workflows/RNA_OntologyEnrichment/RNA_OntologyEnrichment.fsx

+
+//#r @"C:\Users\bvenn\source\repos\FSharp.Stats\bin\FSharp.Stats\net47\FSharp.Stats.dll"
+//#r @"C:\Users\bvenn\source\repos\FSharp.Stats\bin\FSharp.Stats.MSF\net47\FSharp.Stats.MSF.dll"
+//
+//#r @"C:\Users\bvenn\source\repos\BioFSharp\bin\BioFSharp\netstandard2.0\BioFSharp.dll"
+//#r @"C:\Users\bvenn\source\repos\BioFSharp\bin\BioFSharp.Stats\netstandard2.0\BioFSharp.Stats.dll"
+//#r @"C:\Users\bvenn\source\repos\FSharp.Plotly\bin\FSharp.Plotly\netstandard2.0\FSharp.Plotly.dll"
+//#r @"C:\Users\bvenn\source\repos\FSharp.Plotly\bin\FSharp.Plotly\netstandard2.0\Newtonsoft.Json.dll"
+//#r @"C:\Users\bvenn\source\repos\FSharpAux\bin\FSharpAux\netstandard2.0\FSharpAux.dll"
+//#r @"C:\Users\bvenn\source\repos\FSharpAux\bin\FSharpAux.IO\netstandard2.0\FSharpAux.IO.dll"
+//#r @"C:\Users\bvenn\source\repos\FSharpGephiStreamer-developer\bin\FSharpGephiStreamer\netstandard2.0\FSharpGephiStreamer.dll"
+
+#r "nuget: TMEA"
+#r "nuget: Argu"
+
+
+open Argu
+open FSharp.Stats
+//open FSharp.Plotly
+open Plotly.NET
+open System
+open FSharpAux
+//open FSharp.Plotly.Axis
+open Plotly.NET.LayoutObjects
+#time
+
+type Arguments =
+    | [<Mandatory>][<AltCommandLine("-i")>] InputFolder of string
+    | [<Mandatory>][<AltCommandLine("-o")>] OutputFolder of string
+
+    interface IArgParserTemplate with
+        member s.Usage =
+            match s with
+            | InputFolder _ -> "InputFolder"
+            | OutputFolder _ -> "OutputFolder"
+
+let parser = ArgumentParser.Create<Arguments>(programName = "RNA_OntologyEnrichment")
+let results = parser.Parse(fsi.CommandLineArgs |> Array.skip 1)
+
+let inputFolder = results.GetResult(Arguments.InputFolder)
+let outputFolder = results.GetResult(Arguments.OutputFolder)
+
+
+open StyleParam
+
+let myAxis name = LinearAxis.init(Title=name,Mirror=StyleParam.Mirror.All,Ticks=StyleParam.TickOptions.Inside,ShowGrid=false,ShowLine=true)
+let myAxisWODes name = LinearAxis.init(Title=name,Mirror=StyleParam.Mirror.All,Ticks=StyleParam.TickOptions.Inside,ShowGrid=false,ShowLine=true,TickFont=Font.init(StyleParam.FontFamily.Arial,Size=6.))
+let myAxisRange name range = LinearAxis.init(Title=name,Range=StyleParam.Range.MinMax(range), Mirror=StyleParam.Mirror.All,Ticks=StyleParam.TickOptions.Inside,ShowGrid=false,ShowLine=true)
+let myAxisRangeWODes name range = LinearAxis.init(Title=name,Range=StyleParam.Range.MinMax(range), Mirror=StyleParam.Mirror.All,Ticks=StyleParam.TickOptions.Inside,ShowGrid=false,ShowLine=true,TickFont=Font.init(StyleParam.FontFamily.Arial,Size=6.))
+let styleChart x y chart = chart |> Chart.withXAxis (myAxis x) |> Chart.withYAxis (myAxis y)
+let styleChartwR x y (min,max) chart = chart |> Chart.withXAxis (myAxisRange x (min,max)) |> Chart.withYAxis (myAxis y)
+
+let getReg (vec:Vector<float>) rep = vec.[0..rep]
+Vector.init 2 (fun _ -> 2.)
