wip

main.nf

@@ -199,13 +199,21 @@ process MERGE_ALL_PHASED_VCF {
 process COPY_SKIPPED_VCFS {
     label 'copy_skipped_vcfs'
 
-    publishDir "${params.output}/final_vcf", mode: 'copy'
+    publishDir params.output, mode: 'copy'
 
     input:
-    file(skipped_files)
+    path(skipped_files)
+
+    output:
+    path('final_vcf/*'), emit: skipped_files
 
     script:
     """
+    mkdir final_vcf
+    for f in ${skipped_files}; do
+        mv \$f final_vcf/
+
+    done
     echo "Copying skipped VCF file: ${skipped_files}"
     """
 }
@@ -214,18 +222,17 @@ workflow {
     println "Welcome to ${params.service.name} (${workflow.manifest.version})"
 
     if (params.imputation.enabled) {
-        INPUT_VALIDATION()
-
-        // Copy skipped VCFs to the final output directory's subdirectory 'final_vcfs'
-        // COPY_SKIPPED_VCFS(
-        //     INPUT_VALIDATION.out.skipped_files
-        // )
+        def input_validation = INPUT_VALIDATION()
 
-        // Proceed with quality control on validated files
+        // Pass outputs to QUALITY_CONTROL
         QUALITY_CONTROL(
-            INPUT_VALIDATION.out.validated_files,
-            INPUT_VALIDATION.out.validation_report,
-            site_files_ch
+            input_validation.validated_files,
+            input_validation.validation_report,
+            site_files_ch.collect()
+        )
+
+        COPY_SKIPPED_VCFS(
+            input_validation.skipped_files.collect()
         )
 
         // Check if QC chunks exist in case QC failed
@@ -235,7 +242,6 @@ workflow {
 
         if (params.mode == 'imputation') {
             def phased_ch = QUALITY_CONTROL.out.qc_metafiles
-
             if (phasing_engine != 'no_phasing') {
                 PHASING(
                     QUALITY_CONTROL.out.qc_metafiles
@@ -249,16 +255,16 @@ workflow {
                     phased_ch
                 )
 
-                if (params.merge_results == true) {
+                if (params.merge_results) {
                     ENCRYPTION(
-                        IMPUTATION.out.groupTuple()
-                    )
+                    IMPUTATION.out.groupTuple()
+                )
                 }
-            } else {
-                if (params.merge_results == true) {
+        } else {
+                if (params.merge_results) {
                     MERGE_ALL_PHASED_VCF(
-                        phased_ch.groupTuple()
-                    )
+                    phased_ch.groupTuple()
+                )
                 }
             }
         }

modules/local/input_validation/input_validation_vcf.nf

@@ -9,8 +9,8 @@ process INPUT_VALIDATION_VCF {
     path vcf_files
 
     output:
-    path("split_vcfs/*.vcf.gz"), emit: validated_files
-    path("skipped_vcfs/*.vcf.gz"), emit: skipped_files, optional: true
+    path("validated_vcfs/*.vcf.gz"), emit: validated_files
+    path("skipped_vcfs/*.vcf.gz"), emit: skipped_files
     path("validation_report.txt"), emit: validation_report
 
     script:
@@ -42,7 +42,7 @@ EOF
     skipped_vcfs=()
 
     # Create the directories for split VCF files
-    mkdir -p split_vcfs skipped_vcfs
+    mkdir -p split_vcfs skipped_vcfs validated_vcfs
 
     # Process each VCF file
     for vcf in ${vcf_files}; do
@@ -159,23 +159,34 @@ EOF
         base=\$(basename "\$f")
         if [[ "\$base" =~ _([1-9]|1[0-9]|2[0-2]|X|chr([1-9]|1[0-9]|2[0-2]|X))\\.vcf\\.gz\$ ]]; then
             vcf_files_to_validate+=("\$f")
+            mv "\$f" validated_vcfs/
+
+            if [ -f "\$f.csi" ] || [ -f "\$f.tbi" ]; then
+                mv "\$f".csi validated_vcfs/
+                mv "\$f".tbi validated_vcfs/
+            fi
         else
             skipped_vcfs+=("\$f")
+            mv "\$f" skipped_vcfs/
+
+            if [ -f "\$f.csi" ] || [ -f "\$f.tbi" ]; then
+                mv "\$f".csi skipped_vcfs/
+                mv "\$f".tbi skipped_vcfs/
+            fi
         fi
     done
 
-    for f in "\${skipped_vcfs[@]}"; do
-        mv "\$f.*" skipped_vcfs/
-    done
+    final_skip_files=(skipped_vcfs/*.vcf.gz)
+
+    # get the final_vcf_files_to_validate by globbing
 
-    echo "VCF files to validate:"
-    printf '%s\\n' "\${vcf_files_to_validate[@]}"
+    final_vcf_files_to_validate=(validated_vcfs/*.vcf.gz)
 
-    echo "Skipped VCF files:"
-    printf '%s\\n' "\${skipped_vcfs[@]}"
+    echo "skipped_vcfs: \${final_skip_files[@]}"
+    echo "validated_vcfs: \${final_vcf_files_to_validate[@]}"
 
     # Run the validation program only if there are files to validate
-    if [ \${#vcf_files_to_validate[@]} -gt 0 ]; then
+    if [ \${#final_vcf_files_to_validate[@]} -gt 0 ]; then
         java -Xmx${avail_mem}M -jar /opt/imputationserver-utils/imputationserver-utils.jar \\
             validate \\
             --population ${params.population} \\
@@ -189,7 +200,7 @@ EOF
             --no-index \\
             --contactName "${contactName}" \\
             --contactEmail "${contactEmail}" \\
-            "\${vcf_files_to_validate[@]}"
+            "\${final_vcf_files_to_validate[@]}"
         exit_code_a=\$?
     else
         echo "No VCF files to validate."

modules/local/quality_control/quality_control_report.nf

@@ -1,5 +1,4 @@
 process QUALITY_CONTROL_REPORT {
-
     label 'preprocessing'
     publishDir params.output, mode: 'copy'
 
@@ -10,10 +9,11 @@ process QUALITY_CONTROL_REPORT {
     path(qc_report_file)
 
     output:
-    path("*.html")
+    path('*.html')
 
     script:
     """
+    echo "maf_file: ${maf_file}"
     Rscript -e "require( 'rmarkdown' ); render('${qc_report_file}',
         params = list(
             maf_file = '${maf_file}',
@@ -30,5 +30,4 @@ process QUALITY_CONTROL_REPORT {
         output_file='\$PWD/quality-control.html'
     )"
     """
-
 }

workflows/input_validation.nf

@@ -2,16 +2,14 @@ include { INPUT_VALIDATION_VCF } from '../modules/local/input_validation/input_v
 
 Channel
     .fromPath(params.files)
-    .set {files}
+    .set { files }
 
 workflow INPUT_VALIDATION {
-
     main:
     INPUT_VALIDATION_VCF(files.collect())
 
     emit:
     validated_files = INPUT_VALIDATION_VCF.out.validated_files.collect()
+    skipped_files = INPUT_VALIDATION_VCF.out.skipped_files.collect()
     validation_report = INPUT_VALIDATION_VCF.out.validation_report
 }
-
-