Merge pull request #1048 from johnegarza/rnaseq_sequence_type_input

Rnaseq sequence type input

Author: johnegarza

definitions/pipelines/immuno.cwl

@@ -17,8 +17,8 @@ inputs:
         secondaryFiles: [".1.ht2", ".2.ht2", ".3.ht2", ".4.ht2", ".5.ht2", ".6.ht2", ".7.ht2", ".8.ht2"]
     reference_annotation:
         type: File
-    rna_bams:
-        type: File[]
+    rna_sequence:
+        type: ../types/sequence_data.yml#sequence_data[]
     rna_readgroups:
         type: string[]
     read_group_fields:
@@ -788,7 +788,7 @@ steps:
             reference: reference
             reference_index: reference_index
             reference_annotation: reference_annotation
-            instrument_data_bams: rna_bams
+            rna_sequence: rna_sequence
             read_group_id: rna_readgroups
             read_group_fields: read_group_fields
             sample_name: sample_name

definitions/pipelines/rnaseq.cwl

@@ -8,6 +8,9 @@ requirements:
     - class: SubworkflowFeatureRequirement
     - class: ScatterFeatureRequirement
     - class: InlineJavascriptRequirement
+    - class: SchemaDefRequirement
+      types:
+          - $import: ../types/sequence_data.yml
 inputs:
     reference:
         type:
@@ -19,8 +22,8 @@ inputs:
         secondaryFiles: [".1.ht2", ".2.ht2", ".3.ht2", ".4.ht2", ".5.ht2", ".6.ht2", ".7.ht2", ".8.ht2"]
     reference_annotation:
         type: File
-    instrument_data_bams:
-        type: File[]
+    rna_sequence:
+        type: ../types/sequence_data.yml#sequence_data[]
     read_group_id:
         type: string[]
     read_group_fields:
@@ -93,12 +96,12 @@ outputs:
         type: File
         outputSource: cgpbigwig_bamcoverage/outfile
 steps:
-    bam_to_trimmed_fastq_and_hisat_alignments:
-        run: ../subworkflows/bam_to_trimmed_fastq_and_hisat_alignments.cwl
-        scatter: [bam, read_group_id, read_group_fields]
+    sequence_to_trimmed_fastq_and_hisat_alignments:
+        run: ../subworkflows/sequence_to_trimmed_fastq_and_hisat_alignments.cwl
+        scatter: [unaligned, read_group_id, read_group_fields]
         scatterMethod: dotproduct
         in:
-            bam: instrument_data_bams
+            unaligned: rna_sequence
             read_group_id: read_group_id
             read_group_fields: read_group_fields
             adapters: trimming_adapters
@@ -115,7 +118,7 @@ steps:
         in:
             kallisto_index: kallisto_index
             strand: strand
-            fastqs: bam_to_trimmed_fastq_and_hisat_alignments/fastqs
+            fastqs: sequence_to_trimmed_fastq_and_hisat_alignments/fastqs
         out:
             [expression_transcript_table,expression_transcript_h5,fusion_evidence]
     transcript_to_gene:
@@ -128,7 +131,7 @@ steps:
     merge:
         run: ../tools/merge_bams.cwl
         in:
-            bams: bam_to_trimmed_fastq_and_hisat_alignments/aligned_bam
+            bams: sequence_to_trimmed_fastq_and_hisat_alignments/aligned_bam
         out:
             [merged_bam]
     position_sort:

