Merge pull request #74 from SciDAP/master

Test files & sccript for bam-genomecov-bigwig.cwl workflow

test/STAR-alignReads-job.json

@@ -1,6 +1,6 @@
 {
   "readFilesIn": [{"class": "File", "path": "test-files/SRR1031972.fastq"}],
-  "genomeDir": {"class": "File", "path": "test-files/dm3/"},
+  "genomeDir": {"class": "File", "path": "test-files/dm3/Genome"},
   "outFileNamePrefix":"./test-files/SRR1031972.",
   "outSAMunmapped1": "Within",
   "runThreadN": 1

test/STAR-test.yaml

@@ -84,7 +84,7 @@
 
 - args:
     ["STAR",
-    "--genomeDir",  ./test-files/dm3/,
+    "--genomeDir",  ./test-files/dm3,
     "--outBAMcompression", "10",
     "--outFileNamePrefix", "./test-files/SRR1031972.",
     "--outSAMmode", "Full",

test/test-files/prepare_test.sh

@@ -0,0 +1,22 @@
+#!/usr/bin/env bash
+
+# Script has to be run in current directory ./test/test-files
+
+#download test files
+wget -q -O - ftp://hgdownload.cse.ucsc.edu/goldenPath/dm3/chromosomes/chr4.fa.gz|gunzip >dm3_chr4.fa
+wget -q -O - https://scidap.com/tests/dm3.gtf.bz2 | bunzip2 > dm3_chr4.gtf
+wget -q -O - https://scidap.com/tests/SRR1031972.fastq.bz2 |bunzip2 > SRR1031972.fastq
+mkdir -p dm3
+
+#generate required files. Reference genome. Align reads. Create index.
+cd ..
+cwltool ../tools/STAR.cwl ./STAR-genomeGenerate-job.json 
+cwltool ../tools/STAR.cwl ./STAR-alignReads-job.json
+cwltool ../tools/samtools-index.cwl ./samtools-index-job.json 
+mv SRR1031972.Aligned.sortedByCoord.out.bam.bai ./test-files/
+cd test-files
+
+#link the job file into current directory. and run the pipeline
+ln -s ../../workflows/scidap/bam-genomecov-bigwig-job.json  ./bam-genomecov-bigwig-job.json 
+cwltool --tmpdir-prefix $(pwd) --tmp-outdir-prefix $(pwd) --debug ../../workflows/scidap/bam-genomecov-bigwig.cwl ./bam-genomecov-bigwig-job.json 
+rm ./bam-genomecov-bigwig-job.json

tools/STAR.cwl

@@ -84,7 +84,33 @@ inputs:
     description: |
       string: path to the directory where genome files are stored (if
       runMode!=generateGenome) or will be generated (if runMode==generateGenome)
+    secondaryFiles: |
+      ${
+        var p=inputs.genomeDir.path.split('/').slice(0,-1).join('/');
+        return [
+          {"path": p+"/SA", "class":"File"},
+          {"path": p+"/SAindex", "class":"File"},
+          {"path": p+"/chrNameLength.txt", "class":"File"},
+          {"path": p+"/chrLength.txt", "class":"File"},
+          {"path": p+"/chrStart.txt", "class":"File"},
+          {"path": p+"/geneInfo.tab", "class":"File"},
+          {"path": p+"/sjdbList.fromGTF.out.tab", "class":"File"},
+          {"path": p+"/chrName.txt", "class":"File"},
+          {"path": p+"/exonGeTrInfo.tab", "class":"File"},
+          {"path": p+"/genomeParameters.txt", "class":"File"},
+          {"path": p+"/sjdbList.out.tab", "class":"File"},
+          {"path": p+"/exonInfo.tab", "class":"File"},
+          {"path": p+"/sjdbInfo.txt", "class":"File"},
+          {"path": p+"/transcriptInfo.tab", "class":"File"}
+        ];
+      }
     inputBinding:
+      valueFrom: |
+        ${
+              if (inputs.runMode != "genomeGenerate")
+                return inputs.genomeDir.path.split('/').slice(0,-1).join('/');
+              return inputs.genomeDir;
+        }
       position: 1
       prefix: '--genomeDir'
 

workflows/scidap/bam-genomecov-bigwig-job.json

@@ -1,13 +1,13 @@
 {
     "input": {
         "class": "File",
-        "path": "./test-files/rna.SRR948778.bam"
+        "path": "./SRR1031972.Aligned.sortedByCoord.out.bam"
     },
     "genomeFile": {
         "class": "File",
-        "path": "./test-files/mm10-chrNameLength.txt"
+        "path": "./dm3/chrNameLength.txt"
     },
     "scale": 1,
     "dept":"-bg",
-    "bigWig": "./rna.SRR948778.bigwig"
+    "bigWig": "./SRR1031972.bigwig"
 }