update PacBio post-alignment QC optionality

Author: SHuang-Broad

wdl/pipelines/PacBio/Utility/ProcessOnInstrumentDemuxedChunk.wdl

@@ -31,7 +31,7 @@ workflow ProcessOnInstrumentDemuxedChunk {
         "output files will be copied over there"
 
         qc_metrics_config_json:
-        "A config json to for running the QC and metrics-collection sub-workflow 'AlignedBamQCandMetrics'"
+        "A config json to for running the QC and metrics-collection sub-workflow 'AlignedBamQCandMetrics'; could be an empty json file."
 
         fingerprint_sample_id:
         "For fingerprint verification: the ID of the sample supposedly this BAM belongs to; note that the fingerprint VCF is assumed to be located at {fingerprint_store}/{fingerprint_sample_id}*.vcf(.gz)?"
@@ -78,7 +78,7 @@ workflow ProcessOnInstrumentDemuxedChunk {
         String platform
 
         # args for optional QC subworkflows
-        File? qc_metrics_config_json
+        File qc_metrics_config_json
         String? fingerprint_sample_id
         String? expected_sex_type
 
@@ -97,8 +97,8 @@ workflow ProcessOnInstrumentDemuxedChunk {
 
         # metrics block (caveat: always has to keep an eye on the QC subworkflow about outputs)
         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
 
         # fingerprint
         Map[String, String]? fingerprint_check          = QCandMetrics.fingerprint_check
@@ -149,7 +149,7 @@ workflow ProcessOnInstrumentDemuxedChunk {
 
     ###################################################################################
     # QC
-    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 = aBAM,
         bai = aBAI,

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;