From 16fd13f1f6943fefad67214510882a1d0a9707c4 Mon Sep 17 00:00:00 2001
From: Zach Skidmore <zlskidmore@gmail.com>
Date: Tue, 19 Mar 2019 12:36:08 -0500
Subject: [PATCH 1/7] initial test of cram support

---
 definitions/pipelines/rnaseq.cwl              | 16 +++++---
 ...to_trimmed_fastq_and_hisat_alignments.cwl} | 21 ++++++++--
 definitions/tools/input_to_bam.cwl            | 41 +++++++++++++++++++
 definitions/tools/revert_input.cwl            | 30 ++++++++++++++
 example_data/rnaseq/workflow.yaml             |  2 +-
 5 files changed, 99 insertions(+), 11 deletions(-)
 rename definitions/subworkflows/{bam_to_trimmed_fastq_and_hisat_alignments.cwl => input_to_trimmed_fastq_and_hisat_alignments.cwl} (81%)
 create mode 100644 definitions/tools/input_to_bam.cwl
 create mode 100644 definitions/tools/revert_input.cwl

diff --git a/definitions/pipelines/rnaseq.cwl b/definitions/pipelines/rnaseq.cwl
index ce9bcbbb2..cf25742d9 100644
--- a/definitions/pipelines/rnaseq.cwl
+++ b/definitions/pipelines/rnaseq.cwl
@@ -9,12 +9,15 @@ requirements:
     - class: ScatterFeatureRequirement
     - class: InlineJavascriptRequirement
 inputs:
+    cram_reference:
+        type: File?
+        doc: Reference file used for cram decompression
     reference_index:
         type: File #this requires an extra file with the basename
         secondaryFiles: [".1.ht2", ".2.ht2", ".3.ht2", ".4.ht2", ".5.ht2", ".6.ht2", ".7.ht2", ".8.ht2"]
     reference_annotation:
         type: File
-    instrument_data_bams:
+    instrument_data:
         type: File[]
     read_group_id:
         type: string[]
@@ -76,12 +79,13 @@ outputs:
         type: File
         outputSource: kallisto/fusion_evidence
 steps:
-    bam_to_trimmed_fastq_and_hisat_alignments:
-        run: ../subworkflows/bam_to_trimmed_fastq_and_hisat_alignments.cwl
+    input_to_trimmed_fastq_and_hisat_alignments:
+        run: ../subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
         scatter: [bam, read_group_id, read_group_fields]
         scatterMethod: dotproduct
         in:
-            bam: instrument_data_bams
+            cram_reference: cram_reference
+            input: instrument_data
             read_group_id: read_group_id
             read_group_fields: read_group_fields
             adapters: trimming_adapters
@@ -99,7 +103,7 @@ steps:
             kallisto_index: kallisto_index
             strand: strand
             fastqs:
-                source: bam_to_trimmed_fastq_and_hisat_alignments/fastqs
+                source: input_to_trimmed_fastq_and_hisat_alignments/fastqs
                 valueFrom: |
                     ${
                       for(var i=0;i<self.length;i++){self[i] = self[i].reverse()}
@@ -117,7 +121,7 @@ steps:
     merge:
         run: ../tools/merge_bams.cwl
         in:
-            bams: bam_to_trimmed_fastq_and_hisat_alignments/aligned_bam
+            bams: input_to_trimmed_fastq_and_hisat_alignments/aligned_bam
         out:
             [merged_bam]
     index_bam:
diff --git a/definitions/subworkflows/bam_to_trimmed_fastq_and_hisat_alignments.cwl b/definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
similarity index 81%
rename from definitions/subworkflows/bam_to_trimmed_fastq_and_hisat_alignments.cwl
rename to definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
index f8d3170be..a9c1071e6 100644
--- a/definitions/subworkflows/bam_to_trimmed_fastq_and_hisat_alignments.cwl
+++ b/definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
