update ONT post-alignment QC optionality

Author: SHuang-Broad

wdl/pipelines/ONT/Preprocessing/ONTFlowcellWGSuBAM.wdl

@@ -83,7 +83,7 @@ workflow ONTFlowcellWGSuBAM {
         Array[String] uBAM_tags_to_preserve = ['MM', 'ML']
 
         # args for optional QC subworkflows
-        File? qc_metrics_config_json
+        File qc_metrics_config_json
         String? fingerprint_sample_id
         String? expected_sex_type
 
@@ -102,9 +102,10 @@ workflow ONTFlowcellWGSuBAM {
         File aligned_bai = FinalizeAlignedBai.gcs_path
 
         Float wgs_cov                                   = QCandMetrics.wgs_cov
-        Map[String, Float] nanoplot_summ                = QCandMetrics.nanoplot_summ
         Map[String, Float] sam_flag_stats               = QCandMetrics.sam_flag_stats
 
+        Map[String, Float]? nanoplot_summ               = QCandMetrics.nanoplot_summ
+
         Map[String, Float]? seqkit_stats = FASTQstats.stats
 
         Map[String, String]? read_len_summaries = DystPeaker.read_len_summaries
@@ -152,13 +153,15 @@ workflow ONTFlowcellWGSuBAM {
     ###################################################################################
     # QC
     String readgroup_id = bam_SM_ID + "." + flowcell
-    AlignedBamQCnMetricsConfig qcm_config = read_json(select_first([qc_metrics_config_json]))
+    AlignedBamQCnMetricsConfig qcm_config = read_json(qc_metrics_config_json)
     call QCMetrics.Work as QCandMetrics { input:
         bam = ALN.aligned_bam,
         bai = ALN.aligned_bai,
 
         tech = "ONT",
 
+        run_nanoplot = qcm_config.run_nanoplot,
+
         cov_bed            = qcm_config.cov_bed,
         cov_bed_descriptor = qcm_config.cov_bed_descriptor,
 

wdl/pipelines/TechAgnostic/Utility/AlignedBamQCandMetrics.wdl

@@ -80,6 +80,8 @@ workflow Work {
 
         String tech
 
+        Boolean run_nanoplot = true
+
         File?   cov_bed
         String? cov_bed_descriptor
 
@@ -100,9 +102,11 @@ workflow Work {
 
     output {
         Float wgs_cov                     = MosDepthWGS.wgs_cov
-        Map[String, Float] nanoplot_summ  = NanoPlotFromBam.stats_map
         Map[String, Float] sam_flag_stats = ParseFlagStatsJson.qc_pass_reads_SAM_flag_stats
 
+        # nanoplot
+        Map[String, Float]? nanoplot_summ = NanoPlotFromBam.stats_map
+
         # fingerprint
         Map[String, String]? fingerprint_check = fp_res
 
@@ -160,12 +164,14 @@ workflow Work {
 
     ################################
     # nanoplot
-    call NP.NanoPlotFromBam { input: bam = bam, bai = bai, disk_type = disk_type }
-    FinalizationManifestLine b = object
-                                 {files_to_save: flatten([[NanoPlotFromBam.stats], NanoPlotFromBam.plots]),
-                                  is_singleton_file: false,
-                                  destination: metrics_output_dir + "/nanoplot",
-                                  output_attribute_name: "nanoplot"}
+    if (run_nanoplot) {
+        call NP.NanoPlotFromBam { input: bam = bam, bai = bai, disk_type = disk_type }
+        FinalizationManifestLine b = object
+                                    {files_to_save: flatten([[NanoPlotFromBam.stats], NanoPlotFromBam.plots]),
+                                    is_singleton_file: false,
+                                    destination: metrics_output_dir + "/nanoplot",
+                                    output_attribute_name: "nanoplot"}
+    }
 
     ###################################################################################
     # OPTIONAL QC/METRICS
@@ -255,6 +261,8 @@ struct AlignedBamQCnMetricsConfig {
     File?   cov_bed                         # An optional BED file on which coverage will be collected (a mean value for each interval)
     String? cov_bed_descriptor              # A short description of the BED provided for targeted coverage estimation; will be used in naming output files.
 
+    Boolean? run_nanoplot                   # if false or omitted, will NOT run nanoplot, which typically takes a long time to run
+
     String? fingerprint_vcf_store           # A GCS 'folder' holding fingerprint VCF files
 
     File?   vbid2_config_json               # A config json to for running the VBID2 contamination estimation sub-workflow;

wdl/pipelines/TechAgnostic/Utility/SaveFilesToDestination.wdl

@@ -92,15 +92,17 @@ workflow SaveFilestoDestination {
     call GU.CoerceArrayOfPairsToMap as collect { input: keys = attr, values = final_path }
 
     if (defined(already_finalized)) {
-        scatter(mm in select_first([already_finalized])) {
-            call GU.MapToTsv { input: m = mm}
-        }
-        call GU.ConcatenateFiles { input: af = MapToTsv.tsv, out_name = "does_not_matter.tsv" }
+        if (0!=length(select_all([already_finalized]))) {
+            scatter(mm in select_first([already_finalized])) {
+                call GU.MapToTsv { input: m = mm}
+            }
+            call GU.ConcatenateFiles { input: af = MapToTsv.tsv, out_name = "does_not_matter.tsv" }
 
-        Map[String, String] pack_one = read_map(ConcatenateFiles.merged)
+            Map[String, String] pack_one = read_map(ConcatenateFiles.merged)
 
-        call GU.MergeMaps as PackAllSavings { input:
-            one = pack_one, two = collect.output_map
+            call GU.MergeMaps as PackAllSavings { input:
+                one = pack_one, two = collect.output_map
+            }
         }
     }
 }