Add: replaceRG

defaults

@@ -4,9 +4,8 @@
   "normal_fastq_1": "",
   "normal_fastq_2": "",
   "tumor_deduped_bam": "",
-  "tumor_deduped_bam_index": "",
   "normal_deduped_bam": "",
-  "normal_deduped_bam_index": "",
+  "pl": "ILLUMINAL",
   "fasta": "GRCh38.d1.vd1.fa",
   "ref_dir": "oss://pgx-reference-data/GRCh38.d1.vd1/",
   "dbsnp": "dbsnp_146.hg38.vcf",
@@ -23,6 +22,8 @@
   "interval_padding": "0",
   "sentieon_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/sentieon-genomics:v202010.02",
   "varscan_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/varscan2:v2.4.3",
+  "pindel_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/pindel:0.2.5b9_3",
+  "samtools_docker": "registry-vpc.cn-shanghai.aliyuncs.com/pgx-docker-registry/samtools:v1.3.1",
   "annovar_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/annovar:v2019.10",
   "vep_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/vep:v104.0",
   "cache": "oss://pgx-reference-data/ensembl_vep/",

inputs

@@ -5,14 +5,15 @@
   "{{ project_name }}.normal_fastq_1": "{{ normal_fastq_1 }}",
   "{{ project_name }}.normal_fastq_2": "{{ normal_fastq_2 }}",
   "{{ project_name }}.tumor_deduped_bam": "{{ tumor_deduped_bam }}",
-  "{{ project_name }}.tumor_deduped_bam_index": "{{ tumor_deduped_bam_index }}",
   "{{ project_name }}.normal_deduped_bam": "{{ normal_deduped_bam }}",
-  "{{ project_name }}.normal_deduped_bam_index": "{{ normal_deduped_bam_index }}",
   "{{ project_name }}.SENTIEON_INSTALL_DIR": "{{ SENTIEON_INSTALL_DIR }}",
   "{{ project_name }}.SENTIEON_LICENSE": "{{ SENTIEON_LICENSE }}",
   "{{ project_name }}.sentieon_docker": "{{ sentieon_docker }}",
   "{{ project_name }}.varscan_docker": "{{ varscan_docker }}",
+  "{{ project_name }}.pindel_docker": "{{ pindel_docker }}",
+  "{{ project_name }}.samtools_docker": "{{ samtools_docker }}",
   "{{ project_name }}.annovar_docker": "{{ annovar_docker }}",
+  "{{ project_name }}.pl": "{{ pl }}",
   "{{ project_name }}.fasta": "{{ fasta }}",
   "{{ project_name }}.ref_dir": "{{ ref_dir }}",
   "{{ project_name }}.dbsnp": "{{ dbsnp }}",

tasks/Haplotyper.wdl

@@ -33,7 +33,7 @@ task Haplotyper {
     --interval ${regions} -r ${ref_dir}/${fasta} \
     -i ${recaled_bam} \
     --algo Haplotyper -d ${dbsnp_dir}/${dbsnp} \
-    ${sample}_hc.vcf
+    ${sample}.Haplotyper.vcf
   >>>
   
   runtime {
@@ -44,8 +44,8 @@ task Haplotyper {
   }
 
   output {
-    File vcf = "${sample}_hc.vcf"
-    File vcf_idx = "${sample}_hc.vcf.idx"
+    File vcf = "${sample}.Haplotyper.vcf"
+    File vcf_idx = "${sample}.Haplotyper.vcf.idx"
   }
 }
 