+
+let getMap path = 
+    System.IO.File.ReadAllLines (path) 
+    |> Array.map (fun x -> 
+        let tmp = x.Split '\t'
+        tmp.[0],tmp.[1]
+        )
+    |> Map.ofArray
+
+let j2mD = getMap (System.IO.Path.Combine(inputFolder,@"mmSnapShot.txt"))
+let j2GD = getMap (System.IO.Path.Combine(inputFolder,@"goSnapShot.txt"))
+let j2S =  getMap (System.IO.Path.Combine(inputFolder,@"SynonymSnapShot.txt"))
+let j2L =  getMap (System.IO.Path.Combine(inputFolder,@"LocalizationSnapShot.txt"))
+
+let j2GDStr (str :String) =
+    let str' = (str.Split '.').[0..1] |> String.concat "."
+    if (str.Split '.').Length > 3 then ""
+    else 
+        if j2GD.ContainsKey str' then j2GD.[str']
+        else 
+            printfn "Error!!!%s" str'
+            ""        
+let j2mDStr (str :String) =
+    let str' = (str.Split '.').[0..1] |> String.concat "."
+    if (str.Split '.').Length > 3 then ""
+    else 
+        if j2mD.ContainsKey str' then j2mD.[str']
+        else 
+            printfn "Error!!!%s" str'
+            ""
+let j2SStr (str :String) =
+    let str' = (str.Split '.').[0..1] |> String.concat "."
+    if (str.Split '.').Length > 3 then ""
+    else 
+        if j2S.ContainsKey str' then j2S.[str']
+        else 
+            printfn "Error!!!%s" str'
+            ""
+let j2LStr (str :String) =
+    let str' = (str.Split '.').[0..1] |> String.concat "."
+    if (str.Split '.').Length > 3 then ""
+    else 
+        if j2L.ContainsKey str' then j2L.[str']
+        else 
+            printfn "Error!!!%s" str'
+            ""
+
+open BioFSharp.Stats
+
+let (set35,set40) =
+    System.IO.File.ReadAllLines(System.IO.Path.Combine(inputFolder,@"2020_01_29_DEGs.txt"))
+    |> Array.tail
+    |> Array.map (fun x -> 
+        let tmp = x.Split '\t'
+        let id = tmp.[0]
+        let anno = j2mDStr id
+        let create id anno index = OntologyEnrichment.createOntologyItem id anno index id                                                                                                                                                                                                            
+        let s35 = tmp.[1..28] |> Array.chunkBySize 2 |> JaggedArray.transpose |> Seq.head |> Array.map (fun x -> if x = "Down Regulated" then create id anno 1 elif x = "Up Regulated" then create id anno 2 else create id anno 0) 
+        let s40 = tmp.[1..28] |> Array.chunkBySize 2 |> JaggedArray.transpose |> Seq.last |> Array.map (fun x -> if x = "Down Regulated" then create id anno 1 elif x = "Up Regulated" then create id anno 2 else create id anno 0) 
+        s35,s40)
+    |> Array.unzip
+    |> fun (x,y) -> 
+        let xs = 
+            JaggedArray.transpose x 
+            |> Array.map (fun t -> 
+                Seq.collect (OntologyEnrichment.splitMultipleAnnotationsBy ';') t 
+                |>  OntologyEnrichment.expandOntologyTree
+                )
+        let ys = 
+            JaggedArray.transpose y 
+            |> Array.map (fun t -> 
+                Seq.collect (OntologyEnrichment.splitMultipleAnnotationsBy ';') t 
+                |>  OntologyEnrichment.expandOntologyTree
+                )
+        xs,ys
+
+
+System.IO.Directory.CreateDirectory(outputFolder)
+
+set40
+|> Seq.iteri (fun i x -> 
+    let enriched = 
+        OntologyEnrichment.CalcOverEnrichment 2 None None x
+        |> Testing.MultipleTesting.benjaminiHochbergFDRBy (fun x -> x,x.PValue)
+        |> Seq.choose (fun (x,pval) -> 
