Add processSomatic and somaticFilter; motify annovar and vcf2maf module

tasks/annovar.wdl

@@ -1,9 +1,9 @@
 task annovar {
 
-	File varscan_snp
-	File varscan_indel
-	File database
 	String sample
+	File vcf
+	String annotated_vcf = basename(vcf,".vcf")
+	File database
 	String docker
 	String cluster_config
 	String disk_size
@@ -13,8 +13,7 @@ task annovar {
 		set -o pipefail
 		set -e
 		nt=$(nproc)
-		/installations/annovar/table_annovar.pl ${varscan_snp} ${database} -buildver hg38 -out ${sample}_varscan_snp -remove -protocol refGene,ensGene,knownGene,cytoBand,genomicSuperDups,ljb26_all,dbnsfp35c,intervar_20180118,cosmic70,exac03,gnomad211_exome,clinvar_20200316 -operation g,g,g,r,r,f,f,f,f,f,f,f -nastring . -vcfinput -thread $nt
-		/installations/annovar/table_annovar.pl ${varscan_indel} ${database} -buildver hg38 -out ${sample}_varscan_indel -remove -protocol refGene,ensGene,knownGene,cytoBand,genomicSuperDups,ljb26_all,dbnsfp35c,intervar_20180118,cosmic70,exac03,gnomad211_exome,clinvar_20200316 -operation g,g,g,r,r,f,f,f,f,f,f,f -nastring . -vcfinput -thread $nt
+		/installations/annovar/table_annovar.pl ${vcf} ${database} -buildver hg38 -out ${annotated_vcf} -remove -protocol refGene,ensGene,knownGene,cytoBand,genomicSuperDups,ljb26_all,dbnsfp35c,intervar_20180118,cosmic70,exac03,gnomad211_exome,clinvar_20200316 -operation g,g,g,r,r,f,f,f,f,f,f,f -nastring . -vcfinput -thread $nt
 	>>>
 	
 	runtime {
@@ -25,11 +24,8 @@ task annovar {
 	}
 
 	output {
-		File avinput_varscan_snp = "${sample}_varscan_snp.avinput"
-		File multianno_varscan_snp_txt = "${sample}_varscan_snp.hg38_multianno.txt"
-		File multianno_varscan_snp = "${sample}_varscan_snp.hg38_multianno.vcf"
-		File avinput_varscan_indel = "${sample}_varscan_indel.avinput"
-		File multianno_varscan_indel_txt = "${sample}_varscan_indel.hg38_multianno.txt"
-		File multianno_varscan_indel = "${sample}_varscan_indel.hg38_multianno.vcf"
+		File avinput = "${annotated_vcf}.avinput"
+		File multianno_txt = "${annotated_vcf}.hg38_multianno.txt"
+		File multianno = "${annotated_vcf}.hg38_multianno.vcf"
 	}
 }

tasks/processSomatic.wdl

@@ -0,0 +1,31 @@
+task processSomatic {
+    
+	String sample
+	File varscan_snp
+	File varscan_indel
+    
+	File ref_dir
+    String fasta
+    String varscan_docker
+    String cluster_config
+    String disk_size
+
+	command <<<
+		java -Xmx12g -jar /opt/VarScan.v2.4.3.jar processSomatic ${varscan_snp}
+		java -Xmx12g -jar /opt/VarScan.v2.4.3.jar processSomatic ${varscan_indel}
+	>>>
+	
+	runtime {
+		docker: docker
+		cluster: cluster_config
+		systemDisk: "cloud_ssd 40"
+		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+	}
+
+	output {
+		File varscan_snp_somatic_hc = "${sample}.VarScan.TN.SNP.Somatic.hc.vcf"
+		File varscan_snp_loh_hc = "${sample}.VarScan.TN.SNP.LOH.hc.vcf"
+		File varscan_indel_somatic_hc = "${sample}.VarScan.TN.INDEL.Somatic.hc.vcf"
+		File varscan_indel_loh_hc = "${sample}.VarScan.TN.INDEL.LOH.hc.vcf"
+	}
+}

tasks/varscan.wdl → tasks/somatic.wdl

@@ -1,23 +1,23 @@
 task varscan {
 
-    File ref_dir
-    String fasta
+    String sample
     File normal_bam
     File normal_bam_index
     File tumor_bam
     File tumor_bam_index
-    String sample
-    String varscan_docker
+    
+    File ref_dir
+    String fasta    
+    String docker
     String cluster_config
     String disk_size
 