tasks/pindel.wdl

@@ -0,0 +1,76 @@
+task pindel {
+  
+  String sample_id
+  File bam
+  File bam_index
+  File ref_dir
+  String fasta
+
+  String docker
+  String cluster
+  String disk_size
+
+  command <<<
+	
+    set -o pipefail
+    set -e
+
+    mkdir ./pindel_result/
+    mkdir ./input
+
+    cp ${bam} ./input
+
+    bam_file_name=`echo ${bam}|awk -F "/" '{print $NF}'`
+    
+    samtools index -@ 4 ./input/$bam_file_name
+    
+    java "-Xmx16G" -jar /software/picard/picard.jar  CollectInsertSizeMetrics \
+      -H ./pindel_result/${sample_id}_picard.pdf \
+      -I ./input/$bam_file_name \
+      -O ./pindel_result/${sample_id}_picard.txt
+
+    mean_insert_size=`cat pindel_result/${sample_id}_picard.txt|sed -n '8p'|cut -f 6|cut -d . -f 1`
+    bam_file_name=`echo ${bam}|awk -F "/" '{print $NF}'`
+
+    echo -e "./input/$bam_file_name\t$mean_insert_size\t${sample_id}" > ${sample_id}_config.txt
+
+    pindel -i ${sample_id}_config.txt \
+      -f $${ref_dir}/${fasta} \
+      -o ./pindel_result/${sample_id} \
+      -c all \
+      -T 4 \
+      -x 4 \
+      -l \
+      -B 0 \
+      -M 3 \
+      -J /software/picard/hg38_ucsc_centromere.bed
+
+    grep "ChrID" pindel_result/${sample_id}_SI > pindel_result/${sample_id}_all_indel
+    grep "ChrID" pindel_result/${sample_id}_D >> pindel_result/${sample_id}_all_indel
+
+    awk -v chrID="chr1" '$8==chrID {print}' pindel_result/${sample_id}_all_indel > pindel_result/${sample_id}_indel
+    
+    for i in `seq 2 22`  X Y M
+      do
+        awk -v chrID=chr$i '$8==chrID {print}' pindel_result/${sample_id}_all_indel >> pindel_result/${sample_id}_indel
+      done
+
+    pindel2vcf -r ${ref_dir}/${fasta} \
+      -R GRCh38.d1.vd1 \
+      -d GDC \
+      -p pindel_result/${sample_id}_indel \
+      -v pindel_result/${sample_id}.pindel.indel.vcf
+    
+  >>>
+	
+  runtime {
+    docker: docker
+    cluster: cluster
+    systemDisk: "cloud_ssd 40"
+    dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+  }
+
+  output {
+    Array[File] pindel_result = glob("./pindel_result/${sample_id}*")
+  }
+}

tasks/replaceRG.wdl

@@ -0,0 +1,28 @@
+task replaceRG {
+
+  File bam
+  String sample
+  String group
+  String pl
+  String bam_name = basename(bam, ".bam")
+  String docker
+  String cluster_config
+  String disk_size
+
+  command <<<
+    samtools addreplacerg -r "@RG\tID:${group}\tSM:${sample}\tPL:${pl}" -o ${bam_name}.bam ${bam}
+    samtools index ${bam_name}.bam ${bam_name}.bam.bai
+  >>>
+
+  runtime {
+    docker: docker
+    cluster: cluster_config
+    systemDisk: "cloud_ssd 40"
+    dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+  }
+  output {
+    File bam = "${bam_name}.bam"
+    File bam_index = "${bam_name}.bam.bai"
+  }
+}
+

workflow.wdl

@@ -9,6 +9,8 @@ import "./tasks/TNscope.wdl" as TNscope
 import "./tasks/somatic.wdl" as somatic
 import "./tasks/processSomatic.wdl" as processSomatic
 import "./tasks/somaticFilter.wdl" as somaticFilter
+import "./tasks/replaceRG.wdl" as replaceRG
+import "./tasks/pindel.wdl" as pindel
 import "./tasks/ANNOVAR.wdl" as ANNOVAR
 import "./tasks/VEP.wdl" as VEP
 
@@ -21,9 +23,7 @@ workflow {{ project_name }} {
   File? normal_fastq_1
   File? normal_fastq_2
   File? tumor_deduped_bam
-  File? tumor_deduped_bam_index
   File? normal_deduped_bam
-  File? normal_deduped_bam_index
 
   Boolean input_fastq
   Boolean input_bam
@@ -32,9 +32,12 @@ workflow {{ project_name }} {
   String SENTIEON_LICENSE
   String sentieon_docker
   String varscan_docker
+  String pindel_docker
+  String samtools_docker
   String annovar_docker
   String vep_docker
 
