revise fastp

defaults

@@ -21,11 +21,9 @@
     "disable_adapter_trimming": "0",
     "length_required": "50",
     "length_required1": "20",
-    "UMI": "0",
     "umi_len": "0",
     "umi_loc": "umi_loc",
     "qualified_quality_phred": "20",
-    "disable_quality_filtering": "1",
     "hisat2_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/hisat2:v2.1.0-2",
     "hisat2_cluster": "OnDemand bcs.a2.3xlarge img-ubuntu-vpc",
     "samtools_docker": "registry.cn-shanghai.aliyuncs.com/pgx-docker-registry/samtools:v1.3.1",

inputs

@@ -44,7 +44,7 @@
     "{{ project_name }}.count_docker": "{{ count_docker }}",
     "{{ project_name }}.count_cluster": "{{ count_cluster }}",
     "{{ project_name }}.count_length": "{{ count_length }}",
-    "{{ project_name }}.fastp_run": "{{ fastp_run }}",
+    "{{ project_name }}.trim_adapter": "{{ trim_adapter }}",
     "{{ project_name }}.pre_alignment_qc": "{{ pre_alignment_qc }}",
     "{{ project_name }}.qualimap_qc": "{{ qualimap_qc }}"
 }

tasks/fastp.wdl

@@ -8,48 +8,27 @@ task fastp {
     String cluster
     String disk_size
     String umi_loc	
+    String trim_adapter
     Int trim_front1
     Int trim_tail1
     Int max_len1
     Int trim_front2
     Int trim_tail2
     Int max_len2
-    Int disable_adapter_trimming
     Int length_required
     Int umi_len
-    Int UMI
     Int qualified_quality_phred
     Int length_required1
-    Int disable_quality_filtering
-   
+
 	command <<<
     nt=$(nproc)
-    mkdir -p /cromwell_root/tmp/fastp/
-	## 1.Disable_quality_filtering
-	if [ "${disable_quality_filtering}" == 0 ]
-    then
-    cp ${read1} /cromwell_root/tmp/fastp/{sample_id}_R1.fastq.tmp1.gz
-	cp ${read2} /cromwell_root/tmp/fastp/{sample_id}_R2.fastq.tmp1.gz
-    else
-    fastp --thread $nt --trim_front1 ${trim_front1} --trim_tail1 ${trim_tail1} --max_len1 ${max_len1} --trim_front2 ${trim_front2} --trim_tail2 ${trim_tail2} --max_len2 ${max_len2} -i ${read1} -I ${read2} -o /cromwell_root/tmp/fastp/${sample_id}_R1.fastq.tmp1.gz -O /cromwell_root/tmp/fastp/${sample_id}_R2.fastq.tmp1.gz -j ${sample_id}.json -h ${sample_id}.html
-    fi
-
-	## 2.UMI
-	if [ "${UMI}" == 0 ]
-    then
-    cp /cromwell_root/tmp/fastp/${sample_id}_R1.fastq.tmp1.gz /cromwell_root/tmp/fastp/${sample_id}_R1.fastq.tmp2.gz
-    cp /cromwell_root/tmp/fastp/${sample_id}_R2.fastq.tmp1.gz /cromwell_root/tmp/fastp/${sample_id}_R2.fastq.tmp2.gz
-    else
-    fastp --thread $nt -U --umi_loc=${umi_loc} --umi_len=${umi_len} --trim_front1 ${trim_front1} --trim_tail1 ${trim_tail1} --max_len1 ${max_len1} --trim_front2 ${trim_front2} --trim_tail2 ${trim_tail2} --max_len2 ${max_len2} -i /cromwell_root/tmp/fastp/${sample_id}_R1.fastq.tmp1.gz -I /cromwell_root/tmp/fastp/${sample_id}_R2.fastq.tmp1.gz -o /cromwell_root/tmp/fastp/${sample_id}_R1.fastq.tmp2.gz -O /cromwell_root/tmp/fastp/${sample_id}_R2.fastq.tmp2.gz -j ${sample_id}.json -h ${sample_id}.html
-    fi
-
-	## 3.Trim
-    if [ "${disable_adapter_trimming}" == 0 ]
-    then
-    fastp --thread $nt -l ${length_required} -q ${qualified_quality_phred} -u ${length_required1} --adapter_sequence ${adapter_sequence} --adapter_sequence_r2 ${adapter_sequence_r2} --detect_adapter_for_pe --trim_front1 ${trim_front1} --trim_tail1 ${trim_tail1} --max_len1 ${max_len1} --trim_front2 ${trim_front2} --trim_tail2 ${trim_tail2} --max_len2 ${max_len2} -i /cromwell_root/tmp/fastp/${sample_id}_R1.fastq.tmp2.gz -I /cromwell_root/tmp/fastp/${sample_id}_R2.fastq.tmp2.gz -o ${sample_id}_R1.fastq.gz -O ${sample_id}_R2.fastq.gz -j ${sample_id}.json -h ${sample_id}.html
+    
+    ## Trim
+    if [ "${trim_adapter}" != 'true' ]; then
+    cp ${read1} ${sample_id}_R1.fq.gz
+	cp ${read2} ${sample_id}_R2.fq.gz
     else
-    cp /cromwell_root/tmp/fastp/${sample_id}_R1.fastq.tmp2.gz ${sample_id}_R1.fastq.gz
-	cp /cromwell_root/tmp/fastp/${sample_id}_R2.fastq.tmp2.gz ${sample_id}_R2.fastq.gz
+    fastp --thread $nt -l ${length_required} -q ${qualified_quality_phred} -u ${length_required1} --adapter_sequence ${adapter_sequence} --adapter_sequence_r2 ${adapter_sequence_r2} --detect_adapter_for_pe --trim_front1 ${trim_front1} --trim_tail1 ${trim_tail1} --max_len1 ${max_len1} --trim_front2 ${trim_front2} --trim_tail2 ${trim_tail2} --max_len2 ${max_len2} -i ${read1} -I ${read2} -o ${sample_id}_R1.fq.gz -O ${sample_id}_R2.fq.gz -j ${sample_id}.json -h ${sample_id}.html
     fi
    >>>
    