+            if pval < 0.05 && x.NumberOfDEsInBin > 3 then 
+                Some (sprintf "%s\t%i\t%f" x.OntologyTerm x.NumberOfDEsInBin pval)
+            else None)
+    let depleted = 
+        OntologyEnrichment.CalcOverEnrichment 1 None None x
+        |> Testing.MultipleTesting.benjaminiHochbergFDRBy (fun x -> x,x.PValue)
+        |> Seq.choose (fun (x,pval) -> 
+            if pval < 0.05 && x.NumberOfDEsInBin > 3 then 
+                Some (sprintf "%s\t%i\t%f" x.OntologyTerm x.NumberOfDEsInBin pval)
+            else None)
+    System.IO.File.WriteAllLines(System.IO.Path.Combine(outputFolder, "Rna40Enrichments1vs" + string (i+1)+ "_enr.txt"),enriched)
+    System.IO.File.WriteAllLines(System.IO.Path.Combine(outputFolder, "Rna40Enrichments1vs" + string (i+1)+ "_dep.txt"),depleted)
+    )
+
+set35.Length
+
+
+//Seq.item 0 set35 |> Seq.map (fun x-> x.GroupIndex)
+
+////down = 1
+////up = 2
+
+//let print inp co ix =
+//    let data = 
+//        match inp with
+//        | "35" -> set35
+//        | "40" -> set40
+//    let con =
+//        match co with
+//        | "HS" -> [|0..7|]
+//        | "Re" -> [|8..13|]
+//    con
+//    |> Array.map (fun x-> data.[x])
+//    |> Array.map Array.ofSeq
+//    |> JaggedArray.transpose
+//    |> Array.map (fun x -> 
+//        let pos = x |> Array.filter (fun t -> t.GroupIndex = ix) |> fun k -> k.Length >= 2
+//        if pos then {x.[0] with GroupIndex = ix}
+//        else {x.[0] with GroupIndex = 0})
+//    |> OntologyEnrichment.CalcOverEnrichment ix None None
+//    |> Testing.MultipleTesting.benjaminiHochbergFDRBy (fun x -> x,x.PValue)
+//    |> Seq.choose (fun (x,pval) -> 
+//        if pval < 0.05 && x.NumberOfDEsInBin > 3 then 
+//            Some (sprintf "%s\t%i\t%f" x.OntologyTerm x.NumberOfDEsInBin pval)
+//        else None)
+//    |> Seq.iter (printfn "%s")
+
+//print "35" "HS" 1
+//print "35" "HS" 2
+//print "35" "Re" 1
+//print "35" "Re" 2
+
+//print "40" "HS" 1
+//print "40" "HS" 2
+//print "40" "Re" 1
+//print "40" "Re" 2

workflows/RNA_OntologyEnrichment/workflow.cwl

+cwlVersion: v1.2
+class: CommandLineTool
+hints:
+  DockerRequirement:
+    dockerPull: mcr.microsoft.com/dotnet/sdk:6.0
+requirements:
+  - class: InitialWorkDirRequirement
+    listing:
+      - entryname: RNA_OntologyEnrichment.fsx
+        entry:
+          $include: RNA_OntologyEnrichment.fsx
+  - class: EnvVarRequirement
+    envDef:
+      - envName: DOTNET_NOLOGO
+        envValue: "true"
+  - class: NetworkAccess
+    networkAccess: true
+baseCommand: [dotnet, fsi, RNA_OntologyEnrichment.fsx]
+arguments: [-o, $(outputFolder)]
+inputs:
+  inputFolder:
+    type: Directory
+    inputBinding:
+      prefix: -i
+
+outputs:
+  outputFolder:
+    type: Directory
+    outputBinding:
+      glob: $(outputFolder)/*.*
+
+arc:has technology type:
+  - class: arc:technology type
+    arc:annotation value: "Docker Container"
+
+arc:technology platform: ".NET"
+
+arc:performer:
+  - class: arc:Person
+    arc:first name: "Benedikt"
+    arc:last name: "Venn"
+    arc:email: "venn@bio.uni-kl.de "
+    arc:affiliation: "RPTU Kaiserslautern/Landau"
+    arc:has role: 
+      - class: arc:role
+        arc:term accession: "https://credit.niso.org/contributor-roles/formal-analysis/"
+        arc:annotation value: "Formal analysis"
+
+$namespaces:
+  arc: https://github.com/nfdi4plants/ARC_ontology
+
+$schemas:
+  - https://raw.githubusercontent.com/nfdi4plants/ARC_ontology/main/ARC_v2.0.owl