-
     command <<<
-        samtools mpileup -f ${ref_dir}/${fasta} -B ${normal_bam} ${tumor_bam} | java -jar /opt/VarScan.v2.4.3.jar somatic --mpileup 1 --min-coverage 3 --min-coverage-normal 3 --min-coverage-tumor 3 --min-var-freq 0.08 --p-value 0.10 --somatic-p-value 0.05 --output-vcf 1 --output-snp ${sample}.VarScan.TN.SNP --output-indel ${sample}.VarScan.TN.INDEL --strand-filter 1
+        samtools mpileup -f ${ref_dir}/${fasta} -B ${normal_bam} ${tumor_bam} | java -Xmx12g -jar /opt/VarScan.v2.4.3.jar somatic --mpileup 1 --min-coverage 3 --min-coverage-normal 3 --min-coverage-tumor 3 --min-var-freq 0.08 --p-value 0.10 --somatic-p-value 0.05 --output-vcf 1 --output-snp ${sample}.VarScan.TN.SNP --output-indel ${sample}.VarScan.TN.INDEL --strand-filter 1
     >>>
 
     runtime {
-        docker:varscan_docker
+        docker: docker
         cluster: cluster_config
         systemDisk: "cloud_ssd 40"
         dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"

tasks/somaticFilter.wdl

@@ -0,0 +1,37 @@
+task somaticFilter {
+    
+    String sample
+    File varscan_snp_somatic_hc
+    File varscan_snp_loh_hc
+    File varscan_indel_somatic_hc
+    File varscan_indel_loh_hc
+    File varscan_indel
+
+    String docker
+    String cluster_config
+    String disk_size
+
+    command <<<
+        java -Xmx12g -jar /opt/VarScan.v2.4.3.jar somaticFilter ${varscan_snp_somatic_hc} --min-coverage 10 --min-reads2 2 --min-strands2 1 --min-avg-qual 20 --p-value 0.1 --indel-file ${varscan_indel} --output-file ${sample}.VarScan.TN.SNP.Somatic.filter.vcf
+
+        java -Xmx12g -jar /opt/VarScan.v2.4.3.jar somaticFilter ${varscan_snp_loh_hc} --min-coverage 10 --min-reads2 2 --min-strands2 1 --min-avg-qual 20 --p-value 0.1 --indel-file ${varscan_indel} --output-file ${sample}.VarScan.TN.SNP.LOH.filter.vcf
+
+        java -Xmx12g -jar /opt/VarScan.v2.4.3.jar somaticFilter ${varscan_indel_somatic_hc} --min-coverage 10 --min-reads2 2 --min-strands2 1 --min-avg-qual 20 --p-value 0.1 --output-file ${sample}.VarScan.TN.INDEL.Somatic.filter.vcf
+
+        java -Xmx12g -jar /opt/VarScan.v2.4.3.jar somaticFilter ${i}.indel.LOH.hc.vcf --min-coverage 10 --min-reads2 2 --min-strands2 1 --min-avg-qual 20 --p-value 0.1 --output-file ${sample}.VarScan.TN.INDEL.LOH.filter.vcf
+    >>>
+
+    runtime {
+        docker: docker
+        cluster: cluster_config
+        systemDisk: "cloud_ssd 40"
+        dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+    }
+
+    output {
+		File varscan_snp_somatic_filter = "${sample}.VarScan.TN.SNP.Somatic.filter.vcf"
+		File varscan_snp_loh_filter = "${sample}.VarScan.TN.SNP.LOH.filter.vcf"
+		File varscan_indel_somatic_filter = "${sample}.VarScan.TN.INDEL.Somatic.filter.vcf"
+		File varscan_indel_loh_filter = "${sample}.VarScan.TN.INDEL.LOH.filter.vcf"
+    }
+}

tasks/vcf2maf.wdl

@@ -1,15 +1,15 @@
 task vcf2maf {
 
-	File multianno_varscan_snp_txt
-	File multianno_varscan_indel_txt
 	String sample
+	File multianno_txt
+	String maf = basename(multianno_txt,".hg38_multianno.txt")
+		
 	String docker
 	String cluster_config
 	String disk_size
 
 	command <<<
-		vcf2maf ${multianno_varscan_snp_txt} ${sample}_varscan_snp.MAF
-		vcf2maf ${multianno_varscan_indel_txt} ${sample}_varscan_indel.MAF
+		vcf2maf ${multianno_txt} ${maf}.maf
 	>>>
 
 	runtime {
@@ -20,7 +20,6 @@ task vcf2maf {
 	}
 
 	output {
-		File varscan_snp_maf = "${sample}_varscan_snp.MAF"
-		File varscan_indel_maf = "${sample}_varscan_indel.MAF"
+		File maf = "${maf}.maf"
 	}
 }

workflow.wdl

@@ -1,4 +1,6 @@
-import "./tasks/varscan.wdl" as varscan
+import "./tasks/somatic.wdl" as somatic
+import "./tasks/processSomatic.wdl" as processSomatic
+import "./tasks/somaticFilter.wdl" as somaticFilter
 import "./tasks/annovar.wdl" as annovar
 import "./tasks/vcf2maf.wdl" as vcf2maf
 
