gVCF joint: update with latest workflow fixes bcbio/bcbio-nextgen#2473

bcbio · Nov 26, 2018 · 8146478 · 8146478
1 parent 279f053
commit 8146478
Show file tree

Hide file tree

Showing 27 changed files with 537 additions and 276 deletions.
diff --git a/gvcf_joint/gvcf-joint-workflow/main-gvcf-joint-samples.json b/gvcf_joint/gvcf-joint-workflow/main-gvcf-joint-samples.json
@@ -10,8 +10,8 @@
         []
     ],
     "config__algorithm__align_split_size": [
-        "False",
-        "False"
+        null,
+        null
     ],
     "config__algorithm__aligner": [
         "bwa",
@@ -33,8 +33,8 @@
         null
     ],
     "config__algorithm__effects": [
-        "False",
-        "False"
+        "snpeff",
+        "snpeff"
     ],
     "config__algorithm__ensemble": [
         null,
@@ -49,8 +49,8 @@
         "True"
     ],
     "config__algorithm__min_allele_fraction": [
-        10,
-        10
+        10.0,
+        10.0
     ],
     "config__algorithm__nomap_split_size": [
         250,
@@ -87,12 +87,8 @@
         "True"
     ],
     "config__algorithm__tools_off": [
-        [
-            "gemini"
-        ],
-        [
-            "gemini"
-        ]
+        [],
+        []
     ],
     "config__algorithm__tools_on": [
         [
@@ -348,6 +344,28 @@
             ]
         }
     ],
+    "genome_resources__variation__gnomad_exome": [
+        {
+            "class": "File",
+            "path": "/home/chapmanb/drive/work/cwl/test_bcbio_cwl/testdata/genomes/hg19/variation/gnomad_exome.vcf.gz",
+            "secondaryFiles": [
+                {
+                    "class": "File",
+                    "path": "/home/chapmanb/drive/work/cwl/test_bcbio_cwl/testdata/genomes/hg19/variation/gnomad_exome.vcf.gz.tbi"
+                }
+            ]
+        },
+        {
+            "class": "File",
+            "path": "/home/chapmanb/drive/work/cwl/test_bcbio_cwl/testdata/genomes/hg19/variation/gnomad_exome.vcf.gz",
+            "secondaryFiles": [
+                {
+                    "class": "File",
+                    "path": "/home/chapmanb/drive/work/cwl/test_bcbio_cwl/testdata/genomes/hg19/variation/gnomad_exome.vcf.gz.tbi"
+                }
+            ]
+        }
+    ],
     "genome_resources__variation__lcr": [
         null,
         null
@@ -550,14 +568,14 @@
             "path": "/home/chapmanb/drive/work/cwl/test_bcbio_cwl/testdata/genomes/hg19/snpeff--hg19-wf.tar.gz"
         }
     ],
-    "reference__twobit": [
+    "reference__versions": [
         {
             "class": "File",
-            "path": "/home/chapmanb/drive/work/cwl/test_bcbio_cwl/testdata/genomes/hg19/ucsc/hg19.2bit"
+            "path": "/home/chapmanb/drive/work/cwl/test_bcbio_cwl/testdata/genomes/hg19/versions.csv"
         },
         {
             "class": "File",
-            "path": "/home/chapmanb/drive/work/cwl/test_bcbio_cwl/testdata/genomes/hg19/ucsc/hg19.2bit"
+            "path": "/home/chapmanb/drive/work/cwl/test_bcbio_cwl/testdata/genomes/hg19/versions.csv"
         }
     ],
     "resources": [

diff --git a/gvcf_joint/gvcf-joint-workflow/main-gvcf-joint.cwl b/gvcf_joint/gvcf-joint-workflow/main-gvcf-joint.cwl
@@ -5,9 +5,8 @@ inputs:
 - id: config__algorithm__align_split_size
   type:
     items:
-    - string
     - 'null'
-    - boolean
+    - string
     type: array
 - id: files
   type:
@@ -74,12 +73,6 @@ inputs:
   type:
     items: File
     type: array
-- id: genome_resources__variation__train_hapmap
-  secondaryFiles:
-  - .tbi
-  type:
-    items: File
-    type: array
