Add rsq calculation as a workflow

Author: seppinho

.github/workflows/ci-tests.yml

@@ -25,7 +25,7 @@ jobs:
 
       - name: Install nf-test
         run: |
-          wget -qO- get.nf-test.com | bash  -s 0.9.0-rc2
+          wget -qO- get.nf-test.com | bash -s 0.9.0-rc2
           sudo mv nf-test /usr/local/bin/
     
       - name: Set up Docker Buildx

README.md

@@ -11,8 +11,15 @@ nextflow run main.nf -profile test --imputation_token <token>
 ```
 
 ## Run Chip Simulation
+This command simulates array data from sequencing data. For each strand file included, an array will be created. 
 ```
-nextflow run workflows/chip_simulation.nf -c tests/mtdna.config -profile development
+nextflow run main.nf -c tests/simulate_hg19.config -profile development
+```
+
+## Run RSQ Calculation
+This command calculates the RSQ between sequencing data and imputed data. 
+```
+nextflow run main.nf -c tests/r2_calcuation.config -profile development
 ```
 
 ## Run Tests

conf/mtdna.config

@@ -1,23 +1,39 @@
-#!/usr/bin/env nextflow
 /*
 ========================================================================================
-    seppinho/microarray-eval-nf
+    genepi/microarray-eval-nf
 ========================================================================================
-    Github : https://github.com/seppinho/microarray-eval-nf
+    Github : https://github.com/genepi/microarray-eval-nf
     Author: Sebastian Schönherr, Lukas Forer, Martin Eberle
     ---------------------------
 */
 
 nextflow.enable.dsl = 2
 
-switch (params.workflow_name) {
-    case 'simulate':
-        include { ARRAY_SIMULATION } from './workflows/chip_simulation'
-        workflow {
-            ARRAY_SIMULATION ()
-        }
-        break
-    default:
-        error "Unknown workflow: ${params.workflow_name}"
-        break
+if(params.outdir == null) {
+  params.pubDir = "output/${params.project}"
+} else {
+  params.pubDir = params.outdir
+}
+
+include { SIMULATION } from './workflows/simulation'
+include { RSQ } from './workflows/rsq'
+
+workflow {
+    
+    switch (params.workflow_name) {
+
+        case 'simulate':
+            SIMULATION ()
+            break
+
+        case 'r2':
+             RSQ ()
+             break
+
+        default:
+             error "Unknown workflow: ${params.workflow_name}"
+             break
+
+    }
+    
 }

main.nf

@@ -0,0 +1,15 @@
+process CALCULATE_RSQ {
+
+    publishDir "${params.pubDir}/aggRSquare", mode: 'copy', pattern: '*aggRSquare'
+
+    input:
+    tuple val(chr), path(dosage_data), path(sequence_data)
+
+    output:
+    path("*aggRSquare"), emit: agg_rsquare
+
+
+    """
+    aggRSquare -v ${sequence_data} -i ${dosage_data} -o ${sequence_data.baseName} --d
+    """
+}

modules/local/calculate_imputation_accuracy.nf

@@ -1,6 +1,6 @@
 process PREPARE_RSQ_BROWSER_DATA {
 
-publishDir "${params.outdir}/${array_name}", mode: 'copy'
+publishDir "${params.pubDir}/${array_name}", mode: 'copy'
 
     input:
     tuple val(array_name), path(array)

modules/local/calculate_rsq.nf

@@ -1,5 +1,7 @@
 process SIMULATE_ARRAY {
 
+    publishDir "${params.pubDir}/simulation", mode: 'copy', pattern: '*vcf.gz'
+
     input:
     tuple val(chr), path(strand_data), path(sequence_data)
     path sample_file

modules/local/prepare_rsq_browser_data.nf

@@ -0,0 +1,6 @@
+params {
+    project       = "calc_rsq_hg19"
+    workflow_name = "r2"
+    imputed_data  = "$projectDir/tests/data/imputed_data/hg19/*vcf.gz"
+    sequence_data = "$projectDir/tests/data/sequence_data/hg19/*vcf.gz"
+}