@@ -14,7 +16,9 @@ workflow {{ project_name }} {
     String varscan_docker
     String annovar_docker
     String maftools_docker
-    
+    String disk_size
+    String cluster_config
+
     File ref_dir
     String fasta
     File dbmills_dir
@@ -22,42 +26,122 @@ workflow {{ project_name }} {
     File dbsnp_dir
     String dbsnp
     File regions
-    String disk_size
-    String cluster_config
-    
     File database
+    
 
-    call varscan.varscan as varscan {
+    call somatic.somatic as somatic {
         input:
-        ref_dir=ref_dir,
-        fasta=fasta,
+        sample=sample_id,
         normal_bam=normal_recaled_bam,
         tumor_bam=tumor_recaled_bam,
         normal_bam_index=normal_recaled_bam_index,
         tumor_bam_index=tumor_recaled_bam_index,
+        ref_dir=ref_dir,
+        fasta=fasta,        
+        docker=varscan_docker,	
+        disk_size=disk_size,
+        cluster_config=cluster_config
+    }
+
+    call processSomatic.processSomatic as processSomatic {
+        input:
         sample=sample_id,
-        varscan_docker=varscan_docker,	
+        varscan_snp=somatic.varscan_snp,
+        varscan_indel=somatic.varscan_indel,
+        ref_dir=ref_dir,
+        fasta=fasta,        
+        docker=varscan_docker,	
         disk_size=disk_size,
         cluster_config=cluster_config
     }
 
-    call annovar.annovar as annovar {
+    call somaticFilter.somaticFilter as somaticFilter {
         input:
+        sample=sample_id,
+        varscan_snp_somatic_hc=processSomatic.varscan_snp_somatic_hc,
+        varscan_snp_loh_hc=processSomatic.varscan_snp_loh_hc,
+        varscan_indel_somatic_hc=processSomatic.varscan_indel_somatic_hc,
+        varscan_indel_loh_hc=processSomatic.varscan_indel_loh_hc,
+        varscan_indel=somatic.varscan_indel,
+        ref_dir=ref_dir,
+        fasta=fasta,        
+        docker=varscan_docker,	
+        disk_size=disk_size,
+        cluster_config=cluster_config
+    }
+    
+    call annovar.annovar as snp_somatic_annovar {
+        input: 
+        sample=sample_id,
+        vcf=somaticFilter.varscan_snp_somatic_filter,
+        database=database,
         docker=annovar_docker,
+        cluster_config=cluster_config,
+        disk_size=disk_size
+    }
+
+    call vcf2maf.vcf2maf as snp_somatic_vcf2maf {
+        input:
+        sample=sample_id,
+        multianno_txt=snp_somatic_annovar.multianno_txt,
+        docker=maftools_docker,
+        cluster_config=cluster_config,
+        disk_size=disk_size
+    }
+    
+    call annovar.annovar as snp_loh_annovar {
+        input: 
+        sample=sample_id,
+        vcf=somaticFilter.varscan_snp_loh_filter,
         database=database,
-        varscan_snp=varscan.varscan_snp,
-        varscan_indel=varscan.varscan_indel,
+        docker=annovar_docker,
+        cluster_config=cluster_config,
+        disk_size=disk_size
+    }
+
+    call vcf2maf.vcf2maf as snp_loh_vcf2maf {
+        input:
+        sample=sample_id,
+        multianno_txt=snp_loh_annovar.multianno_txt,
+        docker=maftools_docker,
+        cluster_config=cluster_config,
+        disk_size=disk_size
+    }
+    
+    call annovar.annovar as indel_somatic_annovar {
+        input: 
         sample=sample_id,
+        vcf=somaticFilter.varscan_indel_somatic_filter,
+        database=database,
+        docker=annovar_docker,
         cluster_config=cluster_config,
         disk_size=disk_size
     }
 
-    call vcf2maf.vcf2maf as vcf2maf {
+    call vcf2maf.vcf2maf as indel_somatic_vcf2maf {
         input:
+        sample=sample_id,
+        multianno_txt=indel_somatic_annovar.multianno_txt,
         docker=maftools_docker,
-        multianno_varscan_snp_txt=annovar.multianno_varscan_snp_txt,
-        multianno_varscan_indel_txt=annovar.multianno_varscan_indel_txt,
+        cluster_config=cluster_config,
+        disk_size=disk_size
+    }
+    
+    call annovar.annovar as indel_loh_annovar {
+        input: 
+        sample=sample_id,
+        vcf=somaticFilter.varscan_indel_loh_filter,
+        database=database,
+        docker=annovar_docker,
+        cluster_config=cluster_config,
+        disk_size=disk_size
+    }
+
+    call vcf2maf.vcf2maf as indel_loh_vcf2maf {
+        input:
         sample=sample_id,
+        multianno_txt=indel_loh_annovar.multianno_txt,
+        docker=maftools_docker,
         cluster_config=cluster_config,
         disk_size=disk_size
     }