definitions/pipelines/rnaseq_star_fusion.cwl

@@ -7,9 +7,12 @@ requirements:
     - class: MultipleInputFeatureRequirement
     - class: SubworkflowFeatureRequirement
     - class: ScatterFeatureRequirement
+    - class: SchemaDefRequirement
+      types:
+          - $import: ../types/sequence_data.yml
 inputs:
-    instrument_data_bams:
-        type: File[]
+    unaligned:
+        type: ../types/sequence_data.yml#sequence_data[]
     outsam_attrrg_line:
         type: string[]
     star_genome_dir:
@@ -48,6 +51,9 @@ inputs:
         type: File
     sample_name:
         type: string
+    unzip_fastqs:
+        type: boolean?
+        default: true
 outputs:
     cram:
         type: File
@@ -99,17 +105,19 @@ outputs:
         type: File
         outputSource: cgpbigwig_bamcoverage/outfile
 steps:
-    bam_to_trimmed_fastq:
-        run: ../subworkflows/bam_to_trimmed_fastq.cwl
-        scatter: [bam]
+    sequence_to_trimmed_fastq:
+        
+        scatter: [unaligned]
         scatterMethod: dotproduct
+        run: ../subworkflows/sequence_to_trimmed_fastq.cwl
         in:
-            bam: instrument_data_bams
+            unaligned: unaligned
             adapters: trimming_adapters
             adapter_trim_end: trimming_adapter_trim_end
             adapter_min_overlap: trimming_adapter_min_overlap
             max_uncalled: trimming_max_uncalled
             min_readlength: trimming_min_readlength
+            unzip_fastqs: unzip_fastqs
         out:
             [fastqs, fastq1, fastq2]
     strandedness_check:
@@ -120,8 +128,8 @@ steps:
             gtf_file: gtf_file
             kallisto_index: kallisto_index
             cdna_fasta: cdna_fasta
-            reads1: bam_to_trimmed_fastq/fastq1
-            reads2: bam_to_trimmed_fastq/fastq2
+            reads1: sequence_to_trimmed_fastq/fastq1
+            reads2: sequence_to_trimmed_fastq/fastq2
         out:
             [strandedness_check]
     star_align_fusion:
@@ -131,10 +139,10 @@ steps:
             star_genome_dir: star_genome_dir
             gtf_file: gtf_file
             fastq:
-                source: bam_to_trimmed_fastq/fastq1
+                source: sequence_to_trimmed_fastq/fastq1
                 linkMerge: merge_flattened
             fastq2:
-                source: bam_to_trimmed_fastq/fastq2
+                source: sequence_to_trimmed_fastq/fastq2
                 linkMerge: merge_flattened
         out:
             [aligned_bam, chim_junc, splice_junction_out,log_final]
@@ -150,7 +158,7 @@ steps:
         in:
             kallisto_index: kallisto_index
             strand: strand
-            fastqs: bam_to_trimmed_fastq/fastqs
+            fastqs: sequence_to_trimmed_fastq/fastqs
         out:
             [expression_transcript_table,expression_transcript_h5,fusion_evidence]
     transcript_to_gene:

