Merge pull request #58 from mskcc/release/0.3.0

Release 0.3.0
mskcc · Jul 31, 2019 · 5f4a0b5 · 5f4a0b5
2 parents fa8351b + a8a3eb5
commit 5f4a0b5
Show file tree

Hide file tree

Showing 34 changed files with 1,259 additions and 98 deletions.
diff --git a/abra2_2.17/README.md b/abra2_2.17/README.md
@@ -18,4 +18,71 @@
 ```bash
     > toil-cwl-runner abra2_2.17.cwl example_inputs.yaml
 ```
-
+
+**If at MSK, using the JUNO cluster having installed toil version 3.19 and manually modifying [lsf.py](https://github.com/DataBiosphere/toil/blob/releases/3.19.0/src/toil/batchSystems/lsf.py#L170) by removing `type==X86_64 &&` you can use the following command**
+
+```bash
+#Using CWLTOOL
+> cwltool --singularity --non-strict /path/to/abra2_2.17.cwl /path/to/inputs.yaml
+
+#Using toil-cwl-runner
+> mkdir abra2_toil_log
+> toil-cwl-runner --singularity --logFile /path/to/abra2_toil_log/cwltoil.log  --jobStore /path/to/abra2_jobStore --batchSystem lsf --workDir /path/to/abra2_toil_log --outdir . --writeLogs /path/to/abra2_toil_log --logLevel DEBUG --stats --retryCount 2 --disableCaching --maxLogFileSize 20000000000 /path/to/abra2_2.17.cwl /path/to/inputs.yaml > abra2_toil.stdout 2> abra2_toil.stderr &
+```
+
+### Usage  
+
+```
+usage: abra2_2.17.cwl [-h]
+
+positional arguments:
+  job_order             Job input json file
+
+optional arguments:
+  -h, --help            show this help message and exit
+  --memory_per_job MEMORY_PER_JOB
+                        Memory per job in megabytes
+  --memory_overhead MEMORY_OVERHEAD
+                        Memory overhead per job in megabytes
+  --number_of_threads NUMBER_OF_THREADS
+  --working_directory WORKING_DIRECTORY
+                        Set the temp directory (overrides java.io.tmpdir)
+  --reference_fasta REFERENCE_FASTA
+                        Genome reference location
+  --targets TARGETS
+  --kmer_size KMER_SIZE
+                        Optional assembly kmer size(delimit with commas if
+                        multiple sizes specified)
+  --maximum_average_depth MAXIMUM_AVERAGE_DEPTH
+                        Regions with average depth exceeding this value will
+                        be downsampled (default: 1000)
+  --soft_clip_contig SOFT_CLIP_CONTIG
+                        Soft clip contig args [max_contigs,min_base_qual,frac_
+                        high_qual_bases,min_soft_clip_len]
+                        (default:16,13,80,15)
+  --maximum_mixmatch_rate MAXIMUM_MIXMATCH_RATE
+                        Max allowed mismatch rate when mapping reads back to
+                        contigs (default: 0.05)
+  --scoring_gap_alignments SCORING_GAP_ALIGNMENTS
+                        Scoring used for contig alignments(match,
+                        mismatch_penalty,gap_open_penalty,gap_extend_penalty)
+                        (default:8,32,48,1)
+  --contig_anchor CONTIG_ANCHOR
+                        Contig anchor
+                        [M_bases_at_contig_edge,max_mismatches_near_edge]
+                        (default:10,2)
+  --window_size WINDOW_SIZE
+                        Processing window size and overlap (size,overlap)
+                        (default: 400,200)
+  --consensus_sequence  Use positional consensus sequence when aligning high
+                        quality soft clipping
+  --ignore_bad_assembly
+                        Use this option to avoid parsing errors for corrupted
+                        assemblies
+  --bam_index           Enable BAM index generation when outputting sorted
+                        alignments (may require additonal memory)
+  --input_vcf INPUT_VCF
+                        VCF containing known (or suspected) variant sites.
+                        Very large files should be avoided.
