first commit

defaults

@@ -0,0 +1,10 @@
+{   
+    "host_image":"oss://database-shao/pathseq_database/pathseq_host.fa.img",
+    "host_kmer":"oss://database-shao/pathseq_database/pathseq_host.bfi",
+    "microbe_dict":"oss://database-shao/pathseq_database/pathseq_microbe.dict",
+    "microbe_bwa_image":"oss://database-shao/pathseq_database/pathseq_microbe.fa.img",
+    "microbe_taxonomy":"oss://database-shao/pathseq_database/pathseq_taxonomy.db",
+    "pathseq_docker":"registry.cn-shanghai.aliyuncs.com/shaolab_docker/gatk4:4.2.6.1",
+    "pathseq_cluster":"OnDemand bcs.es.r.6xlarge img-ubuntu-vpc",
+    "disk_size":"300"
+}

inputs

@@ -0,0 +1,13 @@
+{
+    "{{ project_name }}.sample_id": "{{ sample_id }}",
+    "{{ project_name }}.fastq1": "{{ fastq1 }}",
+    "{{ project_name }}.fastq2": "{{ fastq2 }}",
+    "{{ project_name }}.host_image": "{{ host_image }}",
+    "{{ project_name }}.host_kmer": "{{ host_kmer }}",
+    "{{ project_name }}.microbe_dict": "{{ microbe_dict }}",
+    "{{ project_name }}.microbe_bwa_image": "{{ microbe_bwa_image }}",
+    "{{ project_name }}.microbe_taxonomy": "{{ microbe_taxonomy }}",
+    "{{ project_name }}.pathseq_docker": "{{ pathseq_docker }}",
+    "{{ project_name }}.pathseq_cluster": "{{ pathseq_cluster }}",
+    "{{ project_name }}.disk_size": "{{ disk_size }}"
+}

tasks/pathseq.wdl

@@ -0,0 +1,79 @@
+task pathseq {
+    String sample_id
+    File fastq1
+    File fastq2
+    File host_image
+    File host_kmer
+    File microbe_dict
+    File microbe_bwa_image
+    File microbe_taxonomy
+
+    String disk_size
+    String docker
+    String cluster
+
+    command <<<
+	
+        set -o pipefail
+        set -e
+
+        mkdir ubam
+        mkdir filter_log
+        mkdir clean_pair_bam
+        mkdir clean_unpaired_bam
+        mkdir pathseq_result
+        mkdir bwa_pair_bam
+        mkdir bwa_pair_unpair_bam
+
+        gatk FastqToSam \
+            -F1 ${fastq1} \
+            -F2 ${fastq2} \
+            -O ./ubam/${sample_id}.bam \
+            -SM ${sample_id}
+
+        time gatk --java-options "-Xmx185g" PathSeqFilterSpark  \
+            --input ./ubam/${sample_id}.bam \
+            --paired-output ./clean_pair_bam/${sample_id}_paired.bam \
+            --unpaired-output ./clean_unpaired_bam/${sample_id}_unpaired.bam \
+            --min-clipped-read-length 70 \
+            --kmer-file ${host_kmer} \
+            --filter-bwa-image ${host_image} \
+            --filter-metrics filter_log/${sample_id}.log 
+
+        time gatk --java-options "-Xmx185g" PathSeqBwaSpark  \
+            --paired-input ./clean_pair_bam/${sample_id}_paired.bam \
+            --unpaired-input ./clean_unpaired_bam/${sample_id}_unpaired.bam \
+            --paired-output bwa_pair_bam/${sample_id}_bwa_paired.bam \
+            --unpaired-output bwa_unpair_bam/${sample_id}_bwa_unpaired.bam \
+            --microbe-bwa-image ${microbe_bwa_image} \
+            --microbe-dict ${microbe_dict}
+
+
+        time gatk --java-options "-Xmx185g" PathSeqScoreSpark  \
+            --paired-input bwa_pair_bam/${sample_id}_bwa_paired.bam \
+            --unpaired-input bwa_unpair_bam/${sample_id}_bwa_unpaired.bam \
+            --taxonomy-file ${microbe_taxonomy} \
+            --scores-output pathseq_result/${sample_id}.pathseq.txt \
+            --output pathseq_result/${sample_id}.pathseq_reads.bam \
+            --min-score-identity 0.90 \
+            --identity-margin 0.02
+
+        
+
+    >>>
+	
+    runtime {
+        docker: docker
+        cluster: cluster
+        systemDisk: "cloud_ssd 40"
+        dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"
+    }
+
+    output {
+        File pathseq_txt = "pathseq_result/${sample_id}.pathseq.txt"
+        File pathseq_bam = "pathseq_result/${sample_id}.pathseq_reads.bam"
+        File pathseq_sbi = "pathseq_result/${sample_id}.pathseq_reads.bam.sbi"
+    }
+
+}
+

workflow.wdl

@@ -0,0 +1,37 @@
+import "./tasks/pathseq.wdl" as pathseq
+
+workflow {{ project_name }} {
+    String sample_id
+    File fastq1
+    File fastq2
+    String disk_size
+   
+    File host_image
+    File host_kmer
+    File microbe_dict
+    File microbe_bwa_image
+    File microbe_taxonomy
+
+    String pathseq_docker
+    String pathseq_cluster
+
+
+
+        call pathseq.pathseq as pathseq {
+                input:
+                sample_id=sample_id,
+                fastq1=fastq1,
+                fastq2=fastq2,
+                host_image=host_image,
+                host_kmer=host_kmer,
+                microbe_dict=microbe_dict,
+                microbe_bwa_image=microbe_bwa_image,
+                microbe_taxonomy=microbe_taxonomy,
+                disk_size=disk_size,
+                docker=pathseq_docker,
+                cluster=pathseq_cluster
+        }
+
+
+
+}