model_validation/tasks/SepSnvIndel.wdl

task SepSnvIndel {
	File vcf
	String sampleName
	String docker
	String cluster_config
	String disk_size
	
	command <<<

		cat ${vcf} | grep '#' > header
		cat ${vcf} | sed '/^#/d' | awk '$5!~/,/' > removed.body
		cat ${vcf} | sed '/^#/d' | awk '$5~/,/' > MNP.body
		cat header removed.body > ${sampleName}.MNPremoved.vcf
		cat header MNP.body > ${sampleName}.MNP.vcf

		rtg bgzip ${sampleName}.MNPremoved.vcf
		rtg index -f vcf ${sampleName}.MNPremoved.vcf.gz

		rtg vcffilter -i ${sampleName}.MNPremoved.vcf.gz -o ${sampleName}.normed.snv.vcf.gz --snps-only 

		rtg vcffilter -i ${sampleName}.MNPremoved.vcf.gz -o ${sampleName}.normed.indel.vcf.gz --non-snps-only
	

		gzip -d ${sampleName}.normed.snv.vcf.gz -c > ${sampleName}.normed.snv.vcf
		gzip -d ${sampleName}.normed.indel.vcf.gz -c > ${sampleName}.normed.indel.vcf
	>>>

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

	output {
		File MNP="${sampleName}.MNP.vcf"
		File nist_snv_gz="${sampleName}.normed.snv.vcf.gz"
		File nist_snv_gz_idx="${sampleName}.normed.snv.vcf.gz.tbi"
		File nist_indel_gz="${sampleName}.normed.indel.vcf.gz"
		File nist_indel_gz_idx="${sampleName}.normed.indel.vcf.gz.tbi"
		File nist_snv="${sampleName}.normed.snv.vcf"
		File nist_indel="${sampleName}.normed.indel.vcf"
	}
}