quartet_qc_vcf_input/tasks/benchmark.wdl

task benchmark {
	File vcf
	File benchmarking_dir
	File contig
	File sdf
	String project
	String sample = basename(vcf,".raw.vcf.gz")
	String docker
	String cluster_config
	String disk_size


	command <<<
		set -o pipefail
		set -e
		nt=$(nproc)
		mkdir -p /cromwell_root/tmp
		cp -r ${benchmarking_dir} /cromwell_root/tmp/

		gunzip ${vcf} -c > unzip.vcf
		cat unzip.vcf | grep '#CHROM' > name
		cat unzip.vcf | grep -v '#' > body
		cat body | grep -w '^chr1\|^chr2\|^chr3\|^chr4\|^chr5\|^chr6\|^chr7\|^chr8\|^chr9\|^chr10\|^chr11\|^chr12\|^chr13\|^chr14\|^chr15\|^chr16\|^chr17\|^chr18\|^chr19\|^chr20\|^chr21\|^chr22\|^chrX' > body.filtered
		cat unzip.vcf | grep '##' | grep -v 'contig' | cat - ${contig} name body.filtered > filtered.vcf

		echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL5" > LCL5_name
		echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL6" > LCL6_name
		echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL7" > LCL7_name
		echo -e "#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tLCL8" > LCL8_name


		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip filtered.vcf -c > ${project}.${sample}.rtg.vcf.gz
		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf ${project}.${sample}.rtg.vcf.gz
		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${project}.${sample}.rtg.vcf.gz -o ${sample}.rtg.SNV.vcf.gz --snps-only
		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${project}.${sample}.rtg.vcf.gz -o ${sample}.rtg.INDEL.vcf.gz --non-snps-only


		if [[ ${sample} =~ "LCL5" ]];then
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL5.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
			cat filtered.vcf | grep '##' > header
			cat filtered.vcf | grep -v '#' > body
			cat header LCL5_name body > LCL5.vcf
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL5.vcf -c > ${project}.${sample}.rtg.vcf.gz
	    elif [[ ${sample} =~ "LCL6" ]]; then
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL6.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
			cat filtered.vcf | grep '##' > header
			cat filtered.vcf | grep -v '#' > body
			cat header LCL6_name body > LCL6.vcf
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL6.vcf -c > ${project}.${sample}.rtg.vcf.gz
        elif [[ ${sample} =~ "LCL7" ]]; then
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL7.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
			cat filtered.vcf | grep '##' > header
			cat filtered.vcf | grep -v '#' > body
			cat header LCL7_name body > LCL7.vcf
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL7.vcf -c > ${project}.${sample}.rtg.vcf.gz
	    elif [[ ${sample} =~ "LCL8" ]]; then
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.SNV.vcf.gz -c ${sample}.rtg.SNV.vcf.gz -o ${sample}_SNV -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcfeval -b /cromwell_root/tmp/reference_datasets_v202103/LCL8.high.confidence.calls.INDEL.vcf.gz -c ${sample}.rtg.INDEL.vcf.gz -o ${sample}_INDEL -t ${sdf} --evaluation-regions=/cromwell_root/tmp/reference_datasets_v202103/Quartet.high.confidence.region.v202103.bed
			cat filtered.vcf | grep '##' > header
			cat filtered.vcf | grep -v '#' > body
			cat header LCL8_name body > LCL8.vcf
			/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip LCL8.vcf -c > ${project}.${sample}.rtg.vcf.gz
        else
        	echo "only for quartet samples"
        fi		


        cat ${sample}_SNV/summary.txt > ${project}.${sample}_SNV_precision_recall.txt
        cat ${sample}_INDEL/summary.txt > ${project}.${sample}_INDEL_precision_recall.txt
        
	>>>

	runtime {
		docker:docker
		cluster:cluster_config
		systemDisk:"cloud_ssd 40"
		dataDisk:"cloud_ssd " + disk_size + " /cromwell_root/"
	}

	output {
		File rtg_vcf = "${project}.${sample}.rtg.vcf.gz"
		File rtg_vcf_index = "${project}.${sample}.rtg.vcf.gz.tbi"
		File SNV_result = "${project}.${sample}_SNV_precision_recall.txt"
		File Indel_result = "${project}.${sample}_INDEL_precision_recall.txt"
	}
}