@@ -8,7 +8,9 @@ requirements:
     - class: SubworkflowFeatureRequirement
     - class: InlineJavascriptRequirement
 inputs:
-    bam:
+    cram_reference:
+        type: File?
+    input:
         type: File
     adapters:
         type: File
@@ -39,10 +41,21 @@ outputs:
         type: File
         outputSource: hisat2_align/aligned_bam
 steps:
+    input_to_bam:
+        run: ../tools/input_to_bam.cwl
+        in:
+            input: input
+            reference_cram: reference_cram
+        out: [bam_file]
+    revert_bam:
+        run: ../tools/revert_bam.cwl
+        in:
+            bam: input_to_bam/bam_file
+        out: [reverted_bam]
     bam_to_fastq:
         run: ../tools/bam_to_fastq.cwl
         in:
-            bam: bam
+            bam: revert_bam/reverted_bam
         out:
             [fastq1, fastq2]
     trim_fastq:
@@ -61,10 +74,10 @@ steps:
         run: ../tools/hisat2_align.cwl
         in:
             reference_index: reference_index
-            fastq1: 
+            fastq1:
                 source: trim_fastq/fastqs
                 valueFrom: $(self[0])
-            fastq2: 
+            fastq2:
                 source: trim_fastq/fastqs
                 valueFrom: $(self[1])
             read_group_id: read_group_id
diff --git a/definitions/tools/input_to_bam.cwl b/definitions/tools/input_to_bam.cwl
new file mode 100644
index 000000000..32e019bd1
--- /dev/null
+++ b/definitions/tools/input_to_bam.cwl
@@ -0,0 +1,41 @@
+#!/usr/bin/env cwl-runner
+
+class: CommandLineTool
+
+cwlVersion: v1.0
+
+baseCommand: [ "samtools", "view" ]
+
+requirements:
+  - class: DockerRequirement
+    dockerImageId: zlskidmore/samtools:1.9
+    dockerPull: zlskidmore/samtools:1.9
+  - class: ResourceRequirement
+    ramMin: 20000
+    coresMin: 8
+
+arguments:
+  - valueFrom: "8"
+    position: 1
+    prefix: "--threads"
+  - valueFrom: "file.bam"
+    position: 2
+    prefix: "-o"
+  - valueFrom: "-b"
+    position: 3
+
+inputs:
+  input:
+    type: File
+    inputBinding:
+      position: 5
+  cram_reference:
+    type: File?
+    inputBinding:
+      position: 4
+
+outputs:
+  bam_file:
+    type: File
+    outputBinding:
+      glob: file.bam
diff --git a/definitions/tools/revert_input.cwl b/definitions/tools/revert_input.cwl
new file mode 100644
index 000000000..a232ab4da
--- /dev/null
+++ b/definitions/tools/revert_input.cwl
@@ -0,0 +1,30 @@
+#!/usr/bin/env cwl-runner
+
+cwlVersion: v1.0
+class: CommandLineTool
+label: "Picard: Revert Input"
+baseCommand: ["/usr/bin/java", "-Xmx4g", "-jar", "/opt/picard/picard.jar", "RevertSam", "VALIDATION_STRINGENCY=SILENT"]
+requirements:
+    - class: ResourceRequirement
+      coresMin: 1
+      ramMin: 6000
+      tmpdirMin: 25000
+    - class: DockerRequirement
+      dockerPull: "mgibio/rnaseq"
+arguments:
+    - valueFrom: "reverted.bam"
+      position: 2
+      prefix: "O="
+
+inputs:
+    bam:
+        type: File
+        inputBinding:
+            prefix: "I="
+            separate: false
+            position: 1
+outputs:
+    reverted_bam:
+        type: File
+        outputBinding:
+            glob: "reverted.bam"
diff --git a/example_data/rnaseq/workflow.yaml b/example_data/rnaseq/workflow.yaml
index 10d30f92c..d9905d366 100644
--- a/example_data/rnaseq/workflow.yaml
+++ b/example_data/rnaseq/workflow.yaml
@@ -1,4 +1,4 @@
-instrument_data_bams:
+instrument_data:
     - class: File
       path: gerald_H3MYFBBXX_4_GCCAAT_1k.bam
     - class: File