 - id: rgnames__lb
   type:
     items:
@@ -118,12 +111,16 @@ inputs:
     type: array
 - id: config__algorithm__min_allele_fraction
   type:
-    items: long
+    items: double
     type: array
 - id: config__algorithm__nomap_split_targets
   type:
     items: long
     type: array
+- id: reference__versions
+  type:
+    items: File
+    type: array
 - id: reference__bwa__indexes
   secondaryFiles:
   - ^.ann
@@ -139,7 +136,9 @@ inputs:
     - 'null'
     - string
     type: array
-- id: reference__twobit
+- id: genome_resources__variation__train_hapmap
+  secondaryFiles:
+  - .tbi
   type:
     items: File
     type: array
@@ -204,6 +203,12 @@ inputs:
   type:
     items: File
     type: array
+- id: genome_resources__variation__gnomad_exome
+  secondaryFiles:
+  - .tbi
+  type:
+    items: File
+    type: array
 - id: config__algorithm__recalibrate
   type:
     items:
@@ -227,7 +232,8 @@ inputs:
 - id: config__algorithm__tools_off
   type:
     items:
-      items: string
+    - 'null'
+    - items: 'null'
       type: array
     type: array
 - id: genome_resources__variation__dbsnp
@@ -289,10 +295,7 @@ inputs:
     type: array
 - id: config__algorithm__effects
   type:
-    items:
-    - string
-    - 'null'
-    - boolean
+    items: string
     type: array
 - id: config__algorithm__variant_regions
   type:
@@ -376,6 +379,20 @@ outputs:
     - File
     - 'null'
     type: array
+- id: versions__tools
+  outputSource: multiqc_summary/versions__tools
+  type:
+    items:
+    - File
+    - 'null'
+    type: array
+- id: versions__data
+  outputSource: multiqc_summary/versions__data
+  type:
+    items:
+    - File
+    - 'null'
+    type: array
 requirements:
 - class: EnvVarRequirement
   envDef:
@@ -416,6 +433,8 @@ steps:
     source: config__algorithm__adapters
   - id: config__algorithm__bam_clean
     source: config__algorithm__bam_clean
+  - id: config__algorithm__variant_regions
+    source: config__algorithm__variant_regions
   - id: config__algorithm__mark_duplicates
     source: config__algorithm__mark_duplicates
   - id: resources
@@ -431,9 +450,9 @@ steps:
     source: alignment_to_rec/alignment_rec
   out:
   - id: align_bam
-  - id: hla__fastq
   - id: work_bam_plus__disc
   - id: work_bam_plus__sr
+  - id: hla__fastq
   run: wf-alignment.cwl
   scatter:
   - alignment_rec
@@ -508,8 +527,6 @@ steps:
     source: genome_resources__variation__polyx
   - id: genome_resources__variation__encode_blacklist
     source: genome_resources__variation__encode_blacklist
-  - id: reference__twobit
-    source: reference__twobit
   - id: reference__fasta__base
     source: reference__fasta__base
   - id: resources
@@ -618,8 +635,6 @@ steps:
     source: config__algorithm__tools_off
   - id: reference__fasta__base
     source: reference__fasta__base
-  - id: reference__twobit
-    source: reference__twobit
   - id: reference__rtg
     source: reference__rtg
   - id: reference__genome_context
@@ -634,6 +649,8 @@ steps:
     source: genome_resources__variation__esp
   - id: genome_resources__variation__exac
     source: genome_resources__variation__exac
+  - id: genome_resources__variation__gnomad_exome
+    source: genome_resources__variation__gnomad_exome
   - id: genome_resources__variation__1000g
     source: genome_resources__variation__1000g
   - id: genome_resources__variation__lcr
@@ -708,6 +725,8 @@ steps:
     source: analysis
   - id: reference__fasta__base
     source: reference__fasta__base
+  - id: reference__versions
+    source: reference__versions
   - id: config__algorithm__tools_on
