IGV check

inputs

@@ -1,16 +1,16 @@
 {
   "{{ project_name }}.LCL5_vcf": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL5.voted.mendelian.vcf.gz",
   "{{ project_name }}.LCL8_callable_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL8.consensus.merged.bed",
+  "{{ project_name }}.RTGdocker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/rtg-hap:v1.1",
   "{{ project_name }}.LCL6_variants_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL6.variants.bed",
   "{{ project_name }}.vcf_info": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/quartet.all.site.info.txt",
   "{{ project_name }}.LCL6_vcf_idx": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL6.voted.mendelian.vcf.gz.tbi",
   "{{ project_name }}.LCL5_variants_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL5.variants.bed",
   "{{ project_name }}.LCL8_HR_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL8.27.homo_ref.consensus.bed",
+  "{{ project_name }}.BEDdocker": "registry-internal.cn-shanghai.aliyuncs.com/pgx-docker-registry/bedtools:v2.27.1", 
   "{{ project_name }}.disk_size": "150",
   "{{ project_name }}.LCL7_variants_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL7.variants.bed",
   "{{ project_name }}.LCL7_callable_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL7.consensus.merged.bed",
-  "{{ project_name }}.filter_vcf.docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/rtg-hap:v1.1",
-  "{{ project_name }}.docker": "registry-internal.cn-shanghai.aliyuncs.com/pgx-docker-registry/bedtools:v2.27.1",
   "{{ project_name }}.LCL5_HR_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL5.27.homo_ref.consensus.bed",
   "{{ project_name }}.LCL6_vcf": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL6.voted.mendelian.vcf.gz",
   "{{ project_name }}.LCL7_HR_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL7.27.homo_ref.consensus.bed",
@@ -18,13 +18,13 @@
   "{{ project_name }}.LCL7_vcf": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL7.voted.mendelian.vcf.gz",
   "{{ project_name }}.LCL6_callable_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL6.consensus.merged.bed",
   "{{ project_name }}.LCL5_vcf_idx": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL5.voted.mendelian.vcf.gz.tbi",
-  "{{ project_name }}.PMRA_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/annotated.notValidatedIGV.vcf.bed",
+  "{{ project_name }}.PMRA_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/validation/annotated.notValidatedIGV.vcf.bed",
   "{{ project_name }}.LCL8_variants_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL8.variants.bed",
   "{{ project_name }}.cluster_config": "OnDemand bcs.a2.large img-ubuntu-vpc",
   "{{ project_name }}.LCL8_vcf_idx": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL8.voted.mendelian.vcf.gz.tbi",
   "{{ project_name }}.LCL8_vcf": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL8.voted.mendelian.vcf.gz",
   "{{ project_name }}.LCL7_vcf_idx": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL7.voted.mendelian.vcf.gz.tbi",
   "{{ project_name }}.LCL6_HR_bed": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/LCL6.27.homo_ref.consensus.bed",
-  "{{ project_name }}.bed_10X": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/linked.10x.26559.removed.vcf.bed"
+  "{{ project_name }}.bed_10X": "oss://pgx-result/renluyao/manuscript_v3.0/benchmark_region/dat/validation/10X.less.0.01.repeat.annotation.bed"
 }
 

tasks/filter_vcf.wdl

@@ -35,10 +35,16 @@ task filter_vcf {
 		gunzip LCL7.high.confidence.calls.vcf.gz
 		gunzip LCL8.high.confidence.calls.vcf.gz
 
-		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL5.high.confidence.calls.vcf -prefix LCL5
-		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL6.high.confidence.calls.vcf -prefix LCL6
-		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL7.high.confidence.calls.vcf -prefix LCL7
-		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL8.high.confidence.calls.vcf -prefix LCL8
+		cat LCL5.high.confidence.calls.vcf | grep -v '#' > LCL5.high.confidence.calls.body
+		cat LCL6.high.confidence.calls.vcf | grep -v '#' > LCL6.high.confidence.calls.body
+		cat LCL7.high.confidence.calls.vcf | grep -v '#' > LCL7.high.confidence.calls.body
+		cat LCL8.high.confidence.calls.vcf | grep -v '#' > LCL8.high.confidence.calls.body
+
+
+		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL5.high.confidence.calls.body -prefix LCL5
+		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL6.high.confidence.calls.body -prefix LCL6
+		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL7.high.confidence.calls.body -prefix LCL7
+		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL8.high.confidence.calls.body -prefix LCL8
 
 
 		cat LCL5.annotated.txt | awk '{print $1"\t"$2"\t.\t"$4"\t"$5"\t.\t.\tVOTE="$13"\tGT:ALT:DP\t"$10":"$18":"$17}' | cat LCL5.header - > LCL5.high.confidence.calls.annotated.vcf