From 99f1ffbee26dd6a98c9af12d2f48b090b14a01a2 Mon Sep 17 00:00:00 2001
From: Zach Skidmore <zlskidmore@gmail.com>
Date: Tue, 19 Mar 2019 13:13:50 -0500
Subject: [PATCH 2/7] fix scatter

---
 definitions/pipelines/rnaseq.cwl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/definitions/pipelines/rnaseq.cwl b/definitions/pipelines/rnaseq.cwl
index cf25742d9..e4478c15b 100644
--- a/definitions/pipelines/rnaseq.cwl
+++ b/definitions/pipelines/rnaseq.cwl
@@ -81,7 +81,7 @@ outputs:
 steps:
     input_to_trimmed_fastq_and_hisat_alignments:
         run: ../subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
-        scatter: [bam, read_group_id, read_group_fields]
+        scatter: [input, read_group_id, read_group_fields]
         scatterMethod: dotproduct
         in:
             cram_reference: cram_reference

From 40604967071b49add0c96f63cbe6531c225ff4e9 Mon Sep 17 00:00:00 2001
From: Zach Skidmore <zlskidmore@gmail.com>
Date: Tue, 19 Mar 2019 13:19:00 -0500
Subject: [PATCH 3/7] fix typos

---
 .../input_to_trimmed_fastq_and_hisat_alignments.cwl           | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl b/definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
index a9c1071e6..335325315 100644
--- a/definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
+++ b/definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
@@ -45,10 +45,10 @@ steps:
         run: ../tools/input_to_bam.cwl
         in:
             input: input
-            reference_cram: reference_cram
+            reference_cram: cram_reference
         out: [bam_file]
     revert_bam:
-        run: ../tools/revert_bam.cwl
+        run: ../tools/revert_input.cwl
         in:
             bam: input_to_bam/bam_file
         out: [reverted_bam]

From 0b793a42583029604d8785889f0290a044802f0d Mon Sep 17 00:00:00 2001
From: Zach Skidmore <zlskidmore@gmail.com>
Date: Tue, 19 Mar 2019 13:24:34 -0500
Subject: [PATCH 4/7] fixed another typo

---
 .../input_to_trimmed_fastq_and_hisat_alignments.cwl             | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl b/definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
index 335325315..dc7186a05 100644
--- a/definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
+++ b/definitions/subworkflows/input_to_trimmed_fastq_and_hisat_alignments.cwl
@@ -45,7 +45,7 @@ steps:
         run: ../tools/input_to_bam.cwl
         in:
             input: input
-            reference_cram: cram_reference
+            cram_reference: cram_reference
         out: [bam_file]
     revert_bam:
         run: ../tools/revert_input.cwl

From c7ef29f0cb6b69e7d428a3c65127e5b22aa5cb71 Mon Sep 17 00:00:00 2001
From: Zach Skidmore <zlskidmore@gmail.com>
Date: Tue, 19 Mar 2019 15:09:12 -0500
Subject: [PATCH 5/7] forgot the -T prefix for samtools

---
 definitions/tools/input_index.cwl  | 30 ++++++++++++++++++++++++++++++
 definitions/tools/input_to_bam.cwl |  1 +
 2 files changed, 31 insertions(+)
 create mode 100644 definitions/tools/input_index.cwl