definitions/pipelines/rnaseq_star_fusion_with_xenosplit.cwl

@@ -8,14 +8,17 @@ requirements:
     - class: SubworkflowFeatureRequirement
     - class: ScatterFeatureRequirement
     - class: StepInputExpressionRequirement
+    - class: SchemaDefRequirement
+      types:
+          - $import: ../types/sequence_data.yml
 inputs:
     reference:
         type:
             - string
             - File
         secondaryFiles: [.fai, ^.dict]
-    instrument_data_bams:
-        type: File[]
+    unaligned:
+        type: ../types/sequence_data.yml#sequence_data[]
     outsam_attrrg_line:
         type: string[]
     graft_star_genome_dir:
@@ -63,6 +66,9 @@ inputs:
         type: File
     sample_name:
         type: string
+    unzip_fastqs:
+        type: boolean?
+        default: true
 outputs:
     final_bam:
         type: File
@@ -114,17 +120,18 @@ outputs:
         type: File
         outputSource: cgpbigwig_bamcoverage/outfile
 steps:
-    bam_to_trimmed_fastq:
-        run: ../subworkflows/bam_to_trimmed_fastq.cwl
-        scatter: [bam]
+    sequence_to_trimmed_fastq:
+        run: ../subworkflows/sequence_to_trimmed_fastq.cwl
+        scatter: [unaligned]
         scatterMethod: dotproduct
         in:
-            bam: instrument_data_bams
+            unaligned: unaligned
             adapters: trimming_adapters
             adapter_trim_end: trimming_adapter_trim_end
             adapter_min_overlap: trimming_adapter_min_overlap
             max_uncalled: trimming_max_uncalled
             min_readlength: trimming_min_readlength
+            unzip_fastqs: unzip_fastqs
         out:
             [fastqs, fastq1, fastq2]
     graft_star_align_fusion:
@@ -135,10 +142,10 @@ steps:
             outfile_name_prefix: graft_outfile_name_prefix
             gtf_file: graft_gtf_file
             fastq:
-                source: bam_to_trimmed_fastq/fastq1
+                source: sequence_to_trimmed_fastq/fastq1
                 linkMerge: merge_flattened
             fastq2:
-                source: bam_to_trimmed_fastq/fastq2
+                source: sequence_to_trimmed_fastq/fastq2
                 linkMerge: merge_flattened
         out:
             [aligned_bam, chim_junc, splice_junction_out,log_final]
@@ -150,10 +157,10 @@ steps:
             outfile_name_prefix: host_outfile_name_prefix
             gtf_file: host_gtf_file
             fastq:
-                source: bam_to_trimmed_fastq/fastq1
+                source: sequence_to_trimmed_fastq/fastq1
                 linkMerge: merge_flattened
             fastq2:
-                source: bam_to_trimmed_fastq/fastq2
+                source: sequence_to_trimmed_fastq/fastq2
                 linkMerge: merge_flattened
         out:
             [aligned_bam, chim_junc, splice_junction_out,log_final]   
@@ -165,9 +172,14 @@ steps:
         out:
             [graft_bam, xenosplit_statistics]
     graftbam_to_fastq:
-        run: ../subworkflows/bam_to_trimmed_fastq.cwl
+        run: ../subworkflows/sequence_to_trimmed_fastq.cwl
         in:
-            bam: xenosplit/graft_bam
+            unaligned:
+                source: xenosplit/graft_bam
+                valueFrom: |
+                    ${
+                        return {'sequence': {'bam': self} };
+                    }
             adapters: trimming_adapters
             adapter_trim_end: trimming_adapter_trim_end
             adapter_min_overlap: trimming_adapter_min_overlap

…ns/subworkflows/bam_to_trimmed_fastq.cwl …bworkflows/sequence_to_trimmed_fastq.cwldefinitions/subworkflows/bam_to_trimmed_fastq.cwl renamed to definitions/subworkflows/sequence_to_trimmed_fastq.cwl definitions/subworkflows/bam_to_trimmed_fastq.cwl renamed to definitions/subworkflows/sequence_to_trimmed_fastq.cwl

@@ -2,14 +2,20 @@
 
 cwlVersion: v1.0
 class: Workflow
-label: "bam to trimmed fastqs"
+label: "sequence (bam or fastqs) to trimmed fastqs"
 requirements:
     - class: MultipleInputFeatureRequirement
     - class: SubworkflowFeatureRequirement
     - class: InlineJavascriptRequirement
+    - class: ScatterFeatureRequirement
+    - class: StepInputExpressionRequirement
+    - class: SchemaDefRequirement
+      types:
+          - $import: ../types/sequence_data.yml
+
 inputs:
-    bam:
-        type: File
+    unaligned:
+        type: ../types/sequence_data.yml#sequence_data
     adapters:
         type: File
     adapter_trim_end:
@@ -20,6 +26,8 @@ inputs:
         type: int
     min_readlength:
         type: int
+    unzip_fastqs:
+        type: boolean?
 
