[WIP] pRESTO pipeline

tools/pRESTO/.dockerignore

@@ -0,0 +1 @@
+*.cwl

tools/pRESTO/Dockerfile

@@ -0,0 +1,19 @@
+FROM debian:testing
+RUN apt-get -y update && apt-get install -y --no-install-recommends \
+	muscle \
+	vsearch \
+	python-biopython \
+	python-numpy \
+	python-scipy \
+	python-numpy \
+	python-pandas \
+        curl \
+ && apt-get clean \
+ && rm -rf /var/lib/apt/lists/*
+ENV p_MAJOR=0
+ENV p_MINOR=4
+ENV p_PATCH=8
+ENV p_ARCHIVE=pRESTO_v${p_MAJOR}.${p_MINOR}.${p_PATCH}.tar.gz
+RUN curl -SL \
+	"http://clip.med.yale.edu/presto/downloadHelper.php?dbTable=presto&downloadFile=${p_ARCHIVE}" \
+    | tar xzC /usr/local

tools/pRESTO/MaskPrimers.py.cwl

@@ -0,0 +1,112 @@
+class: CommandLineTool
+id: MaskPrimers.py
+label: pRESTO MaskPrimers.py
+
+description: |
+  usage: MaskPrimers.py [-h] [-v]  ...
+
+  Removes primers and annotates sequences with primer and barcode identifiers
+
+  optional arguments:
+    -h, --help     show this help message and exit
+    -v, --version  show program's version number and exit
+
+  subcommands:
+                   Alignment method
+      align        Find primer matches using pairwise local alignment
+      score        Find primer matches by scoring primers at a fixed position
+
+  output files:
+    mask-pass      processed reads with successful primer matches.
+    mask-fail      raw reads failing primer identification.
+
+  output annotation fields:
+    SEQORIENT      the orientation of the output sequence. Either F (input)
+                   or RC (reverse complement of input).
+    PRIMER         name of the best primer match.
+    BARCODE        the sequence preceding the primer match. Only output when
+                   the --barcode flag is specified.
+
+requirements:
+  - import: node-engine.cwl
+
+inputs:
+  - id: "#subcommand"
+    type: enum
+    symbols: ["score", "align"]
+    inputBinding:
+      position: 0 
+
+  - id: "#input_files"
+    type: File 
+    inputBinding:
+      prefix: "-s"
+
+  - id: "#primer_file"
+    type: File 
+    inputBinding:
+      prefix: "-p"
+
+  - id: "#mode"
+    type: enum
+    symbols: ["cut","mask","trim","tag"]
+    inputBinding:
+      prefix: "--mode"
+
+  - id: "#barcode"
+    type: boolean
+    inputBinding:
+      prefix: "--barcode"
+
+  - id: "#start"
+    type: int
+    inputBinding:
+      prefix: "--start"
+
+  - id: "#maxerror"
+    type: float
+    inputBinding:
+      prefix: "--maxerror"
+
+  - id: "#nproc"
+    type: int
+    inputBinding:
+      prefix: "--nproc"
+
+  - id: "#outname"
+    type: string
+    inputBinding:
+      prefix: "--outname"
+
+  - id: "#outdir"
+    type: string
+    default:"outdir"
+    inputBinding:
+      prefix: "--outdir"
+
+  - id: "#error_log"
+    type: string
+    default: "Pipeline.err"
+
+  - id: "#pipeline_log"
+    type: string
+    default: "Pipeline.log"
+
+outputs:
+  - id: "#outputs"
+    type: File
+    outputBinding:
+      glob:
+        engine: "node-engine.cwl"
+        script: >
+        $job.outname + '_quality-pass.fastq'
+
+stdout:
+  engine: cwl:JsonPointer
+  script: /job/pipeline_log
+
+# stderr:
+#   engine: cwl:JsonPointer
+#   script: /job/error_log
+
+# _primers-pass.fastq