Making code Dockstore compatible

Dockerfile

@@ -1,13 +1,12 @@
 FROM ubuntu
 
-RUN apt-get update && apt-get install -y wget build-essential git-core cmake zlib1g-dev libncurses-dev python python-dev python-pip
+RUN apt-get update && apt-get install -y wget build-essential git-core cmake zlib1g-dev libncurses-dev python python-dev python-pip samtools
 RUN	pip install pysam
 
 WORKDIR /opt
 RUN wget https://github.com/genome/somatic-sniper/archive/v1.0.5.0.tar.gz && tar xvzf v1.0.5.0.tar.gz && rm v1.0.5.0.tar.gz
 
 RUN cd /opt/somatic-sniper-1.0.5.0 && mkdir build && cd build && cmake ../ && make deps && make -j && make install
-RUN cp /opt/somatic-sniper-1.0.5.0/build/vendor/samtools/samtools /usr/bin/
 
 COPY SomaticSniper.py /opt/ 
-RUN chmod +x /opt/SomaticSniper.py
+RUN chmod +x /opt/SomaticSniper.py

Dockstore.json

@@ -0,0 +1,18 @@
+{
+  "reference": {
+    "path": "http://hgwdev.cse.ucsc.edu/~jeltje/public_data/genome.fa.gz",
+    "class": "File"
+  },
+  "normalbam": {
+    "path":"https://dcc.icgc.org/api/v1/download?fn=/PCAWG/reference_data/data_for_testing/HCC1143_ds/HCC1143_BL.bam",
+    "class": "File"
+  },
+  "mutations": {
+    "path": "/tmp/out.vcf",
+    "class": "File"
+  },
+  "tumorbam": {
+    "path":"https://dcc.icgc.org/api/v1/download?fn=/PCAWG/reference_data/data_for_testing/HCC1143_ds/HCC1143.bam",
+    "class": "File"
+  }
+}

SomaticSniper.py

@@ -45,6 +45,15 @@ def sniper_argparser():
     group.add_argument('--sniper-exe', dest='sniper_exe', default='bam-somaticsniper', help='SomaticSniper Exec Name')
     return parser
 
+def gunzip(infile, outfile):
+    cmd = ' '.join(['zcat', infile])
+    with open(outfile, 'w') as outF:
+        p = subprocess.Popen(cmd, shell=True, stdout=outF, stderr=subprocess.PIPE)
+    stdout,stderr =  p.communicate()
+    if len(stderr):
+        print "unzip command failed:", stderr
+        raise Exception("unzip failed")
+
 def tcga_header_arguments():
     return set(('reference_id', 'center',
         'tumor_uuid', 'tumor_barcode', 'tumor_accession', 'tumor_platform',
@@ -87,7 +96,13 @@ def create_sniper_cmdline(namespace_dict, reference, tumor_bam, normal_bam, temp
 
 def create_workspace(workdir, reference, tumor_bam, normal_bam):
 
-    new_ref = symlink_workspace_file(workdir, reference, "ref_genome.fasta")
+    if reference.endswith('.gz'):
+        new_ref = os.path.join(args.workdir, "ref_genome.fasta")
+        gunzip(args.f, new_ref)
+        args.f = new_ref
+    else:
+        new_ref = symlink_workspace_file(workdir, reference, "ref_genome.fasta")
+
     if not os.path.exists(reference + ".fai"):
         print "Indexing", new_ref
         subprocess.check_call( ["/usr/bin/samtools", "faidx", new_ref] )

somatic_sniper.cwl

@@ -0,0 +1,128 @@
+#!/usr/bin/env cwl-runner
+#
+# Author: Jeltje van Baren jeltje.van.baren@gmail.com
+
+cwlVersion: v1.0
+class: CommandLineTool
+baseCommand: [ /opt/SomaticSniper.py, -F, vcf, --workdir, ./ ]
+
+doc: "Runs somatic sniper snp caller on input bam files"
+
+hints:
+  DockerRequirement:
+    dockerPull: quay.io/opengenomics/somatic-sniper
+
+#requirements:
+#  - class: InlineJavascriptRequirement
+
+inputs:
+
+  reference:
+    type: File
+    doc: |
+      reference sequence in the FASTA format
+    inputBinding:
+      position: 3
+      prefix: -f
+
+  tumor_name:
+    type: string?
+    doc: |
+      tumor sample id (for VCF header) [TUMOR]
+    inputBinding:
+      position: 3
+      prefix: -t
+
+  normal_name:
+    type: string?
+    doc: |
+      normal sample id (for VCF header) [NORMAL]
+    inputBinding:
+      position: 3
+      prefix: -n
+
+  minmapqual:
+    type: int?
+    doc: |
+         filtering reads with mapping quality less than [0]
+    inputBinding:
+      position: 3
+      prefix: -q
+
+  snvqual:
+    type: int?
+    doc: |
+         filtering somatic snv output with somatic quality less than  [15]
+    inputBinding:
+      position: 3
+      prefix: -Q
+
+  noLOH:
+    type: boolean?
+    doc: |
+        do not report LOH variants as determined by genotypes
+    inputBinding:
+      position: 3
+      prefix: -L
+
+  noGainOfRef:
+    type: boolean?
+    doc: |
+        do not report Gain of Reference variants as determined by genotypes
+    inputBinding:
+      position: 3
+      prefix: -G
+
+  noSomaticPriors:
+    type: boolean?
+    doc: |
+        disable priors in the somatic calculation. Increases sensitivity for solid tumors
+    inputBinding:
+      position: 3
+      prefix: -p
+
+  doPriors:
+    type: boolean?
+    doc: |
+        Use prior probabilities accounting for the somatic mutation rate
+    inputBinding:
+      position: 3
+      prefix: -J
+
+  priorProb:
+    type: float?
+    doc: |
+       prior probability of a somatic mutation (implies -J) [0.010000]
+    inputBinding:
+      position: 3
+      prefix: -s
+
+  tumorbam:
+    type: File
+    doc: |
+      tumor bamfile
+    inputBinding:
+      position: 1
+
+  normalbam:
+    type: File
+    doc: |
+      normal bamfile
+    inputBinding:
+      position: 2
+
+  output_name:
+    type: string
+    default: mutations.vcf
+    doc: |
+      Name of output file
+    inputBinding:
+      position: 5
+
+outputs:
+
+  mutations:
+    type: File?
+    outputBinding:
+      glob: $(inputs.output_name)
+