 outputs:
     fastqs:
@@ -33,17 +41,26 @@ outputs:
          outputSource: trim_fastq/fastq2
 
 steps:
-    bam_to_fastq:
-        run: ../tools/bam_to_fastq.cwl
-        in:
-            bam: bam
+    sequence_to_fastq:
+        run: ../tools/sequence_to_fastq.cwl
+        in: 
+            bam:
+                source: unaligned
+                valueFrom: "$(self.sequence.hasOwnProperty('bam')? self.sequence.bam : null)"
+            fastq1:
+                source: unaligned
+                valueFrom: "$(self.sequence.hasOwnProperty('fastq1')? self.sequence.fastq1 : null)"
+            fastq2:
+                source: unaligned
+                valueFrom: "$(self.sequence.hasOwnProperty('fastq2')? self.sequence.fastq2 : null)"
+            unzip_fastqs: unzip_fastqs
         out:
             [fastq1, fastq2]
     trim_fastq:
         run: ../tools/trim_fastq.cwl
         in:
-            reads1: bam_to_fastq/fastq1
-            reads2: bam_to_fastq/fastq2
+            reads1: sequence_to_fastq/fastq1
+            reads2: sequence_to_fastq/fastq2
             adapters: adapters
             adapter_trim_end: adapter_trim_end
             adapter_min_overlap: adapter_min_overlap

…o_trimmed_fastq_and_hisat_alignments.cwl …o_trimmed_fastq_and_hisat_alignments.cwldefinitions/subworkflows/bam_to_trimmed_fastq_and_hisat_alignments.cwl renamed to definitions/subworkflows/sequence_to_trimmed_fastq_and_hisat_alignments.cwl definitions/subworkflows/bam_to_trimmed_fastq_and_hisat_alignments.cwl renamed to definitions/subworkflows/sequence_to_trimmed_fastq_and_hisat_alignments.cwl

@@ -2,15 +2,19 @@
 
 cwlVersion: v1.0
 class: Workflow
-label: "bam to trimmed fastqs and HISAT alignments"
+label: "sequence (bam or fastqs) to trimmed fastqs and HISAT alignments"
 requirements:
     - class: MultipleInputFeatureRequirement
     - class: SubworkflowFeatureRequirement
     - class: InlineJavascriptRequirement
     - class: StepInputExpressionRequirement
+    - class: SchemaDefRequirement
+      types:
+          - $import: ../types/sequence_data.yml
+
 inputs:
-    bam:
-        type: File
+    unaligned:
+        type: ../types/sequence_data.yml#sequence_data
     adapters:
         type: File
     adapter_trim_end:
@@ -35,6 +39,8 @@ inputs:
           - "null"
           - type: enum
             symbols: ["first", "second", "unstranded"]
+    unzip_fastqs:
+        type: boolean?
 outputs:
     fastqs:
         type: File[]
@@ -43,17 +49,26 @@ outputs:
         type: File
         outputSource: hisat2_align/aligned_bam
 steps:
-    bam_to_fastq:
-        run: ../tools/bam_to_fastq.cwl
-        in:
-            bam: bam
+    sequence_to_fastq:
+        run: ../tools/sequence_to_fastq.cwl
+        in: 
+            bam:
+                source: unaligned
+                valueFrom: "$(self.sequence.hasOwnProperty('bam')? self.sequence.bam : null)"
+            fastq1:
+                source: unaligned
+                valueFrom: "$(self.sequence.hasOwnProperty('fastq1')? self.sequence.fastq1 : null)"
+            fastq2:
+                source: unaligned
+                valueFrom: "$(self.sequence.hasOwnProperty('fastq2')? self.sequence.fastq2 : null)"
+            unzip_fastqs: unzip_fastqs
         out:
             [fastq1, fastq2]
     trim_fastq:
         run: ../tools/trim_fastq.cwl
         in:
-            reads1: bam_to_fastq/fastq1
-            reads2: bam_to_fastq/fastq2
+            reads1: sequence_to_fastq/fastq1
+            reads2: sequence_to_fastq/fastq2
             adapters: adapters
             adapter_trim_end: adapter_trim_end
             adapter_min_overlap: adapter_min_overlap