+  --no_sort             Do not attempt to sort final output
+  ```
diff --git a/abra2_2.17/abra2_2.17.cwl b/abra2_2.17/abra2_2.17.cwl
@@ -170,8 +170,10 @@ arguments:
     valueFrom: /usr/local/bin/abra2.jar
 requirements:
   - class: ResourceRequirement
-    ramMin: "${\r  if(inputs.memory_per_job && inputs.memory_overhead) {\r   \r    return inputs.memory_per_job + inputs.memory_overhead\r  }\r  else if (inputs.memory_per_job && !inputs.memory_overhead){\r    \r   \treturn inputs.memory_per_job + 2000\r  }\r  else if(!inputs.memory_per_job && inputs.memory_overhead){\r    \r    return 15000 + inputs.memory_overhead\r  }\r  else {\r    \r  \treturn 17000 \r  }\r}"
-    coresMin: "${\r  if (inputs.number_of_threads) {\r    \r   \treturn inputs.number_of_threads \r  }\r  else {\r    \r    return 4\r  }\r}"
+    ramMin: 48000
+    coresMin: 4
+    #ramMin: "${\r  if(inputs.memory_per_job && inputs.memory_overhead) {\r   \r    return inputs.memory_per_job + inputs.memory_overhead\r  }\r  else if (inputs.memory_per_job && !inputs.memory_overhead){\r    \r   \treturn inputs.memory_per_job + 2000\r  }\r  else if(!inputs.memory_per_job && inputs.memory_overhead){\r    \r    return 15000 + inputs.memory_overhead\r  }\r  else {\r    \r  \treturn 17000 \r  }\r}"
+    #coresMin: "${\r  if (inputs.number_of_threads) {\r    \r   \treturn inputs.number_of_threads \r  }\r  else {\r    \r    return 4\r  }\r}"
   - class: DockerRequirement
     dockerPull: 'mskcc/abra2:0.1.0'
   - class: InlineJavascriptRequirement
@@ -193,6 +195,3 @@ requirements:
   - class: 'doap:Version'
     'doap:name': abra2
     'doap:revision': 2.17
-  - class: 'doap:Version'
-    'doap:name': cwl-wrapper
-    'doap:revision': 1.0.0
diff --git a/alfred_0.1.17/alfred_0.1.17.cwl b/alfred_0.1.17/alfred_0.1.17.cwl
@@ -0,0 +1,109 @@
+class: CommandLineTool
+cwlVersion: v1.0
+id: alfred_0_1_17
+baseCommand:
+  - alfred
+  - qc
+inputs:
+  - id: reference
+    type: File
+    inputBinding:
+      position: 0
+      prefix: '--reference'
+  - id: ignore_rg
+    type: boolean?
+    inputBinding:
+      position: 0
+      prefix: '--ignore'
+  - id: bed
+    type: File?
+    inputBinding:
+      position: 0
+      prefix: '--bed'
+  - id: bam
+    type: File
+    inputBinding:
+      position: 3
+    secondaryFiles:
+      - .bai
+  - id: read_group
+    type: string?
+    inputBinding:
+      position: 0
+      prefix: '--rg'
+  - id: supplementary
+    type: boolean?
+    inputBinding:
+      position: 0
+      prefix: '--supplementary'
+  - id: secondary
+    type: boolean?
+    inputBinding:
+      position: 0
+      prefix: '--secondary'
+  - id: jsonout
+    type: string?
+    inputBinding:
+      position: 0
+      prefix: '--jsonout'
+  - id: sample_name
+    type: string?
+    inputBinding:
+      position: 0
+      prefix: '--name'
+  - id: outfile
+    type: string?
+outputs:
+  - id: output
+    type: File?
+    outputBinding:
+      glob: '*.pdf'
+  - id: output_pdf
+    type: File?