+  String pl
   File ref_dir
   String fasta
   File dbmills_dir
@@ -72,13 +75,13 @@ workflow {{ project_name }} {
     if (tumor_fastq_1!= "") {
       call mapping.mapping as tumor_mapping {
         input: 
-        group=sample_id + '_tumor',
-        sample=sample_id + '_tumor',
+        group=sample_id + '.T',
+        sample=sample_id + '.T',
         fastq_1=tumor_fastq_1,
         fastq_2=tumor_fastq_2,
         SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
         SENTIEON_LICENSE=SENTIEON_LICENSE,
-        pl="ILLUMINAL",
+        pl=pl,
         fasta=fasta,
         ref_dir=ref_dir,
         docker=sentieon_docker,
@@ -94,7 +97,7 @@ workflow {{ project_name }} {
         ref_dir=ref_dir,
         sorted_bam=tumor_mapping.sorted_bam,
         sorted_bam_index=tumor_mapping.sorted_bam_index,
-        sample=sample_id + '_tumor',
+        sample=sample_id + '.T',
         docker=sentieon_docker,
         disk_size=disk_size,
         cluster_config=cluster_config
@@ -106,7 +109,7 @@ workflow {{ project_name }} {
         SENTIEON_LICENSE=SENTIEON_LICENSE,
         sorted_bam=tumor_mapping.sorted_bam,
         sorted_bam_index=tumor_mapping.sorted_bam_index,
-        sample=sample_id + '_tumor',
+        sample=sample_id + '.T',
         docker=sentieon_docker,
         disk_size=disk_size,
         cluster_config=cluster_config
@@ -120,7 +123,7 @@ workflow {{ project_name }} {
         ref_dir=ref_dir,
         deduped_bam=tumor_Dedup.deduped_bam,
         deduped_bam_index=tumor_Dedup.deduped_bam_index,
-        sample=sample_id + '_tumor',
+        sample=sample_id + '.T',
         docker=sentieon_docker,
         disk_size=disk_size,
         cluster_config=cluster_config
@@ -138,7 +141,7 @@ workflow {{ project_name }} {
         dbmills_dir=dbmills_dir,
         dbsnp=dbsnp,
         dbsnp_dir=dbsnp_dir,
-        sample=sample_id + '_tumor',
+        sample=sample_id + '.T',
         docker=sentieon_docker,
         disk_size=disk_size,
         cluster_config=cluster_config
@@ -148,13 +151,13 @@ workflow {{ project_name }} {
     if (normal_fastq_1!= "") {
       call mapping.mapping as normal_mapping {
         input: 
-        group=sample_id + '_normal',
-        sample=sample_id + '_normal',
+        group=sample_id + '.N',
+        sample=sample_id + '.N',
         fastq_1=normal_fastq_1,
         fastq_2=normal_fastq_2,
         SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
         SENTIEON_LICENSE=SENTIEON_LICENSE,
-        pl="ILLUMINAL",
+        pl=pl,
         fasta=fasta,
         ref_dir=ref_dir,
         docker=sentieon_docker,
@@ -170,7 +173,7 @@ workflow {{ project_name }} {
         ref_dir=ref_dir,
         sorted_bam=normal_mapping.sorted_bam,
         sorted_bam_index=normal_mapping.sorted_bam_index,
-        sample=sample_id + '_normal',
+        sample=sample_id + '.N',
         regions=regions,
         docker=sentieon_docker,
         disk_size=disk_size,
@@ -183,7 +186,7 @@ workflow {{ project_name }} {
         SENTIEON_LICENSE=SENTIEON_LICENSE,
         sorted_bam=normal_mapping.sorted_bam,
         sorted_bam_index=normal_mapping.sorted_bam_index,
-        sample=sample_id + '_normal',
+        sample=sample_id + '.N',
         docker=sentieon_docker,
         disk_size=disk_size,
         cluster_config=cluster_config
@@ -197,7 +200,7 @@ workflow {{ project_name }} {
         ref_dir=ref_dir,
         deduped_bam=normal_Dedup.deduped_bam,
         deduped_bam_index=normal_Dedup.deduped_bam_index,
-        sample=sample_id + '_normal',
+        sample=sample_id + '.N',
         regions=regions,
         docker=sentieon_docker,
         disk_size=disk_size,
@@ -216,19 +219,19 @@ workflow {{ project_name }} {
         dbmills_dir=dbmills_dir,
         dbsnp=dbsnp,
         dbsnp_dir=dbsnp_dir,
-        sample=sample_id + '_normal',
+        sample=sample_id + '.N',
         docker=sentieon_docker,
         disk_size=disk_size,
         cluster_config=cluster_config
       }
     }
-    
+
     if (haplotyper) {
       call Haplotyper.Haplotyper as Haplotyper {
