Merge pull request #190 from msk-access/MafAnnotationWithBed

Maf annotation with bed

Author: rhshah

variant_annotation/README.md

@@ -6,16 +6,25 @@
 
 - [snpift](https://msk-access.gitbook.io/command-line-tools-cwl/snpsift_annotate_5.0/)
 - [vcf2maf](https://msk-access.gitbook.io/command-line-tools-cwl/vcf2maf_1.6.21/)
+- [maf_annotated_by_bed](https://msk-access.gitbook.io/command-line-tools-cwl/postprocessing_variant_calls/0.2.0/maf_annotated_by_bed/)
 
 #### Usage
 
 ```bash
-usage: variant_annotation.cwl [-h]  --input_cosmicCountDB_vcf INPUT_COSMICCOUNTDB_VCF
+usage: variant_annotation.cwl
+       [-h] --input_cosmicCountDB_vcf INPUT_COSMICCOUNTDB_VCF
        --vardict_input_vcf VARDICT_INPUT_VCF --input_cosmicprevalenceDB_vcf
        INPUT_COSMICPREVALENCEDB_VCF [--min_hom_vaf MIN_HOM_VAF]
-       [--output_mafName OUTPUT_MAFNAME] [--retain_info RETAIN_INFO]
+       [--output_vcf2mafName OUTPUT_VCF2MAFNAME] [--retain_info RETAIN_INFO]
        [--tumor_id TUMOR_ID] [--snpsift_countOpName SNPSIFT_COUNTOPNAME]
-       [--snpsift_prevalOpName SNPSIFT_PREVALOPNAME]
+       [--snpsift_prevalOpName SNPSIFT_PREVALOPNAME] --opOncoKbMafName
+       OPONCOKBMAFNAME --oncoKbApiToken ONCOKBAPITOKEN
+       [--oncoKbAnnotateHotspots] --input_complexity_bed INPUT_COMPLEXITY_BED
+       --input_DUST_bed INPUT_DUST_BED
+       [--output_DUST_filename OUTPUT_DUST_FILENAME]
+       [--column_name_DUST COLUMN_NAME_DUST]
+       [--output_complexity_filename OUTPUT_COMPLEXITY_FILENAME]
+       [--column_name_complexity COLUMN_NAME_COMPLEXITY]
        [job_order]
 
 positional arguments:
@@ -27,10 +36,19 @@ optional arguments:
   --vardict_input_vcf VARDICT_INPUT_VCF
   --input_cosmicprevalenceDB_vcf INPUT_COSMICPREVALENCEDB_VCF
   --min_hom_vaf MIN_HOM_VAF
-  --output_mafName OUTPUT_MAFNAME
+  --output_vcf2mafName OUTPUT_VCF2MAFNAME
   --retain_info RETAIN_INFO
   --tumor_id TUMOR_ID
   --snpsift_countOpName SNPSIFT_COUNTOPNAME
   --snpsift_prevalOpName SNPSIFT_PREVALOPNAME
+  --opOncoKbMafName OPONCOKBMAFNAME
+  --oncoKbApiToken ONCOKBAPITOKEN
+  --oncoKbAnnotateHotspots
+  --input_complexity_bed INPUT_COMPLEXITY_BED
+  --input_DUST_bed INPUT_DUST_BED
+  --output_DUST_filename OUTPUT_DUST_FILENAME
+  --column_name_DUST COLUMN_NAME_DUST
+  --output_complexity_filename OUTPUT_COMPLEXITY_FILENAME
+  --column_name_complexity COLUMN_NAME_COMPLEXITY
 ```
 

variant_annotation/example_inputs.yaml

@@ -11,7 +11,18 @@ input_cosmicprevalenceDB_vcf:
     /work/cch/production/resources/cosmic/versions/v96/CosmicCodingMuts_GRCh37_processed.vcf.gz
 snpsift_countOpName: tumorId_cosmic.vcf
 snpsift_prevalOpName: tumorId_preval.vcf
-output_mafName: tumorId_annotated.maf
+opOncoKbMafName: tumorId_annotated.maf
 retain_info: CNT,TUMOR_TYPE
 tumor_id: tumorId
 min_hom_vaf: 0
+column_name_mappability: mappability
+column_name_complexity: complexity
+input_mappability_bed:
+  class: File
+  path: /path/to/mappability_algo/bedfile.bed
+input_complexity_bed:
+  class: File
+  path: /path/to/complexity/bedfile.bed
+output_mappability_filename: mappability.maf
+output_complexity_filename: complexity.maf
+output_vcf2mafName: vcf2maf.maf