+    outputBinding:
+      glob: '*.tsv.gz'
+label: alfred_0.1.17
+arguments:
+  - position: 4
+    prefix: ''
+    shellQuote: false
+    valueFrom: '&&'
+  - position: 5
+    prefix: ''
+    shellQuote: false
+    valueFrom: Rscript
+  - position: 6
+    prefix: ''
+    shellQuote: false
+    valueFrom: /opt/alfred/scripts/stats.R
+  - position: 7
+    prefix: ''
+    shellQuote: false
+    valueFrom: |-
+      ${
+          if (inputs.outname) {
+              return inputs.outname + '.tsv.gz';
+          }
+          if (inputs.ignore_rg) {
+              return inputs.bam.basename.replace('.bam', '') + '.alfred.tsv.gz'
+          }
+          return inputs.bam.basename.replace('.bam', '') + 'RG.alfred.tsv.gz'
+      }
+  - position: 2
+    prefix: '--outfile'
+    shellQuote: false
+    valueFrom: |-
+      ${
+          if (inputs.outname) {
+              return inputs.outname + '.tsv.gz';
+          }
+          if (inputs.ignore_rg) {
+              return inputs.bam.basename.replace('.bam', '') + '.alfred.tsv.gz';
+          }
+          return inputs.bam.basename.replace('.bam', '') + 'RG.alfred.tsv.gz';
+      }
+requirements:
+  - class: ShellCommandRequirement
+  - class: DockerRequirement
+    dockerPull: 'cmopipeline/alfred:v0.1.17'
+  - class: InlineJavascriptRequirement
diff --git a/bedtools_genomecov_v2.28.0_cv2/README.md b/bedtools_genomecov_v2.28.0_cv2/README.md
@@ -0,0 +1,56 @@
+# CWL and Dockerfile for running Bedtools GenomeCov
+
+## Version of tools in [docker image ](https://hub.docker.com/r/biocontainers/bedtools)
+
+| Tool	| Version	| Location	|
+|---	|---	|---	|
+| Bedtools  	| v2.28.0_cv2   	|  https://github.com/arq5x/bedtools2/releases/tag/v2.28.0	|
+
+[![](https://img.shields.io/badge/version-2.28.0_cv2-blue)](https://github.com/arq5x/bedtools2/releases/tag/v2.28.0)
+## CWL
+
+- CWL specification 1.0
+- Use example_inputs.yml to see the inputs to the cwl
+- Example Command using [toil](https://toil.readthedocs.io):
+
+```bash
+    > toil-cwl-runner bedtools_genomecov_v2.28.0_cv2.cwl example_inputs.yml
+```
+
+**If at MSK, using the JUNO cluster having installed toil version 3.19 and manually modifying [lsf.py](https://github.com/DataBiosphere/toil/blob/releases/3.19.0/src/toil/batchSystems/lsf.py#L170) by removing `type==X86_64 &&` you can use the following command**
+
+```bash
+#Using CWLTOOL
+> cwltool --singularity --non-strict bedtools_genomecov_v2.28.0_cv2.cwl inputs.yaml
+
+#Using toil-cwl-runner
+> mkdir run_directory
+> toil-cwl-runner --singularity --logFile path/to/run_directory/cwltoil.log  --jobStore path/to/jobStore --batchSystem lsf --workDir /path/to/run_directory --outdir /path/to/run_directory --writeLogs /path/to/run_directory --logLevel DEBUG --stats --retryCount 2 --disableCaching --maxLogFileSize 20000000000 bedtools_genomecov_v2.28.0_cv2.cwl inputs.yaml > file.stdout 2> file.stderr &
+```
+
+## Usage
+
+```bash
+> toil-cwl-runner bedtools_genomecov_v2.28.0_cv2.cwl --help
+
+usage: bedtools_genomecov_v2.28.0_cv2.cwl [-h] --input INPUT
+                                          --output_file_name OUTPUT_FILE_NAME
+                                          [--memory_overhead MEMORY_OVERHEAD]
+                                          [--memory_per_job MEMORY_PER_JOB]
+                                          [--number_of_threads NUMBER_OF_THREADS]
+                                          [--option_bedgraph]
+                                          [job_order]
+
+positional arguments:
+  job_order             Job input json file
+
+optional arguments:
+  -h, --help            show this help message and exit
+  --input INPUT         The input file can be in BAM format (Note: BAM must be
+                        sorted by position)
+  --output_file_name OUTPUT_FILE_NAME
+  --memory_overhead MEMORY_OVERHEAD
+  --memory_per_job MEMORY_PER_JOB
+  --number_of_threads NUMBER_OF_THREADS
+  --option_bedgraph     option flag parameter to choose output file format.