diff --git a/definitions/tools/input_index.cwl b/definitions/tools/input_index.cwl
new file mode 100644
index 000000000..ccc543696
--- /dev/null
+++ b/definitions/tools/input_index.cwl
@@ -0,0 +1,30 @@
+#!/usr/bin/env cwl-runner
+
+class: CommandLineTool
+
+cwlVersion: v1.0
+
+baseCommand: [ "samtools", "index" ]
+
+requirements:
+  - class: DockerRequirement
+    dockerImageId: zlskidmore/samtools:1.9
+    dockerPull: zlskidmore/samtools:1.9
+  - class: InitialWorkDirRequirement
+    listing:
+      - $(inputs.bam_file)
+  - class: ResourceRequirement
+    ramMin: 4000
+
+inputs:
+  input_file:
+    type: File
+    inputBinding:
+      position: 1
+
+outputs:
+  input_index:
+    type: File
+    secondaryFiles: [.bai, .crai]
+    outputBinding:
+      glob: "*.bam"
diff --git a/definitions/tools/input_to_bam.cwl b/definitions/tools/input_to_bam.cwl
index 32e019bd1..d2cd11f74 100644
--- a/definitions/tools/input_to_bam.cwl
+++ b/definitions/tools/input_to_bam.cwl
@@ -33,6 +33,7 @@ inputs:
     type: File?
     inputBinding:
       position: 4
+      prefix: "-T"
 
 outputs:
   bam_file:

From 3181f61a2fd2cc1e069b7a315f92c36eaa058417 Mon Sep 17 00:00:00 2001
From: Zach Skidmore <zlskidmore@gmail.com>
Date: Tue, 19 Mar 2019 15:36:37 -0500
Subject: [PATCH 6/7] removed uneccessary file

---
 definitions/tools/input_index.cwl | 30 ------------------------------
 1 file changed, 30 deletions(-)
 delete mode 100644 definitions/tools/input_index.cwl

diff --git a/definitions/tools/input_index.cwl b/definitions/tools/input_index.cwl
deleted file mode 100644
index ccc543696..000000000
--- a/definitions/tools/input_index.cwl
+++ /dev/null
@@ -1,30 +0,0 @@
-#!/usr/bin/env cwl-runner
-
-class: CommandLineTool
-
-cwlVersion: v1.0
-
-baseCommand: [ "samtools", "index" ]
-
-requirements:
-  - class: DockerRequirement
-    dockerImageId: zlskidmore/samtools:1.9
-    dockerPull: zlskidmore/samtools:1.9
-  - class: InitialWorkDirRequirement
-    listing:
-      - $(inputs.bam_file)
-  - class: ResourceRequirement
-    ramMin: 4000
-
-inputs:
-  input_file:
-    type: File
-    inputBinding:
-      position: 1
-
-outputs:
-  input_index:
-    type: File
-    secondaryFiles: [.bai, .crai]
-    outputBinding:
-      glob: "*.bam"

From 1bc747a0a1819591f016ed8b4116aedf71d62dd3 Mon Sep 17 00:00:00 2001
From: Zach Skidmore <zlskidmore@gmail.com>
Date: Tue, 19 Mar 2019 16:58:58 -0500
Subject: [PATCH 7/7] bumped java heap space

---
 definitions/tools/revert_input.cwl | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/definitions/tools/revert_input.cwl b/definitions/tools/revert_input.cwl
index a232ab4da..c4098819f 100644
--- a/definitions/tools/revert_input.cwl
+++ b/definitions/tools/revert_input.cwl
@@ -3,11 +3,11 @@
 cwlVersion: v1.0
 class: CommandLineTool
 label: "Picard: Revert Input"
-baseCommand: ["/usr/bin/java", "-Xmx4g", "-jar", "/opt/picard/picard.jar", "RevertSam", "VALIDATION_STRINGENCY=SILENT"]
+baseCommand: ["/usr/bin/java", "-Xmx16g", "-jar", "/opt/picard/picard.jar", "RevertSam", "VALIDATION_STRINGENCY=SILENT"]
 requirements:
     - class: ResourceRequirement
       coresMin: 1
-      ramMin: 6000
+      ramMin: 18000
       tmpdirMin: 25000
     - class: DockerRequirement
       dockerPull: "mgibio/rnaseq"