tasks/remove_IGVrm_vcf.wdl

@@ -1,11 +1,10 @@
 task remove_IGVrm_vcf {
 	
-	File benchmark_region
+	File benchmark_filtered_region
 	File LCL5_annotated_vcf
 	File LCL6_annotated_vcf
 	File LCL7_annotated_vcf
 	File LCL8_annotated_vcf
-	File vcf_info
 	String docker
 	String disk_size
 	String cluster_config
@@ -13,35 +12,27 @@ task remove_IGVrm_vcf {
 
 	command <<<
 
-		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${LCL5_vcf} --include-bed=${benchmark_region} -o LCL5.high.confidence.calls.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip ${LCL5_annotated_vcf} -c > LCL5.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf LCL5.vcf.gz
 
-		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${LCL6_vcf} --include-bed=${benchmark_region} -o LCL6.high.confidence.calls.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip ${LCL6_annotated_vcf} -c > LCL6.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf LCL6.vcf.gz
 
-		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${LCL7_vcf} --include-bed=${benchmark_region} -o LCL7.high.confidence.calls.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip ${LCL7_annotated_vcf} -c > LCL7.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf LCL7.vcf.gz
 
-		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i ${LCL8_vcf} --include-bed=${benchmark_region} -o LCL8.high.confidence.calls.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg bgzip ${LCL8_annotated_vcf} -c > LCL8.vcf.gz
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg index -f vcf LCL8.vcf.gz
 
-		zcat LCL5.high.confidence.calls.vcf.gz | grep '#' > LCL5.header
-		zcat LCL6.high.confidence.calls.vcf.gz | grep '#' > LCL6.header
-		zcat LCL7.high.confidence.calls.vcf.gz | grep '#' > LCL7.header
-		zcat LCL8.high.confidence.calls.vcf.gz | grep '#' > LCL8.header
 
-		gunzip LCL5.high.confidence.calls.vcf.gz
-		gunzip LCL6.high.confidence.calls.vcf.gz
-		gunzip LCL7.high.confidence.calls.vcf.gz
-		gunzip LCL8.high.confidence.calls.vcf.gz
 
-		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL5.high.confidence.calls.vcf -prefix LCL5
-		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL6.high.confidence.calls.vcf -prefix LCL6
-		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL7.high.confidence.calls.vcf -prefix LCL7
-		python /opt/annotate_vcf.py -info ${vcf_info} -vcf LCL8.high.confidence.calls.vcf -prefix LCL8
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i LCL5.vcf.gz --include-bed=${benchmark_filtered_region} -o LCL5.high.confidence.calls.vcf.gz
 
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i LCL6.vcf.gz --include-bed=${benchmark_filtered_region} -o LCL6.high.confidence.calls.vcf.gz
 
-		cat LCL5.annotated.txt | awk '{print $1"\t"$2"\t.\t"$4"\t"$5"\t.\t.\tVOTE="$13"\tGT:ALT:DP\t"$10":"$18":"$17}' | cat LCL5.header - > LCL5.high.confidence.calls.annotated.vcf
-		cat LCL6.annotated.txt | awk '{print $1"\t"$2"\t.\t"$4"\t"$5"\t.\t.\tVOTE="$14"\tGT:ALT:DP\t"$10":"$20":"$19}' | cat LCL6.header - > LCL6.high.confidence.calls.annotated.vcf
-		cat LCL7.annotated.txt | awk '{print $1"\t"$2"\t.\t"$4"\t"$5"\t.\t.\tVOTE="$15"\tGT:ALT:DP\t"$10":"$22":"$21}' | cat LCL7.header - > LCL7.high.confidence.calls.annotated.vcf
-		cat LCL8.annotated.txt | awk '{print $1"\t"$2"\t.\t"$4"\t"$5"\t.\t.\tVOTE="$16"\tGT:ALT:DP\t"$10":"$24":"$23}' | cat LCL8.header - > LCL8.high.confidence.calls.annotated.vcf
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i LCL7.vcf.gz --include-bed=${benchmark_filtered_region} -o LCL7.high.confidence.calls.vcf.gz
 
+		/opt/rtg-tools/dist/rtg-tools-3.10.1-4d58ead/rtg vcffilter -i LCL8.vcf.gz --include-bed=${benchmark_filtered_region} -o LCL8.high.confidence.calls.vcf.gz
 