+                        -bg refers to bedgraph format
diff --git a/...omecov_v2.28.0_cv2/bedtools_genomecov.cwl → ....0_cv2/bedtools_genomecov_v2.28.0_cv2.cwl b/...omecov_v2.28.0_cv2/bedtools_genomecov.cwl → ....0_cv2/bedtools_genomecov_v2.28.0_cv2.cwl
@@ -4,6 +4,7 @@ $namespaces:
   dct: 'http://purl.org/dc/terms/'
   doap: 'http://usefulinc.com/ns/doap#'
   foaf: 'http://xmlns.com/foaf/0.1/'
+  edam: http://edamontology.org/
 id: bedtools_genomecov
 baseCommand:
   - bedtools
@@ -51,8 +52,10 @@ label: bedtools_genomecov
 requirements:
   - class: ShellCommandRequirement
   - class: ResourceRequirement
-    ramMin: "${\r  if(inputs.memory_per_job && inputs.memory_overhead) {\r   \r    return inputs.memory_per_job + inputs.memory_overhead\r  }\r  else if (inputs.memory_per_job && !inputs.memory_overhead){\r    \r   \treturn inputs.memory_per_job + 2000\r  }\r  else if(!inputs.memory_per_job && inputs.memory_overhead){\r    \r    return 8000 + inputs.memory_overhead\r  }\r  else {\r    \r  \treturn 8000 \r  }\r}"
-    coresMin: "${\r  if (inputs.number_of_threads) {\r    \r   \treturn inputs.number_of_threads \r  }\r  else {\r    \r    return 1\r  }\r}"
+    ramMin: 20000
+    coresMin: 1
+    #ramMin: "${\r  if(inputs.memory_per_job && inputs.memory_overhead) {\r   \r    return inputs.memory_per_job + inputs.memory_overhead\r  }\r  else if (inputs.memory_per_job && !inputs.memory_overhead){\r    \r   \treturn inputs.memory_per_job + 2000\r  }\r  else if(!inputs.memory_per_job && inputs.memory_overhead){\r    \r    return 8000 + inputs.memory_overhead\r  }\r  else {\r    \r  \treturn 8000 \r  }\r}"
+    #coresMin: "${\r  if (inputs.number_of_threads) {\r    \r   \treturn inputs.number_of_threads \r  }\r  else {\r    \r    return 1\r  }\r}"
   - class: DockerRequirement
     dockerPull: 'biocontainers/bedtools:v2.28.0_cv2'
   - class: InlineJavascriptRequirement
@@ -80,6 +83,3 @@ stdout: |-
   - class: 'doap:Version'
     'doap:name': bedtools
     'doap:revision': v2.28.0_cv2
-  - class: 'doap:Version'
-    'doap:name': cwl-wrapper
-    'doap:revision': 1.0.0
diff --git a/bedtools_merge_v2.28.0_cv2/README.md b/bedtools_merge_v2.28.0_cv2/README.md
@@ -0,0 +1,57 @@
+# CWL and Dockerfile for running Bedtools Merge
+
+## Version of tools in [docker image ](https://hub.docker.com/r/biocontainers/bedtools)
+
+| Tool	| Version	| Location	|
+|---	|---	|---	|
+| Bedtools  	| v2.28.0_cv2   	|  https://github.com/arq5x/bedtools2/releases/tag/v2.28.0	|
+
+[![](https://img.shields.io/badge/version-2.28.0_cv2-blue)](https://github.com/arq5x/bedtools2/releases/tag/v2.28.0)
+## CWL
+
+- CWL specification 1.0
+- Use example_inputs.yml to see the inputs to the cwl
+- Example Command using [toil](https://toil.readthedocs.io):
+
+```bash
+    > toil-cwl-runner bedtools_merge_v2.28.0_cv2.cwl example_inputs.yml
+```
+
+**If at MSK, using the JUNO cluster having installed toil version 3.19 and manually modifying [lsf.py](https://github.com/DataBiosphere/toil/blob/releases/3.19.0/src/toil/batchSystems/lsf.py#L170) by removing `type==X86_64 &&` you can use the following command**
+
+```bash
+#Using CWLTOOL
+> cwltool --singularity --non-strict bedtools_merge_v2.28.0_cv2.cwl inputs.yaml
+
+#Using toil-cwl-runner
+> mkdir run_directory