         input:
         SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
         SENTIEON_LICENSE=SENTIEON_LICENSE,
-        sample=sample_id + '_normal',
+        sample=sample_id + '.N',
         fasta=fasta,
         ref_dir=ref_dir,
         recaled_bam=normal_BQSR.recaled_bam,
@@ -241,7 +244,21 @@ workflow {{ project_name }} {
         cluster_config=cluster_config
       }
     }
-    
+     
+    if (pindel) {
+      call pindel.pindel as pindel {
+        input:	
+        sample_id=sample_id + '.N',
+        bam=normal_BQSR.recaled_bam,
+        bam_index=normal_BQSR.recaled_bam_index,
+        fasta=fasta,
+        ref_dir=ref_dir,
+        docker=pindel_docker,
+        disk_size=disk_size,
+        cluster_config=cluster_config
+      }
+    }
+
     if (tnseq) {
       call TNseq.TNseq as TNseq {
         input:
@@ -252,8 +269,8 @@ workflow {{ project_name }} {
         normal_recaled_bam_index=normal_BQSR.recaled_bam_index,
         tumor_recaled_bam=tumor_BQSR.recaled_bam,
         tumor_recaled_bam_index=tumor_BQSR.recaled_bam_index,
-        normal_name=sample_id + "_normal",
-        tumor_name=sample_id + "_tumor",
+        normal_name=sample_id + ".N",
+        tumor_name=sample_id + ".T",
         fasta=fasta,
         ref_dir=ref_dir,
         regions=regions,
@@ -286,8 +303,8 @@ workflow {{ project_name }} {
           hg=hg,
           only_pass=only_pass,
           sample_id=sample_id,
-          tumor_id=sample_id + "_tumor",
-          normal_id=sample_id + "_normal",
+          tumor_id=sample_id + ".T",
+          normal_id=sample_id + ".N",
           ref_dir=ref_dir,
           fasta=fasta,
           vep_path=vep_path,
@@ -311,8 +328,8 @@ workflow {{ project_name }} {
         normal_recaled_bam_index=normal_BQSR.recaled_bam_index,
         tumor_recaled_bam=tumor_BQSR.recaled_bam,
         tumor_recaled_bam_index=tumor_BQSR.recaled_bam_index,
-        normal_name=sample_id + "_normal",
-        tumor_name=sample_id + "_tumor",
+        normal_name=sample_id + ".N",
+        tumor_name=sample_id + ".T",
         fasta=fasta,
         ref_dir=ref_dir,
         regions=regions,
@@ -345,8 +362,8 @@ workflow {{ project_name }} {
           hg=hg,
           only_pass=only_pass,
           sample_id=sample_id,
-          tumor_id=sample_id + "_tumor",
-          normal_id=sample_id + "_normal",
+          tumor_id=sample_id + ".T",
+          normal_id=sample_id + ".N",
           ref_dir=ref_dir,
           fasta=fasta,
           vep_path=vep_path,
@@ -420,8 +437,8 @@ workflow {{ project_name }} {
           hg=hg,
           only_pass=only_pass,
           sample_id=sample_id,
-          tumor_id=sample_id + "_tumor",
-          normal_id=sample_id + "_normal",
+          tumor_id=sample_id + ".T",
+          normal_id=sample_id + ".N",
           ref_dir=ref_dir,
           fasta=fasta,
           vep_path=vep_path,
@@ -435,22 +452,33 @@ workflow {{ project_name }} {
       }
     }
   }
-  
+
   if (input_bam) {
     if (tumor_deduped_bam != "") {
+      call replaceRG.replaceRG as tumor_replaceRG {
+        input:
+        bam=tumor_deduped_bam,
+        group=group_id + '.T',
+        sample=sample_id + '.T',
+        pl=pl,
+        docker=samtools_docker,
+        disk_size=disk_size,
+        cluster_config=cluster_config
+      }
+
       call BQSR.BQSR as tumor_BQSR_fb {
         input:
         SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
         SENTIEON_LICENSE=SENTIEON_LICENSE,
         fasta=fasta,
         ref_dir=ref_dir,
-        deduped_bam=tumor_deduped_bam,
-        deduped_bam_index=tumor_deduped_bam_index,
+        deduped_bam=tumor_replaceRG.bam,
+        deduped_bam_index=tumor_replaceRG.bam_index,
         db_mills=db_mills,
         dbmills_dir=dbmills_dir,
         dbsnp=dbsnp,
         dbsnp_dir=dbsnp_dir,
-        sample=sample_id + '_tumor',
+        sample=sample_id + '.T',
         docker=sentieon_docker,
         disk_size=disk_size,
         cluster_config=cluster_config
@@ -458,19 +486,30 @@ workflow {{ project_name }} {
     }
     
