Execute subsampling before QC

cloudgene.yaml

@@ -109,12 +109,14 @@ workflow:
         off: Off
 
     - id: subsampling
-      description: Subsample to 2000x
+      description: Coverage Subsampling
       type: number
       visible: true 
+      default: off
       values:
-        on: On
-        off: Off 
+        off: No subsampling
+        on: Subsample to 2000x 
+
 
     - id: myseparator0
       type: separator

modules/local/subsampling.nf

@@ -16,12 +16,16 @@ process SUBSAMPLING {
     # convert to integer
     mean_cov_int=\$(printf "%.0f" "\$mean_cov")
     
-    # set seed to 1 (for reproducbility) and FRAC to coverage/mean_cov (e.g. 1.2)
-    fraction=\$(echo "scale=4; 1+(${coverage} / \${mean_cov})" | bc)
+    fraction=\$(echo "scale=4; ${coverage} / \${mean_cov}" | bc)
     
     if [ \${mean_cov_int} -gt ${coverage} ]
     then
-        samtools view -s \$fraction -b -o ${bam_file.baseName}.subsampled.bam ${bam_file}
+        samtools view \
+        --subsample-seed 1 \
+        --subsample \$fraction \
+        -b \
+        -o ${bam_file.baseName}.subsampled.bam ${bam_file}
+        
         mv ${bam_file.baseName}.subsampled.bam ${bam_file}
     fi 
     """

nextflow.config

@@ -23,7 +23,7 @@ params {
     alignQ                 = 30
     coverage_estimation    = "on"
     max_samples            = 0
-    subsampling            = "on"
+    subsampling            = "off"
     subsampling_coverage   = 2000
 
 

workflows/mtdna_server_2.nf

@@ -62,6 +62,14 @@ workflow MTDNA_SERVER_2 {
         ref_file_mutect2
     )
 
+    if(params.subsampling.equals("on") ) {
+        SUBSAMPLING (
+            bams_ch,
+            params.subsampling_coverage
+        )
+        bams_ch = SUBSAMPLING.out.subsampled_bam_ch
+    }
+
     CALCULATE_STATISTICS(
         bams_ch
     )
@@ -84,14 +92,6 @@ workflow MTDNA_SERVER_2 {
 
     validated_files = INPUT_VALIDATION.out.validated_files.flatten()
 
-    if(params.subsampling.equals("on") ) {
-        SUBSAMPLING (
-            validated_files,
-            params.subsampling_coverage
-        )
-        validated_files = SUBSAMPLING.out.subsampled_bam_ch
-    }
-
     if (params.mode == 'mutserve') {
 
         MUTSERVE(