From b7b4c55c2ee412894a920f103837dc4a55feb29b Mon Sep 17 00:00:00 2001
From: Yueyao Gao <gaoyueya@broadinstitute.org>
Date: Sat, 6 Jul 2024 16:24:36 -0400
Subject: [PATCH] first commit of cnv prober

---
 .dockstore.yml                              |  5 ++
 CNV_Array_Prober/cnvArrayProber.inputs.json | 11 ++++
 CNV_Array_Prober/cnvArrayProber.wdl         | 63 +++++++++++++++++++++
 3 files changed, 79 insertions(+)
 create mode 100644 CNV_Array_Prober/cnvArrayProber.inputs.json
 create mode 100644 CNV_Array_Prober/cnvArrayProber.wdl

diff --git a/.dockstore.yml b/.dockstore.yml
index 4ce4927..60df846 100644
--- a/.dockstore.yml
+++ b/.dockstore.yml
@@ -129,3 +129,8 @@ workflows:
    primaryDescriptorPath: /PECGS-QUICviz/QUICviz.wdl
    testParameterFiles:
     - /PECGS-QUICviz/QUICviz.inputs.json
+ - name: cnvArrayProber
+   subclass: WDL
+   primaryDescriptorPath: /CNV_Array_Prober/cnvArrayProber.wdl
+   testParameterFiles:
+    - /CNV_Array_Prober/cnvArrayProber.inputs.json
\ No newline at end of file
diff --git a/CNV_Array_Prober/cnvArrayProber.inputs.json b/CNV_Array_Prober/cnvArrayProber.inputs.json
new file mode 100644
index 0000000..de8564f
--- /dev/null
+++ b/CNV_Array_Prober/cnvArrayProber.inputs.json
@@ -0,0 +1,11 @@
+{
+  "cnvArrayProber.cnvArrayProber.cpu": "Int (optional, default = 1)",
+  "cnvArrayProber.cnvProberDocker": "String (optional, default = \"us.gcr.io/tag-public/cnv-array-prober:0.0.0\")",
+  "cnvArrayProber.cnvBedFile": "File",
+  "cnvArrayProber.CytoSNP850K_Support_Csv": "File",
+  "cnvArrayProber.sampleName": "String",
+  "cnvArrayProber.GDA_Support_Csv": "File",
+  "cnvArrayProber.cnvArrayProber.disk": "Int (optional, default = 100)",
+  "cnvArrayProber.cnvArrayProber.memory": "Int (optional, default = 4)"
+}
+
diff --git a/CNV_Array_Prober/cnvArrayProber.wdl b/CNV_Array_Prober/cnvArrayProber.wdl
new file mode 100644
index 0000000..ba07f38
--- /dev/null
+++ b/CNV_Array_Prober/cnvArrayProber.wdl
@@ -0,0 +1,63 @@
+version 1.0
+
+workflow cnvArrayProber {
+    input{
+        String sampleName
+        File cnvBedFile
+        File CytoSNP850K_Support_Csv
+        File GDA_Support_Csv
+        String cnvProberDocker = "us.gcr.io/tag-public/cnv-array-prober:0.0.0"
+    }
+    call cnvArrayProber {
+        input:
+            sampleName = sampleName,
+            cnvBedFile = cnvBedFile,
+            CytoSNP850K_Support_Csv = CytoSNP850K_Support_Csv,
+            GDA_Support_Csv = GDA_Support_Csv,
+            cnvProberDocker = cnvProberDocker
+    }
+    output{
+        File cnvProbeAnnotation = cnvArrayProber.cnvProbeAnnotation
+        File cnvProbePlots = cnvArrayProber.cnvProbePlots
+    }
+    meta {
+        author: "Yueyao Gao"
+        email: "tag@broadinstitute.org"
+        description: "This workflow takes a CNV bed file and CytoSNP-850K and GDA support files as input and outputs a csv file with probe information for each CNV interval. Additionally,  output a PDF file with plots for each CNV interval the number of probes in the CytoSNP-850K and GDA arrays."
+    }
+}
+
+task cnvArrayProber {
+    input{
+        String sampleName
+        File cnvBedFile
+        File CytoSNP850K_Support_Csv
+        File GDA_Support_Csv
+        String cnvProberDocker
+        Int memory = 4
+        Int cpu = 1
+        Int disk = 100
+    }
+    command <<<
+    set -e
+    mkdir output
+
+    conda run --no-capture-output \
+            -n prober_env \
+            python3 /BaseImage/cnvArrayProber/scripts/cnvArrayProber.py \
+            -b ~{cnvBedFile} \
+            -c ~{CytoSNP850K_Support_Csv} \
+            -g ~{GDA_Support_Csv} \
+            -o output/~{sampleName}
+    >>>
+    output{
+        File cnvProbeAnnotation = "output/~{sampleName}CNV_Probe_Mappings.csv"
+        File cnvProbePlots = "output/~{sampleName}CNV_Probe_Mappings_Plots.pdf"
+    }
+    runtime {
+        docker: cnvProberDocker
+        memory: memory
+        cpu: cpu
+        disk: disk
+    }
+}
\ No newline at end of file