variant_annotation/variant_annotation.cwl

@@ -26,18 +26,18 @@ inputs:
     type: float?
     'sbg:x': 416.7921447753906
     'sbg:y': 654.78125
-  - id: output_mafName
+  - id: output_vcf2mafName
     type: string?
     'sbg:x': 416.7921447753906
     'sbg:y': 441.1875
   - id: retain_info
     type: string?
-    'sbg:x': 416.7921447753906
-    'sbg:y': 227.59375
+    'sbg:x': 407
+    'sbg:y': 277
   - id: tumor_id
     type: string?
-    'sbg:x': 416.7921447753906
-    'sbg:y': 0
+    'sbg:x': 469
+    'sbg:y': -64
   - id: snpsift_countOpName
     type: string?
     'sbg:x': 16.4202823638916
@@ -46,6 +46,42 @@ inputs:
     type: string?
     'sbg:x': 214.9839324951172
     'sbg:y': -51.58765411376953
+  - id: opOncoKbMafName
+    type: string
+    'sbg:x': 953.6817626953125
+    'sbg:y': 129.14283752441406
+  - id: oncoKbApiToken
+    type: File
+    'sbg:x': 934
+    'sbg:y': 402
+  - id: oncoKbAnnotateHotspots
+    type: boolean?
+    'sbg:x': 914.09375
+    'sbg:y': 582.5
+  - id: input_mappability_bed
+    type: File
+    'sbg:x': 1137.1201171875
+    'sbg:y': 407.4906311035156
+  - id: output_mappability_filename
+    type: string?
+    'sbg:x': 1172.9835205078125
+    'sbg:y': 68.39226531982422
+  - id: column_name_mappability
+    type: string?
+    'sbg:x': 1273.205322265625
+    'sbg:y': 514.48193359375
+  - id: input_complexity_bed
+    type: File
+    'sbg:x': 1455.4620361328125
+    'sbg:y': 427.4315490722656
+  - id: column_name_complexity
+    type: string?
+    'sbg:x': 1688.9140625
+    'sbg:y': 449.50762939453125
+  - id: output_complexity_filename
+    type: string?
+    'sbg:x': 1609.2314453125
+    'sbg:y': 21.32272720336914
 outputs:
   - id: cosmicCount_annotatedOutput
     outputSource:
@@ -63,8 +99,20 @@ outputs:
     outputSource:
       - vcf2maf_v1_6_21/vcf2maf_maf
     type: File
-    'sbg:x': 1113.535400390625
-    'sbg:y': 265.8045959472656
+    'sbg:x': 1148.8089599609375
+    'sbg:y': 594.5475463867188
+  - id: output_mappability_maf
+    outputSource:
+      - maf_annotated_by_bed_mappability/output
+    type: File
+    'sbg:x': 1448.455810546875
+    'sbg:y': 54.44126510620117
+  - id: output_complexity_maf
+    outputSource:
+      - maf_annotated_by_bed_lowComplexity/output
+    type: File
+    'sbg:x': 1815.8809814453125
+    'sbg:y': 248.1118621826172
 steps:
   - id: snpsift_annotate_5_0
     in:
@@ -101,7 +149,7 @@ steps:
       - id: min_hom_vaf
         source: min_hom_vaf
       - id: output_maf
-        source: output_mafName
+        source: output_vcf2mafName
       - id: ref_fasta
         default: /.vep/homo_sapiens/105_GRCh37/Homo_sapiens.GRCh37.dna.toplevel.fa.gz
       - id: retain_info
@@ -115,6 +163,58 @@ steps:
     run: ../command_line_tools//vcf2maf_1.6.21/vcf2maf_1.6.21.cwl
     'sbg:x': 833.5098266601562
     'sbg:y': 276.9501953125
+  - id: oncokb_annotator
+    in:
+      - id: inputMafFile
+        source: vcf2maf_v1_6_21/vcf2maf_maf
+      - id: outputMafName
+        source: opOncoKbMafName
+      - id: apiToken
+        source: oncoKbApiToken
+      - id: referenceGenome
+        default: GRCh37
+      - id: annotateHotspots
+        source: oncoKbAnnotateHotspots
+    out:
+      - id: outputMaf
+    run: ../command_line_tools/oncokb_annotator_3.2.2/oncokb_annotator_3-2-2.cwl
+    label: oncokb_annotator
+    'sbg:x': 1096.5919189453125
+    'sbg:y': 261
+  - id: maf_annotated_by_bed_mappability
+    in:
+      - id: input_maf
+        source: oncokb_annotator/outputMaf
+      - id: input_bed
+        source: input_mappability_bed
+      - id: output_filename
+        source: output_mappability_filename
+      - id: column_name
+        source: column_name_mappability
+    out:
+      - id: output
+    run: >-
+      ../command_line_tools/postprocessing_variant_calls/0.2.2/maf_annotated_by_bed/maf_annotated_by_bed.cwl
+    label: maf_annotated_by_bed
+    'sbg:x': 1317.3984375
+    'sbg:y': 267
+  - id: maf_annotated_by_bed_lowComplexity
+    in:
+      - id: input_maf
+        source: maf_annotated_by_bed_mappability/output
+      - id: input_bed
+        source: input_complexity_bed
+      - id: output_filename
+        source: output_complexity_filename
+      - id: column_name
+        source: column_name_complexity
+    out:
+      - id: output
+    run: >-
+      ../command_line_tools/postprocessing_variant_calls/0.2.2/maf_annotated_by_bed/maf_annotated_by_bed.cwl
+    label: maf_annotated_by_bed
+    'sbg:x': 1600
+    'sbg:y': 262.8888854980469
 requirements: []
 $schemas:
   - 'http://schema.org/version/latest/schemaorg-current-http.rdf'