     source: config__algorithm__tools_on
   - id: config__algorithm__tools_off
@@ -718,6 +737,8 @@ steps:
     source: config__algorithm__qc
   - id: metadata__batch
     source: metadata__batch
+  - id: metadata__phenotype
+    source: metadata__phenotype
   - id: config__algorithm__coverage_interval
     source: postprocess_alignment/config__algorithm__coverage_interval
   - id: depth__variant_regions__regions
@@ -771,4 +792,6 @@ steps:
     source: pipeline_summary/qcout_rec
   out:
   - id: summary__multiqc
+  - id: versions__tools
+  - id: versions__data
   run: steps/multiqc_summary.cwl
diff --git a/gvcf_joint/gvcf-joint-workflow/steps/alignment_to_rec.cwl b/gvcf_joint/gvcf-joint-workflow/steps/alignment_to_rec.cwl
@@ -4,8 +4,8 @@ arguments:
 - position: 0
   valueFrom: sentinel_runtime=cores,$(runtime['cores']),ram,$(runtime['ram'])
 - sentinel_parallel=multi-combined
-- sentinel_outputs=alignment_rec:resources;description;config__algorithm__align_split_size;files;config__algorithm__trim_reads;reference__fasta__base;config__algorithm__adapters;rgnames__lb;rgnames__rg;rgnames__lane;reference__bwa__indexes;config__algorithm__bam_clean;config__algorithm__aligner;rgnames__pl;rgnames__pu;config__algorithm__mark_duplicates;analysis;rgnames__sample
-- sentinel_inputs=files:var,analysis:var,config__algorithm__align_split_size:var,reference__fasta__base:var,rgnames__pl:var,rgnames__sample:var,rgnames__pu:var,rgnames__lane:var,rgnames__rg:var,rgnames__lb:var,reference__bwa__indexes:var,config__algorithm__aligner:var,config__algorithm__trim_reads:var,config__algorithm__adapters:var,config__algorithm__bam_clean:var,config__algorithm__mark_duplicates:var,resources:var,description:var
+- sentinel_outputs=alignment_rec:resources;description;config__algorithm__align_split_size;files;config__algorithm__trim_reads;reference__fasta__base;config__algorithm__adapters;rgnames__lb;rgnames__rg;rgnames__lane;reference__bwa__indexes;config__algorithm__bam_clean;config__algorithm__aligner;rgnames__pl;rgnames__pu;config__algorithm__mark_duplicates;analysis;rgnames__sample;config__algorithm__variant_regions
+- sentinel_inputs=files:var,analysis:var,config__algorithm__align_split_size:var,reference__fasta__base:var,rgnames__pl:var,rgnames__sample:var,rgnames__pu:var,rgnames__lane:var,rgnames__rg:var,rgnames__lb:var,reference__bwa__indexes:var,config__algorithm__aligner:var,config__algorithm__trim_reads:var,config__algorithm__adapters:var,config__algorithm__bam_clean:var,config__algorithm__variant_regions:var,config__algorithm__mark_duplicates:var,resources:var,description:var
 - run_number=0
 baseCommand:
 - bcbio_nextgen.py
@@ -39,9 +39,8 @@ inputs:
 - id: config__algorithm__align_split_size
   type:
     items:
-    - string
     - 'null'
-    - boolean
+    - string
     type: array
 - id: reference__fasta__base
   secondaryFiles:
@@ -112,6 +111,10 @@ inputs:
     - 'null'
     - boolean
     type: array
+- id: config__algorithm__variant_regions
+  type:
+    items: File
+    type: array
 - id: config__algorithm__mark_duplicates
   type:
     items:
@@ -138,9 +141,8 @@ outputs:
         type: string
       - name: config__algorithm__align_split_size
         type:
-        - string
         - 'null'
-        - boolean
+        - string
       - name: files
         type:
           items: File
@@ -189,6 +191,8 @@ outputs:
         type: string
       - name: rgnames__sample
         type: string
+      - name: config__algorithm__variant_regions
+        type: File
       name: alignment_rec
       type: record
     type: array