     if (normal_deduped_bam != "") {
+      call replaceRG.replaceRG as normal_replaceRG {
+        input:
+        bam=normal_deduped_bam,
+        group=group_id + '.N',
+        sample=sample_id + '.N',
+        pl=pl,
+        docker=samtools_docker,
+        disk_size=disk_size,
+        cluster_config=cluster_config
+      }
+      
       call BQSR.BQSR as normal_BQSR_fb {
         input:
         SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
         SENTIEON_LICENSE=SENTIEON_LICENSE,
         fasta=fasta,
         ref_dir=ref_dir,
-        deduped_bam=normal_deduped_bam,
-        deduped_bam_index=normal_deduped_bam_index,
+        deduped_bam=normal_replaceRG.bam,
+        deduped_bam_index=normal_replaceRG.bam_index,
         db_mills=db_mills,
         dbmills_dir=dbmills_dir,
         dbsnp=dbsnp,
         dbsnp_dir=dbsnp_dir,
-        sample=sample_id + '_normal',
+        sample=sample_id + '.N',
         docker=sentieon_docker,
         disk_size=disk_size,
         cluster_config=cluster_config
@@ -482,7 +521,7 @@ workflow {{ project_name }} {
         input:
         SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
         SENTIEON_LICENSE=SENTIEON_LICENSE,
-        sample=sample_id + '_normal',
+        sample=sample_id + '.N',
         fasta=fasta,
         ref_dir=ref_dir,
         recaled_bam=normal_BQSR_fb.recaled_bam,
@@ -495,7 +534,21 @@ workflow {{ project_name }} {
         cluster_config=cluster_config
       }
     }
-    
+
+    if (pindel) {
+      call pindel.pindel as pindel_fb {
+        input:	
+        sample_id=sample_id + '.N',
+        bam=normal_BQSR_fb.recaled_bam,
+        bam_index=normal_BQSR_fb.recaled_bam_index,
+        fasta=fasta,
+        ref_dir=ref_dir,
+        docker=pindel_docker,
+        disk_size=disk_size,
+        cluster_config=cluster_config
+      }
+    }
+
     if (tnseq) {
       call TNseq.TNseq as TNseq_fb {
         input:
@@ -506,8 +559,8 @@ workflow {{ project_name }} {
         normal_recaled_bam_index=normal_BQSR_fb.recaled_bam_index,
         tumor_recaled_bam=tumor_BQSR_fb.recaled_bam,
         tumor_recaled_bam_index=tumor_BQSR_fb.recaled_bam_index,
-        normal_name=sample_id + "_normal",
-        tumor_name=sample_id + "_tumor",
+        normal_name=sample_id + ".N",
+        tumor_name=sample_id + ".T",
         fasta=fasta,
         ref_dir=ref_dir,
         regions=regions,
@@ -540,8 +593,8 @@ workflow {{ project_name }} {
           hg=hg,
           only_pass=only_pass,
           sample_id=sample_id,
-          tumor_id=sample_id + "_tumor",
-          normal_id=sample_id + "_normal",
+          tumor_id=sample_id + ".T",
+          normal_id=sample_id + ".N",
           ref_dir=ref_dir,
           fasta=fasta,
           vep_path=vep_path,
@@ -565,8 +618,8 @@ workflow {{ project_name }} {
         normal_recaled_bam_index=normal_BQSR_fb.recaled_bam_index,
         tumor_recaled_bam=tumor_BQSR_fb.recaled_bam,
         tumor_recaled_bam_index=tumor_BQSR_fb.recaled_bam_index,
-        normal_name=sample_id + "_normal",
-        tumor_name=sample_id + "_tumor",
+        normal_name=sample_id + ".N",
+        tumor_name=sample_id + ".T",
         fasta=fasta,
         ref_dir=ref_dir,
         regions=regions,
@@ -599,8 +652,8 @@ workflow {{ project_name }} {
           hg=hg,
           only_pass=only_pass,
           sample_id=sample_id,
-          tumor_id=sample_id + "_tumor",
-          normal_id=sample_id + "_normal",
+          tumor_id=sample_id + ".T",
+          normal_id=sample_id + ".N",
           ref_dir=ref_dir,
           fasta=fasta,
           vep_path=vep_path,
@@ -674,8 +727,8 @@ workflow {{ project_name }} {
           hg=hg,
           only_pass=only_pass,
           sample_id=sample_id,
-          tumor_id=sample_id + "_tumor",
-          normal_id=sample_id + "_normal",
+          tumor_id=sample_id + ".T",
+          normal_id=sample_id + ".N",
           ref_dir=ref_dir,
           fasta=fasta,
           vep_path=vep_path,