Add a simple bioinformatics workflow w/ scatter/gather

bioinformatics/data/get_bam.sh

@@ -0,0 +1,4 @@
+#!/bin/bash 
+
+wget ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/phase1/data/HG00154/alignment/HG00154.chrom20.ILLUMINA.bwa.GBR.low_coverage.20101123.bam
+wget ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/phase1/data/HG00154/alignment/HG00154.chrom20.ILLUMINA.bwa.GBR.low_coverage.20101123.bam.bai

bioinformatics/tools/bcftools_filter.cwl

@@ -0,0 +1,29 @@
+#!/usr/bin/env cwl-runner
+
+cwlVersion: v1.0
+class: CommandLineTool
+label: "filter variants"
+
+baseCommand: "bcftools"
+stdout: $(inputs.output_filename)
+arguments: ["filter",  "-i"]
+
+inputs:
+    condition:
+        type: string
+        inputBinding:
+            position: 1
+
+    input_file:
+        type: File
+        streamable: true
+        inputBinding:
+            position: 2
+
+    output_filename:
+        type: string?
+        default: output_filtered.vcf
+
+outputs:
+    subset:
+        type: stdout

bioinformatics/tools/samtools_merge.cwl

@@ -0,0 +1,25 @@
+#!/usr/bin/env cwl-runner
+
+cwlVersion: v1.0
+class: CommandLineTool
+label: "merge bams"
+
+baseCommand: samtools
+arguments: ["merge"]
+
+inputs:
+    output_bam:
+        type: string
+        inputBinding:
+            position: 2
+
+    input_bams:
+        type: File[]
+        inputBinding:
+            position: 3
+
+outputs:
+    merged_bam:
+        type: File
+        outputBinding:
+            glob: $(inputs.output_bam)

bioinformatics/tools/samtools_sort.cwl

@@ -0,0 +1,25 @@
+#!/usr/bin/env cwl-runner
+
+cwlVersion: v1.0
+class: CommandLineTool
+label: "sort bam"
+
+baseCommand: samtools
+arguments: ["sort", "-o"]
+
+inputs:
+    output_bam:
+        type: string
+        inputBinding:
+            position: 3
+
+    input_bam:
+        type: File
+        inputBinding:
+            position: 4
+
+outputs:
+    sorted_bam:
+        type: File
+        outputBinding:
+            glob: $(inputs.output_bam)

bioinformatics/tools/samtools_view.cwl

@@ -0,0 +1,28 @@
+#!/usr/bin/env cwl-runner
+
+cwlVersion: v1.0
+class: CommandLineTool
+label: "filter bam"
+
+baseCommand: samtools_view_region.sh
+stdout: $(inputs.output_bam)
+
+inputs:
+    region:
+        type: File
+        inputBinding:
+            position: 1
+
+    input_bam:
+        type: File
+        secondaryFiles: .csi
+        inputBinding:
+            position: 2
+
+    output_bam:
+        type: string?
+        default: out.bam
+
+outputs:
+    filtered_bam:
+        type: stdout

bioinformatics/tools/samtools_view_region.sh

@@ -0,0 +1,8 @@
+#!/bin/bash
+
+readonly REGION_FILE="${1}"
+readonly INPUTFILE="${2:-/dev/stdin}"
+readonly OUTPUTFILE="${3:-/dev/stdout}"
+
+region=$( cat "${REGION_FILE}" )
+samtools view -b ${INPUTFILE} "${region}"

bioinformatics/tools/tabix_extract.cwl

@@ -0,0 +1,29 @@
+#!/usr/bin/env cwl-runner
+
+cwlVersion: v1.0
+class: CommandLineTool
+label: "extract region"
+
+baseCommand: "tabix"
+stdout: $(inputs.output_filename)
+arguments: ["-h"]   # include header
+
+inputs:
+    input_file:
+        type: File
+        secondaryFiles: .tbi
+        inputBinding:
+            position: 1
+
+    region:
+        type: string
+        inputBinding:
+            position: 2
+
+    output_filename:
+        type: string?
+        default: output_tabix.vcf
+
+outputs:
+    subset_by_region:
+        type: stdout