 		>>>
 
@@ -53,15 +44,10 @@ task remove_IGVrm_vcf {
 	}
 
 	output {
-		File LCL5_filtered_vcf = "LCL5.high.confidence.calls.vcf"
-		File LCL6_filtered_vcf = "LCL6.high.confidence.calls.vcf"
-		File LCL7_filtered_vcf = "LCL7.high.confidence.calls.vcf"
-		File LCL8_filtered_vcf = "LCL8.high.confidence.calls.vcf"
-		File LCL5_annotated_vcf = "LCL5.high.confidence.calls.annotated.vcf"
-		File LCL6_annotated_vcf = "LCL6.high.confidence.calls.annotated.vcf"
-		File LCL7_annotated_vcf = "LCL7.high.confidence.calls.annotated.vcf"
-		File LCL8_annotated_vcf = "LCL8.high.confidence.calls.annotated.vcf"
-
+		File LCL5_IGVrm_vcf = "LCL5.high.confidence.calls.vcf.gz"
+		File LCL6_IGVrm_vcf = "LCL6.high.confidence.calls.vcf.gz"
+		File LCL7_IGVrm_vcf = "LCL7.high.confidence.calls.vcf.gz"
+		File LCL8_IGVrm_vcf = "LCL8.high.confidence.calls.vcf.gz"
 	}
 }
 

workflow.wdl

@@ -3,6 +3,9 @@ import "./tasks/final_merge.wdl" as final_merge
 import "./tasks/homo_bed.wdl" as homo_bed
 import "./tasks/variant_bed.wdl" as variant_bed
 import "./tasks/filter_vcf.wdl" as filter_vcf
+import "./tasks/remove_IGVrm_bed.wdl" as remove_IGVrm_bed
+import "./tasks/remove_IGVrm_vcf.wdl" as remove_IGVrm_vcf
+
 
 workflow {{ project_name }} {
 
@@ -29,7 +32,8 @@ workflow {{ project_name }} {
 	File LCL7_vcf_idx
 	File LCL8_vcf_idx
 	File vcf_info
-	String docker
+	String RTGdocker
+	String BEDdocker	
 	String disk_size
 	String cluster_config
 	
@@ -39,7 +43,7 @@ workflow {{ project_name }} {
 		LCL6_callable_bed=LCL6_callable_bed,
 		LCL7_callable_bed=LCL7_callable_bed,
 		LCL8_callable_bed=LCL8_callable_bed,
-		docker=docker,
+		docker=BEDdocker,
 		disk_size=disk_size,
 		cluster_config=cluster_config
 	}
@@ -50,7 +54,7 @@ workflow {{ project_name }} {
 		LCL6_HR_bed=LCL6_HR_bed,
 		LCL7_HR_bed=LCL7_HR_bed,
 		LCL8_HR_bed=LCL8_HR_bed,
-		docker=docker,
+		docker=BEDdocker,
 		disk_size=disk_size,
 		cluster_config=cluster_config
 	}
@@ -61,7 +65,7 @@ workflow {{ project_name }} {
 		LCL6_variants_bed=LCL6_variants_bed,
 		LCL7_variants_bed=LCL7_variants_bed,
 		LCL8_variants_bed=LCL8_variants_bed,
-		docker=docker,
+		docker=BEDdocker,
 		disk_size=disk_size,
 		cluster_config=cluster_config
 	}
@@ -71,7 +75,7 @@ workflow {{ project_name }} {
 		callable_merged_intersect_bed=callable_loci.callable_merged_intersect_bed,
 		HR_merged_intersect_bed=homo_bed.HR_merged_intersect_bed,
 		variants_merged_bed=variant_bed.variants_merged_bed,
-		docker=docker,
+		docker=BEDdocker,
 		disk_size=disk_size,
 		cluster_config=cluster_config
 	}
@@ -88,6 +92,29 @@ workflow {{ project_name }} {
 		LCL6_vcf_idx=LCL6_vcf_idx,	
 		LCL7_vcf_idx=LCL7_vcf_idx,
 		LCL8_vcf_idx=LCL8_vcf_idx,
+		docker=RTGdocker,
+		disk_size=disk_size,
+		cluster_config=cluster_config
+	}
+
+	call remove_IGVrm_bed.remove_IGVrm_bed as remove_IGVrm_bed{
+		input:
+		benchmark_region=final_merge.benchmark_region,
+		bed_10X=bed_10X,
+		PMRA_bed=PMRA_bed,
+		docker=BEDdocker,
+		disk_size=disk_size,
+		cluster_config=cluster_config
+	}
+
+	call remove_IGVrm_vcf.remove_IGVrm_vcf as remove_IGVrm_vcf{
+		input:	
+		benchmark_filtered_region=remove_IGVrm_bed.benchmark_filtered_region,
+		LCL5_annotated_vcf=filter_vcf.LCL5_annotated_vcf,
+		LCL6_annotated_vcf=filter_vcf.LCL6_annotated_vcf,
+		LCL7_annotated_vcf=filter_vcf.LCL7_annotated_vcf,
+		LCL8_annotated_vcf=filter_vcf.LCL8_annotated_vcf,
+		docker=RTGdocker,
 		disk_size=disk_size,
 		cluster_config=cluster_config
 	}