From 28e57415662c2ebdaa85063425ff340362064f6e Mon Sep 17 00:00:00 2001
From: Dominik Brilhaus <brilhaus@nfdi4plants.org>
Date: Mon, 4 Nov 2024 18:00:38 +0100
Subject: [PATCH] [needs f# fix] add isaSampleToRawDataSeq

---
 runs/isaSampleToRawDataSeq/README.md          | 10 ++
 runs/isaSampleToRawDataSeq/job.yml            |  6 ++
 workflows/isaSampleToRawDataSeq/README.md     |  7 ++
 .../isaSampleToRawDataSeq.cwl                 | 43 ++++++++
 .../isaSampleToRawDataSeq.fsx                 | 98 +++++++++++++++++++
 5 files changed, 164 insertions(+)
 create mode 100644 runs/isaSampleToRawDataSeq/README.md
 create mode 100644 runs/isaSampleToRawDataSeq/job.yml
 create mode 100644 workflows/isaSampleToRawDataSeq/README.md
 create mode 100644 workflows/isaSampleToRawDataSeq/isaSampleToRawDataSeq.cwl
 create mode 100644 workflows/isaSampleToRawDataSeq/isaSampleToRawDataSeq.fsx

diff --git a/runs/isaSampleToRawDataSeq/README.md b/runs/isaSampleToRawDataSeq/README.md
new file mode 100644
index 0000000..d8e1dcb
--- /dev/null
+++ b/runs/isaSampleToRawDataSeq/README.md
@@ -0,0 +1,10 @@
+
+
+
+```bash
+cd runs/isaSampleToRawDataSeq
+```
+
+```bash
+cwltool ../../workflows/isaSampleToRawDataSeq/isaSampleToRawDataSeq.cwl job.yml
+```
diff --git a/runs/isaSampleToRawDataSeq/job.yml b/runs/isaSampleToRawDataSeq/job.yml
new file mode 100644
index 0000000..803a637
--- /dev/null
+++ b/runs/isaSampleToRawDataSeq/job.yml
@@ -0,0 +1,6 @@
+arcPath: 
+  class: Directory
+  path: ../../
+assayName: "Talinum_RNASeq_minimal"
+outName: "rnaseq-samples"
+startingNodeNum: 1
diff --git a/workflows/isaSampleToRawDataSeq/README.md b/workflows/isaSampleToRawDataSeq/README.md
new file mode 100644
index 0000000..e831cf1
--- /dev/null
+++ b/workflows/isaSampleToRawDataSeq/README.md
@@ -0,0 +1,7 @@
+
+
+```bash
+dotnet fsi isa-sampleToRawDataSeq.fsx ../../ Talinum_RNASeq_minimal 1 rnaseq-samples
+
+
+```
diff --git a/workflows/isaSampleToRawDataSeq/isaSampleToRawDataSeq.cwl b/workflows/isaSampleToRawDataSeq/isaSampleToRawDataSeq.cwl
new file mode 100644
index 0000000..9916ed4
--- /dev/null
+++ b/workflows/isaSampleToRawDataSeq/isaSampleToRawDataSeq.cwl
@@ -0,0 +1,43 @@
+cwlVersion: v1.2
+class: CommandLineTool
+hints:
+  DockerRequirement:
+    dockerPull: mcr.microsoft.com/dotnet/sdk:6.0
+requirements:
+  - class: InitialWorkDirRequirement
+    listing:
+      - entryname: isa-sampleToRawDataSeq.fsx
+        entry:
+          $include: isa-sampleToRawDataSeq.fsx
+  - class: EnvVarRequirement
+    envDef:
+      - envName: DOTNET_NOLOGO
+        envValue: "true"
+  - class: NetworkAccess
+    networkAccess: true
+baseCommand: [dotnet, fsi, isa-sampleToRawDataSeq.fsx]
+inputs:
+  arcPath:
+    type: Directory
+    inputBinding:
+      position: 1
+  assayName:
+    type: string
+    inputBinding:
+      position: 2
+  outName:
+    type: string
+    inputBinding:
+      position: 3
+  startingNodeNum:
+    type: int
+    inputBinding:
+      position: 4
+
+outputs:
+  output:
+    type: File[]
+    outputBinding:
+      glob:
+        - "*.tsv"
+        - "*.xlsx"
diff --git a/workflows/isaSampleToRawDataSeq/isaSampleToRawDataSeq.fsx b/workflows/isaSampleToRawDataSeq/isaSampleToRawDataSeq.fsx
new file mode 100644
index 0000000..530a186
--- /dev/null
+++ b/workflows/isaSampleToRawDataSeq/isaSampleToRawDataSeq.fsx
@@ -0,0 +1,98 @@
+// Pull out the full ISA process sequence (incl. all metadata) leading to the first Raw Data Node 
+
+// Dependencies
+
+#r "nuget: ARCtrl.NET, 2.0.2"
+#r "nuget: ARCtrl.QueryModel, 1.0.5"
+#r "nuget: FsSpreadsheet.CsvIO, 6.2.0"
+
+open FsSpreadsheet.CsvIO
+open FsSpreadsheet.Net
+open System.IO
+open ARCtrl.NET
+open ARCtrl
+open ARCtrl.ISA
+open ARCtrl.QueryModel
+
+// input parameters
+
+let args : string array = fsi.CommandLineArgs |> Array.tail
+let arcPath = args.[0]
+let assayName = args.[1]
+let outName = args.[2]
+
+let startingNodeNum = args.[3] |> int
+
+
+// test parameters
+
+// let source = __SOURCE_DIRECTORY__
+// let arcPath = Path.Combine(source, "../../")
+
+// let assayName = "pick2012_illumina_rnaseq"
+
+// let outName = "out.csv"
+
+// Load ARC
+
+let arc = ARC.load(arcPath)
+
+let inv = arc.ISA.Value 
+
+// Load first data node
+
+let firstData = inv.GetAssay(assayName).FirstData
+
+// Create headers for output table
+let headers = [
+    CompositeHeader.Input IOType.Sample
+    for v in inv.ArcTables.ValuesOf firstData.[0] do
+        if v.IsCharacteristicValue then
+            CompositeHeader.Characteristic v.Category
+        elif v.IsParameterValue then
+            CompositeHeader.Parameter v.Category
+        elif v.IsFactorValue then
+            CompositeHeader.Factor v.Category
+        elif v.IsComponent then
+            CompositeHeader.Component v.Category
+        else failwithf "what the f is %O" v
+
+    CompositeHeader.Output IOType.RawDataFile
+]
+
+// Create rows
+
+let getRow (d: QNode) = 
+    [|
+
+    CompositeCell.createFreeText (inv.ArcTables.SamplesOf d).[startingNodeNum].Name
+
+    for v in inv.ArcTables.ValuesOf d do
+        if v.HasUnit then
+            CompositeCell.Unitized(v.ValueText, v.Unit)
+        else
+            CompositeCell.Term(v.Value.AsOntology())
+
+    CompositeCell.FreeText d.Name
+    
+    |]
+
+// Combine into table
+
+let t = ArcTable.init "FullTable"
+t.Headers <- ResizeArray headers
+
+for d in firstData do
+    t.AddRow (getRow d)
+
+// Small detour via workbook
+let ws = Spreadsheet.ArcTable.toFsWorksheet t
+
+let wb = new FsSpreadsheet.FsWorkbook()
+
+wb.AddWorksheet ws
+
+// Write to csv
+
+wb.ToCsvFile (outName + ".tsv", Separator = '\t')
+wb.ToXlsxFile (outName + ".xlsx")
-- 
GitLab