl

task mendelian {
	File mother_vcf_gz
	File father_vcf_gz
	File child_vcf_gz
	File mother_vcf_idx
	File father_vcf_idx
	File child_vcf_idx
	String mother_name
	String father_name
	String child_name
	String family_name
	File sdf
	String docker
	String cluster_config
	String disk_size
	
	command <<<
		rtg vcfmerge -o family.vcf.gz ${child_vcf_gz} ${mother_vcf_gz} ${father_vcf_gz}

		rtg vcfannotate -i family.vcf.gz -o family.anno.vcf.gz \
		--add-header "##SAMPLE=<ID=${mother_name},Sex=FEMALE>" \
		--add-header "##SAMPLE=<ID=${father_name},Sex=MALE>" \
		--add-header "##SAMPLE=<ID=${child_name},Sex=FEMALE>" \
		--add-header "##PEDIGREE=<Child=${child_name},Mother=${mother_name},Father=${father_name}>"

		rtg mendelian -i family.anno.vcf.gz -t ${sdf} -o ${family_name}.family.anno.mendelian.vcf.gz --lenient

		zcat ${family_name}.family.anno.mendelian.vcf.gz | grep '#' > header
		zcat ${family_name}.family.anno.mendelian.vcf.gz | grep -v '#'| grep 'MCU\|MCV' | cat header -  > ${family_name}.violation.vcf
		zcat ${family_name}.family.anno.mendelian.vcf.gz | grep -v '#'| grep -v 'MCV' | grep -v 'MCU' | cat header - > ${family_name}.consistent.vcf

		rtg bgzip ${family_name}.violation.vcf 
		rtg index -f vcf ${family_name}.violation.vcf.gz

		rtg bgzip ${family_name}.consistent.vcf
		rtg index -f vcf ${family_name}.consistent.vcf.gz

		zcat ${child_vcf_gz} | grep '##' > child_header
		zcat ${family_name}.violation.vcf.gz | grep -v '##' | cut -f1-10 | awk '$10!~/\./' | cat header - > ${child_name}.violation.vcf		
		zcat ${family_name}.consistent.vcf.gz | grep -v '##' | cut -f1-10 | awk '$10!~/\./' | cat header - > ${child_name}.consistent.vcf

		rtg bgzip ${child_name}.violation.vcf
		rtg index -f vcf ${child_name}.violation.vcf.gz

		rtg bgzip ${child_name}.consistent.vcf
		rtg index -f vcf ${child_name}.consistent.vcf.gz

	>>>

	runtime {
		docker:docker
		cluster: cluster_config
		systemDisk: "cloud_ssd 40"
		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
	}
	output {
		File family_vcf_gz = "${family_name}.family.anno.mendelian.vcf.gz"
		File family_vcf_index = "${family_name}.family.anno.mendelian.vcf.gz.tbi"
		File family_violation_vcf_gz = "${family_name}.violation.vcf.gz"
		File family_violation_vcf_idx = "${family_name}.violation.vcf.gz.tbi"
		File family_consistent_vcf_gz = "${family_name}.consistent.vcf.gz"
		File family_consistent_vcf_idx = "${family_name}.consistent.vcf.gz.tbi"
		File violation_vcf_gz = "${child_name}.violation.vcf.gz"
		File violation_vcf_idx = "${child_name}.violation.vcf.gz.tbi"
		File consistent_vcf_gz = "${child_name}.consistent.vcf.gz"
		File consistent_vcf_idx = "${child_name}.consistent.vcf.gz.tbi"
	}
}