Merge pull request #12 from mskcc/develop

Release - 1.1.1

.gitmodules

@@ -1,6 +1,3 @@
-[submodule "modules/sample/resolve-pdx"]
-	path = modules/sample/resolve-pdx
-	url = https://github.com/mskcc/resolve-pdx
 [submodule "modules/pair/helix_filters_01"]
 	path = modules/pair/helix_filters_01
 	url = https://github.com/mskcc/helix_filters_01

modules/sample/align_sample.cwl

@@ -0,0 +1,92 @@
+class: Workflow
+cwlVersion: v1.0
+id: align_sample
+label: align_sample
+inputs:
+  - id: reference_sequence
+    type: File
+    secondaryFiles:
+      - .amb
+      - .ann
+      - .bwt
+      - .pac
+      - .sa
+      - .fai
+  - id: r1
+    type: 'File[]'
+  - id: r2
+    type: 'File[]'
+  - id: sample_id
+    type: string
+  - id: lane_id
+    type: 'string[]'
+outputs:
+  - id: sample_id_output
+    outputSource:
+      - bwa_sort/sample_id_output
+    type:
+      - string
+      - type: array
+        items: string
+  - id: output_md_metrics
+    outputSource:
+      - gatk_markduplicatesgatk/output_md_metrics
+    type: File
+  - id: output_merge_sort_bam
+    outputSource:
+      - samtools_merge/output_file
+    type: File
+  - id: output_md_bam
+    outputSource:
+      - gatk_markduplicatesgatk/output_md_bam
+    type: File
+
+steps:
+  - id: samtools_merge
+    in:
+      - id: input_bams
+        source:
+          - bwa_sort/output_file
+    out:
+      - id: output_file
+    run: ../../tools/samtools.merge/1.9/samtools.merge.cwl
+  - id: bwa_sort
+    in:
+      - id: r1
+        source: r1
+      - id: r2
+        source: r2
+      - id: reference_sequence
+        source: reference_sequence
+      - id: read_pair
+        valueFrom: ${ var data = []; data.push(inputs.r1); data.push(inputs.r2); return data; }
+      - id: sample_id
+        source: sample_id
+      - id: lane_id
+        source: lane_id
+    out:
+      - id: output_file
+      - id: sample_id_output
+      - id: lane_id_output
+    run: align_sort_bam.cwl
+    label: bwa_sort
+    scatter:
+      - r1
+      - r2
+      - lane_id
+    scatterMethod: dotproduct
+  - id: gatk_markduplicatesgatk
+    in:
+      - id: input
+        source: samtools_merge/output_file
+    out:
+      - id: output_md_bam
+      - id: output_md_metrics
+    run: >-
+      ../../tools/gatk.mark_duplicates/4.1.0.0/gatk.mark_duplicates.cwl
+    label: GATK MarkDuplicates
+requirements:
+  - class: SubworkflowFeatureRequirement
+  - class: ScatterFeatureRequirement
+  - class: InlineJavascriptRequirement
+  - class: StepInputExpressionRequirement

modules/sample/align_sort_bam.cwl

@@ -0,0 +1,63 @@
+class: Workflow
+cwlVersion: v1.0
+id: bwa_sort
+label: bwa_sort
+inputs:
+  - id: reference_sequence
+    type: File
+    secondaryFiles:
+      - .amb
+      - .ann
+      - .bwt
+      - .pac
+      - .sa
+      - .fai
+  - id: read_pair
+    type: 'File[]'
+  - id: sample_id
+    type: string
+  - id: lane_id
+    type: string
+outputs:
+  - id: output_file
+    outputSource:
+      - samtools_sort/output_file
+    type: File
+  - id: sample_id_output
+    outputSource:
+      - sample_id
+    type: string
+  - id: lane_id_output
+    outputSource:
+      - lane_id
+    type: string
+steps:
+  - id: bwa_mem
+    in:
+      - id: reads
+        source:
+          - read_pair
+      - id: reference
+        source: reference_sequence
+      - id: sample_id
+        source: sample_id
+      - id: lane_id
+        source: lane_id
+    out:
+      - id: output_sam
+    run: ../../tools/bwa-mem/0.7.12/bwa_mem_with_read_groups.cwl
+  - id: samtools_sort
+    in:
+      - id: input
+        source: sam_to_bam/output_bam
+    out:
+      - id: output_file
+    run: ../../tools/samtools.sort/1.3.1/samtools.sort.cwl
+  - id: sam_to_bam
+    in:
+      - id: input
+        source: bwa_mem/output_sam
+    out:
+      - id: output_bam
+    run: ../../tools/samtools.view/1.3.1/samtools.view.cwl
+requirements: []

modules/sample/resolve-pdx.cwl

@@ -0,0 +1,104 @@
+#!/usr/bin/env cwl-runner
+cwlVersion: v1.0
+class: Workflow
+id: resolve-pdx
+requirements:
+  StepInputExpressionRequirement: {}
+  MultipleInputFeatureRequirement: {}
+  ScatterFeatureRequirement: {}
+  SubworkflowFeatureRequirement: {}
+  InlineJavascriptRequirement: {}
+
+inputs:
+  r1:
+    type: File[]
+
+  r2:
+    type: File[]
+
+  sample_id:
+    type: string
+
+  lane_id:
+    type: string[]
+
+  mouse_reference:
+    type: File
+    secondaryFiles:
+      - .amb
+      - .ann
+      - .bwt
+      - .pac
+      - .sa
+      - .fai
+
+  human_reference:
+    type: File
+    secondaryFiles:
+      - .amb
+      - .ann
+      - .bwt
+      - .pac
+      - .sa
+      - .fai
+
+outputs:
+  disambiguate_bam:
+    type: File
+    outputSource: run_disambiguate/disambiguate_a_bam
+  summary:
+    type: File
+    outputSource: run_disambiguate/summary
+
+steps:
+  align_to_human:
+    run: align_sample.cwl
+    in:
+      prefix: sample_id
+      reference_sequence: human_reference
+      r1: r1
+      r2: r2
+      sample_id:
+        valueFrom: ${ return inputs.prefix + "_human"; }
+      lane_id: lane_id
+    out: [ output_merge_sort_bam ]
+
+  align_to_mouse:
+    run: align_sample.cwl
+    in:
+      prefix: sample_id
+      reference_sequence: mouse_reference
+      r1: r1
+      r2: r2
+      sample_id:
+        valueFrom: ${ return inputs.prefix + "_mouse"; }
+      lane_id: lane_id
+    out: [ output_merge_sort_bam ]
+
+  name_sort_human:
+    run: ../../tools/samtools.sort/1.3.1/samtools.sort.cwl
+    in:
+      input: align_to_human/output_merge_sort_bam
+      sort_by_name:
+        valueFrom: ${ return true; }
+    out: [ output_file ]
+
+  name_sort_mouse:
+    run: ../../tools/samtools.sort/1.3.1/samtools.sort.cwl
+    in:
+      input: align_to_mouse/output_merge_sort_bam
+      sort_by_name:
+        valueFrom: ${ return true; }
+    out: [ output_file ]
+
+  run_disambiguate:
+    run: ../../tools/disambiguate/1.0.0/disambiguate.cwl
+    in:
+      prefix: sample_id
+      aligner:
+        valueFrom: ${ return "bwa"; }
+      output_dir:
+        valueFrom: ${ return inputs.prefix + "_disambiguated"; }
+      species_a_bam: name_sort_human/output_file
+      species_b_bam: name_sort_mouse/output_file
+    out: [ disambiguate_a_bam, disambiguate_b_bam, summary ]