bioinformatics/tools/tabix_index.cwl

@@ -0,0 +1,24 @@
+#!/usr/bin/env cwl-runner
+
+cwlVersion: v1.0
+class: CommandLineTool
+label: "vcf index"
+baseCommand: "tabix"
+arguments: ["-p", "vcf"]
+requirements:
+    - class: InitialWorkDirRequirement
+      listing:
+          - $(inputs.vcf)
+inputs:
+    vcf:
+        type: File
+        inputBinding:
+            valueFrom:
+                $(self.basename)
+            position: 1
+outputs:
+    indexed_vcf:
+        type: File
+        secondaryFiles: .tbi
+        outputBinding:
+            glob: $(inputs.vcf.basename)

bioinformatics/tools/vcf_to_region.cwl

@@ -0,0 +1,28 @@
+#!/usr/bin/env cwl-runner
+
+cwlVersion: v1.0
+class: CommandLineTool
+label: "VCF to regions"
+
+baseCommand: vcf_to_region.sh
+
+inputs:
+    windowsize:
+        type: int
+        default: 50
+        inputBinding:
+            position: 1
+
+    input_file:
+        type: File
+        streamable: true
+        inputBinding:
+            position: 2
+
+outputs:
+  regions:
+    type:
+      type: array
+      items: File
+    outputBinding:
+      glob: "regions.*.txt"

bioinformatics/tools/vcf_to_region.sh

@@ -0,0 +1,26 @@
+#!/bin/bash
+
+function vcf_to_region {
+    local window="$1"
+    local input="$2"
+
+    local tmpfile=tmp.$$
+    awk -v window="$window" '/^[^#]/{
+        s = $2 - window;
+        e = $2 + window;
+        s = (s < 1) ? 1 : s;
+        printf "%s_%d\t%s:%d-%d\n", $1, $2, $1, s, e
+    }' "$input" > "$tmpfile"
+
+
+    while IFS=$'\t' read -ra line
+    do
+        echo "${line[1]}" > regions."${line[0]}".txt
+    done < "$tmpfile"
+    rm -f "$tmpfile"
+}
+
+readonly WINDOWSIZE=${1:-50}
+readonly INPUTFILE="${2:-/dev/stdin}"
+
+vcf_to_region "$WINDOWSIZE" "$INPUTFILE"

bioinformatics/tools/workflow.cwl

@@ -0,0 +1,85 @@
+#!/usr/bin/env cwl-runner
+cwlVersion: v1.0
+class: Workflow
+
+requirements:
+    - class: ScatterFeatureRequirement
+    - class: StepInputExpressionRequirement
+
+inputs:
+    input_vcf_file:
+        type: File
+    input_bam_file:
+        type: File
+    region:
+        type: string
+    condition:
+        type: string
+    windowsize:
+        type: int
+        default: 50
+    output_filename:
+        type: string
+
+outputs:
+    smallsortbam:
+        type: File
+        outputSource: samtools_sort/sorted_bam
+
+steps:
+    extract:
+        run: tabix_extract.cwl
+        in:
+            input_file:
+                source: input_vcf_file
+            region:
+                source: region
+        out: [subset_by_region]
+
+    filter:
+        run: bcftools_filter.cwl
+        in:
+            input_file: 
+                source: extract/subset_by_region
+            condition: 
+                source: condition
+        out: [subset]
+
+    regions:
+        run: vcf_to_region.cwl
+        in:
+            windowsize:
+                source: windowsize
+            input_file:
+                source: filter/subset
+        out: [regions]
+
+    samtools_view:
+        run: samtools_view.cwl
+        scatter: region
+        in:
+            input_bam:
+                source: input_bam_file
+            region:
+                source: regions/regions
+            output_bam:
+                valueFrom: $(inputs.region.nameroot).bam
+        out: [filtered_bam]
+
+    samtools_merge:
+        run: samtools_merge.cwl
+        in:
+            input_bams:
+                source: samtools_view/filtered_bam
+            output_bam:
+                source: output_filename
+        out: [merged_bam]
+
+    samtools_sort:
+        run: samtools_sort.cwl
+        in:
+            input_bam:
+                source: samtools_merge/merged_bam
+            output_bam:
+                source: output_filename
+        out: [sorted_bam]

bioinformatics/work-inputs.yml

@@ -0,0 +1,13 @@
+region: "20:50000-350000"
+
+condition: "INFO/AF < 0.05"
+
+input_vcf_file: 
+    class: File
+    path: data/chr20.vcf.gz
+
+input_bam_file: 
+    class: File
+    path: data/HG00154.chrom20.ILLUMINA.bwa.GBR.low_coverage.20101123.bam
+
+output_filename: sorted.bam