+> toil-cwl-runner --singularity --logFile path/to/run_directory/cwltoil.log  --jobStore path/to/jobStore --batchSystem lsf --workDir /path/to/run_directory --outdir /path/to/run_directory --writeLogs /path/to/run_directory --logLevel DEBUG --stats --retryCount 2 --disableCaching --maxLogFileSize 20000000000 bedtools_merge_v2.28.0_cv2.cwl inputs.yaml > file.stdout 2> file.stderr &
+```
+
+## Usage
+
+```bash
+> toil-cwl-runner bedtools_merge_v2.28.0_cv2.cwl --help
+
+usage: bedtools_merge_v2.28.0_cv2.cwl [-h] --input INPUT --output_file_name
+                                      OUTPUT_FILE_NAME
+                                      [--memory_overhead MEMORY_OVERHEAD]
+                                      [--memory_per_job MEMORY_PER_JOB]
+                                      [--number_of_threads NUMBER_OF_THREADS]
+                                      [--distance_between_features DISTANCE_BETWEEN_FEATURES]
+                                      [job_order]
+
+positional arguments:
+  job_order             Job input json file
+
+optional arguments:
+  -h, --help            show this help message and exit
+  --input INPUT         BEDgraph format file generated from Bedtools Genomecov
+                        module
+  --output_file_name OUTPUT_FILE_NAME
+  --memory_overhead MEMORY_OVERHEAD
+  --memory_per_job MEMORY_PER_JOB
+  --number_of_threads NUMBER_OF_THREADS
+  --distance_between_features DISTANCE_BETWEEN_FEATURES
+                        Maximum distance between features allowed for features
+                        to be merged.
diff --git a/...ools_merge_v2.28.0_cv2/bedtools_merge.cwl → ...2.28.0_cv2/bedtools_merge_v2.28.0_cv2.cwl b/...ools_merge_v2.28.0_cv2/bedtools_merge.cwl → ...2.28.0_cv2/bedtools_merge_v2.28.0_cv2.cwl
@@ -46,8 +46,10 @@ label: bedtools_merge
 requirements:
   - class: ShellCommandRequirement
   - class: ResourceRequirement
-    ramMin: "${\r  if(inputs.memory_per_job && inputs.memory_overhead) {\r   \r    return inputs.memory_per_job + inputs.memory_overhead\r  }\r  else if (inputs.memory_per_job && !inputs.memory_overhead){\r    \r   \treturn inputs.memory_per_job + 2000\r  }\r  else if(!inputs.memory_per_job && inputs.memory_overhead){\r    \r    return 8000 + inputs.memory_overhead\r  }\r  else {\r    \r  \treturn 8000 \r  }\r}"
-    coresMin: "${\r  if (inputs.number_of_threads) {\r    \r   \treturn inputs.number_of_threads \r  }\r  else {\r    \r    return 1\r  }\r}"
+    ramMin: 20000
+    coresMin: 1
+#     ramMin: "${\r  if(inputs.memory_per_job && inputs.memory_overhead) {\r   \r    return inputs.memory_per_job + inputs.memory_overhead\r  }\r  else if (inputs.memory_per_job && !inputs.memory_overhead){\r    \r   \treturn inputs.memory_per_job + 2000\r  }\r  else if(!inputs.memory_per_job && inputs.memory_overhead){\r    \r    return 8000 + inputs.memory_overhead\r  }\r  else {\r    \r  \treturn 8000 \r  }\r}"
+#     coresMin: "${\r  if (inputs.number_of_threads) {\r    \r   \treturn inputs.number_of_threads \r  }\r  else {\r    \r    return 1\r  }\r}"
   - class: DockerRequirement
     dockerPull: 'biocontainers/bedtools:v2.28.0_cv2'
   - class: InlineJavascriptRequirement
@@ -75,6 +77,3 @@ stdout: |-
   - class: 'doap:Version'
     'doap:name': bedtools
     'doap:revision': v2.28.0_cv2
-  - class: 'doap:Version'
-    'doap:name': cwl-wrapper
-    'doap:revision': 1.0.0