@@ -63,7 +42,7 @@ task fastp {
    output {
       File json = "${sample_id}.json"
       File report = "${sample_id}.html"
-      File Trim_R1 = "${sample_id}_R1.fastq.gz"
-      File Trim_R2 = "${sample_id}_R2.fastq.gz"
+      File trim_R1 = "${sample_id}_R1.fq.gz"
+      File trim_R2 = "${sample_id}_R2.fq.gz"
    }
 }

tasks/hisat2_fp.wdl

@@ -1,29 +0,0 @@
-task hisat2_fp {
-   File idx
-   File read_1P
-   File read_2P
-   String idx_prefix
-   String sample_id
-   String docker
-   String cluster
-   String disk_size
-   
-
-   command {
-      nt=$(nproc)
-      hisat2 -t -p $nt -x ${idx}/${idx_prefix} -1 ${read_1P} -2 ${read_2P} -S ${sample_id}.sam  --un-conc-gz ${sample_id}_un.fq.gz
-   }
-   
-   runtime { 
-      docker: docker
-      cluster: cluster
-      systemDisk: "cloud_ssd 40"
-      dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
-   }
-
-   output {
-      File sam=sample_id + ".sam"
-      File unmapread_1p=sample_id + "_un.fq.1.gz"
-      File unmapread_2p=sample_id + "_un.fq.2.gz"
-   }
-}

workflow.wdl

@@ -53,15 +53,13 @@ workflow {{ project_name }} {
 	Int disable_adapter_trimming
 	Int length_required
 	Int umi_len
-	Int UMI
 	Int qualified_quality_phred
 	Int length_required1
-	Int disable_quality_filtering
 	Int insert_size
 
 	Boolean pre_alignment_qc
 	Boolean qualimap_qc
-	Boolean fastp_run
+	Boolean trim_adapter
 	
 	if (pre_alignment_qc) {
 		call fastqc.fastqc as fastqc {
@@ -84,60 +82,44 @@ workflow {{ project_name }} {
 			disk_size=disk_size
 			}	
 			}
-	
-	if (fastp_run){
-		call fastp.fastp as fastp {
-			input:
-			sample_id=sample_id,
-			read1=read1,
-			read2=read2,
-			docker=fastp_docker,
-			cluster=fastp_cluster,
-			disk_size=disk_size,
-			adapter_sequence=adapter_sequence,
-			adapter_sequence_r2=adapter_sequence_r2,
-			umi_loc=umi_loc,
-			trim_front1=trim_front1,
-			trim_tail1=trim_tail1,
-			max_len1=max_len1,
-			trim_front2=trim_front2,
-			trim_tail2=trim_tail2,
-			max_len2=max_len2,
-			disable_adapter_trimming=disable_adapter_trimming,
-			length_required=length_required,
-			umi_len=umi_len,
-			UMI=UMI,
-			qualified_quality_phred=qualified_quality_phred,
-			length_required1=length_required1,
-			disable_quality_filtering=disable_quality_filtering
-			}
-
-		call hisat2_fp.hisat2_fp as hisat2_fp {
-			input:
-			sample_id=sample_id,
-			idx=idx,
-			idx_prefix=idx_prefix,
-			read_1P=fastp.Trim_R1,
-			read_2P=fastp.Trim_R2,
-			docker=hisat2_docker,
-			cluster=hisat2_cluster,
-			disk_size=disk_size
-			}
-			}
 
-	if (!fastp_run){
-		call hisat2.hisat2 as hisat2 {
-			input:
-			sample_id=sample_id,
-			docker=hisat2_docker,
-			cluster=hisat2_cluster,
-			idx=idx,
-			idx_prefix=idx_prefix,
-			read_1P=read1,
-			read_2P=read2,
-			disk_size=disk_size
-			}
-			}
+	call fastp.fastp as fastp {
+		input:
+		sample_id=sample_id,
+		read1=read1,
+		read2=read2,
+		docker=fastp_docker,
+		cluster=fastp_cluster,
+		disk_size=disk_size,
+		adapter_sequence=adapter_sequence,
+		adapter_sequence_r2=adapter_sequence_r2,
+		umi_loc=umi_loc,
+		trim_front1=trim_front1,
+		trim_tail1=trim_tail1,
+		max_len1=max_len1,
+		trim_front2=trim_front2,
+		trim_tail2=trim_tail2,
+		max_len2=max_len2,
+		disable_adapter_trimming=disable_adapter_trimming,
+		length_required=length_required,
+		umi_len=umi_len,
+		qualified_quality_phred=qualified_quality_phred,
+		length_required1=length_required1
+		trim_adapter=trim_adapter
+		}
+			
+	call hisat2.hisat2 as hisat2 {
+		input:
+		sample_id=sample_id,
+		docker=hisat2_docker,
+		cluster=hisat2_cluster,
+		idx=idx,
+		idx_prefix=idx_prefix,
+		read_1P=fastp.trim_R1,
+		read_2P=fastp.trim_R2,
+		disk_size=disk_size
+		}
+		}
 	
 	call samtools.samtools as samtools {
 		input: