Merge pull request #4 from madetunj/frommain

dx-cwl from the main seaseq
madetunj · Jun 24, 2020 · 45fffdb · 45fffdb
2 parents 3ee6a5c + 8d093dc
commit 45fffdb
Show file tree

Hide file tree

Showing 54 changed files with 331 additions and 1,376 deletions.
diff --git a/LSFconfig.json b/LSFconfig.json
diff --git a/README.md b/README.md
@@ -1,5 +1,7 @@
 # (S)ingle (E)nd (A)ntibody (SEQ)uencing pipeline
 
+### DX-CWL COMPATIBLE
+
 Chromatin Single-End analysis pipeline
 
 The SEASEQ Pipeline is a complete analysis pipeline for CHiP 
@@ -20,14 +22,16 @@ and about 30GB of supplemental data.
 
 ## PROGRAMS & VERSIONS
 
+Programs and versions used to build and test the pipeline.
+
 * bowtie v. 1.2.2
 * fastqc v. 0.11.5
 * samtools v. 1.9
 * R v. 3.6.1
 * macs v. 041014
-* SICER2 v. 1.0.1
+* SICER2 v. 1.0.2
 * meme v. 4.11.2
-* phantompeakqualtools v. 1.2.1.1
+* spp v. 1.16.0
 * bedtools/2.25.0
 * python v. 3.7.0
 * java v. 1.8.0_60
@@ -39,18 +43,22 @@ and about 30GB of supplemental data.
 * BAM2GFF v. 1.1.0
 
 
-## REQUIREMENTS
+## REQUIREMENT
+
+* INPUT
 
-INPUT YML
+**inputyml.yml** file 
+
+Example of an **inputyml.yml** file.
 
 ```
 reference: 
   class: Directory
   location: /path/to/genome_reference+index
 
 fastqfile: 
-  - { class: File, path: /path/to/fastqfile1 }
-  - { class: File, path: /path/to/fastqfile2 }
+  class: File
+  path: /path/to/fastqfile
 
 keep_dup: all
 
@@ -71,10 +79,11 @@ motifdatabases:
   - { class: File, path: /path/to/meme_motif2 }
 ```
 
+* OUTPUTS
 
-## EXAMPLE
+All outputfiles will be saved in current working directory
 
-We provided example instructions for running under [Toil]
-(https://toil.readthedocs.io/en/latest/) on our St. Jude HPC LSF cluster.
 
+## EXAMPLE SYNTAX using CWLTOOL
 
+cwltool /path/to/folder/cwl/seaseq_pipeline.cwl inputyml.yml
diff --git a/cwl/ame.cwl b/cwl/ame.cwl
@@ -2,14 +2,15 @@
 cwlVersion: v1.0
 baseCommand: ame
 class: CommandLineTool
+label: AME - Analysis of Motif Enrichment
+doc: |
+  ame <convert fasta> <motif-databases>
+
 
 hints:
   DockerRequirement:
-    dockerPull: madetunj/memesuite:v5.1.1 
+    dockerPull: madetunj/memesuite:v5.1.1
 
-label: AME - Analysis of Motif Enrichment
-doc: |
-  ame <convert fasta> <motif-databases>
 
 requirements:
 - class: InlineJavascriptRequirement
@@ -18,6 +19,7 @@ requirements:
       return 'bklist'+inputs.convertfasta.nameroot.split('bklist').slice(-1)+'-ame_out';
    };
 
+
 inputs:
   convertfasta:
     label: "BED converted FASTA file"
@@ -48,8 +50,7 @@ inputs:
             }
         }
     default: ""
-
-
+
 
 outputs:
   outDir:

diff --git a/cwl/bamtobed.cwl b/cwl/bamtobed.cwl
@@ -2,26 +2,27 @@
 cwlVersion: v1.0
 baseCommand: [bamToBed]
 class: CommandLineTool
+label: convert bam to bed
+doc: |
+  bamToBed -i <bam file> > <bed file>
+
 
 hints:
   DockerRequirement:
-    dockerPull: madetunj/bedtools:v2.25.0
+    dockerPull: madetunj/bedtools:v2.25.0 
 
-label: convert bam to bed
-doc: |
-  bamToBed -i <bam file> > <bed file>
 
 requirements:
 - class: ShellCommandRequirement
 - class: InlineJavascriptRequirement
-
   expressionLib:
   - var var_output_name = function() {
       if (inputs.infile != null) {
          return inputs.infile.nameroot.split('.bam')[0]+'.bam2bed.bed';
       }
    };
 
+
 inputs:
   infile:
     label: "BAM file"
@@ -35,6 +36,7 @@ inputs:
     label: "Output BED file name"
     default: ""
 
+
 stdout: |
   ${
     if (inputs.outputfile == "") {
@@ -44,6 +46,7 @@ stdout: |
     }
   }
 
+
 outputs:
   outfile:
     type: stdout

diff --git a/cwl/bamtogff-scatter.cwl b/cwl/bamtogff-scatter.cwl
@@ -3,13 +3,14 @@ cwlVersion: v1.0
 baseCommand: [BAM2GFF-local.sh]
 class: CommandLineTool
 label: BAM to GFF for MetaGenes calculation v1 on bam file for all metagenes
+doc: |
+  BAM2GFF_call.sh <gtf file> <feature type> <bam file> <chromsizes file> <samplename>
+
 
 hints:
   DockerRequirement:
-    dockerPull: madetunj/bam2gff:v1.1.0 
+    dockerPull: madetunj/bam2gff:v1.1.0
 
-doc: |
-  BAM2GFF_call.sh <gtf file> <feature type> <bam file> <chromsizes file> <samplename>
 
 requirements:
 - class: ShellCommandRequirement
@@ -21,6 +22,7 @@ requirements:
       }
     };
 
+
 inputs:
   bamfile:
     label: "BAM file"
@@ -83,6 +85,7 @@ inputs:
       prefix: '" && mkdir -p $nameoffolder && mv matrix *png *pdf $nameoffolder' 
     default: true
 
+
 outputs:
   metagenesDir:
     type: Directory

diff --git a/cwl/bamtogff.cwl b/cwl/bamtogff.cwl
@@ -3,13 +3,14 @@ cwlVersion: v1.0
 baseCommand: [BAM2GFF-local.sh]
 class: CommandLineTool
 label: BAM to GFF for MetaGenes calculation v1 on bam file for all metagenes
+doc: |
+  BAM2GFF_call.sh <gtf file> <feature type> <bam file> <chromsizes file> <samplename>
+
 
 hints:
   DockerRequirement:
-    dockerPull: madetunj/bam2gff:v1.1.0 
+    dockerPull: madetunj/bam2gff:v1.1.0
 
-doc: |
-  BAM2GFF_call.sh <gtf file> <feature type> <bam file> <chromsizes file> <samplename>
 
 requirements:
 - class: ShellCommandRequirement
@@ -21,6 +22,7 @@ requirements:
       }
     };
 
+
 inputs:
   bamfile:
     label: "BAM file"
@@ -83,6 +85,7 @@ inputs:
       prefix: '" && mkdir -p $nameoffolder && mv matrix *png *pdf $nameoffolder' 
     default: true
 
+
 outputs:
   metagenesDir:
     type: Directory

diff --git a/cwl/basicfastqstats.cwl b/cwl/basicfastqstats.cwl
@@ -3,20 +3,22 @@ cwlVersion: v1.0
 baseCommand: [basicfastqstats.sh]
 class: CommandLineTool
 
+
 hints:
   DockerRequirement:
     dockerPull: madetunj/seaseq:v0.0.1
 
+
 requirements:
 - class: InlineJavascriptRequirement
-
   expressionLib:
   - var var_output_name = function() {
       if (inputs.fastqfile != null) {
          return inputs.fastqfile.nameroot.split('.fastq')[0]+'-fastq.metrics.txt';
       }
    };
 
+
 inputs:
   fastqfile:
     type: File

diff --git a/cwl/bedfasta.cwl b/cwl/bedfasta.cwl
@@ -26,22 +26,6 @@ requirements:
   listing: [ $(inputs.ref_fasta), $(inputs.ref_fasta_index) ]
 
 inputs:
-  reference:
-    label: "Genome reference directory"
-    type: Directory?
-    inputBinding:
-      prefix: -fi
-      position: 1
-      valueFrom: |
-        ${
-            for (var i = 0; i < self.listing.length; i++) {
-                if (self.listing[i].path.split('.').slice(-1) == 'fa') {
-                  return self.listing[i].path;
-                }
-            }
-            return null;
-        }
-
   ref_fasta:
     type: File?
     inputBinding:

diff --git a/cwl/blacklist.cwl b/cwl/blacklist.cwl
@@ -2,26 +2,27 @@
 cwlVersion: v1.0
 baseCommand: [intersectBed, -v]
 class: CommandLineTool
+label: bedtools intersect, to remove blacklist
+doc: |
+  intersectBed -v -a KOPTK1_DMSO.rmdup.bam -b ~/.genomes/hg19/hg19-blacklist.v2.bed > ooo
+
 
 hints:
   DockerRequirement:
     dockerPull: madetunj/bedtools:v2.25.0
 
-label: bedtools intersect, to remove blacklist
-doc: |
-  intersectBed -v -a KOPTK1_DMSO.rmdup.bam -b ~/.genomes/hg19/hg19-blacklist.v2.bed > ooo
 
 requirements:
 - class: ShellCommandRequirement
 - class: InlineJavascriptRequirement
-
   expressionLib:
   - var var_output_name = function() {
       if (inputs.infile != null) {
          return inputs.infile.nameroot.split('.bam')[0]+'.bklist.bam';
       }
    };
 
+
 inputs:
   infile:
     type: File
@@ -43,6 +44,7 @@ inputs:
     label: "Output file name"
     default: ""
 
+
 stdout: |
   ${
     if (inputs.outputfile == "") {
@@ -52,6 +54,7 @@ stdout: |
     }
   }
 
+
 outputs:
   outfile:
     type: stdout

diff --git a/cwl/bowtie.cwl b/cwl/bowtie.cwl
@@ -99,23 +99,6 @@ inputs:
       prefix: --best
       position: 5
 
-  reference:
-    type: Directory?
-    label: "Genome reference directory"
-    inputBinding:
-      position: 6
-      valueFrom: |
-        ${
-            for (var i = 0; i < self.listing.length; i++) {
-                if (self.listing[i].path.split('.').slice(-3).join('.') == 'rev.1.ebwt') {
-                  return self.listing[i].path.split('.').slice(0,-3).join('.');
-                }
-            }
-            return null;
-        }
-    doc: |
-      Folder with Bowtie indices
-
   bowtieindex_1:
     type: File?