guard against empty readgroups (they exist by chance, because of downsampling)

Author: SHuang-Broad

wdl/pipelines/PacBio/Utility/ProcessOnDownsampledMultiReadgroupUBAM.wdl

@@ -2,6 +2,7 @@ version 1.0
 
 import "ProcessOnInstrumentDemuxedChunk.wdl" as TreatPerReadGroup
 
+import "../../../tasks/Utility/Utils.wdl" as U
 import "../../../tasks/Utility/BAMutils.wdl" as BU
 import "../../../tasks/Utility/GeneralUtils.wdl" as GU
 import "../../../tasks/Utility/Finalize.wdl" as FF
@@ -54,16 +55,16 @@ workflow ProcessOnDownsampledMultiReadgroupUBAM {
     output {
         String last_processing_date = today.yyyy_mm_dd
 
-        Array[File] readgroup_aligned_bams = eachRG.aligned_bam
-        Array[File] readgroup_aligned_bais = eachRG.aligned_bai
-        Array[File] readgroup_aligned_pbis = eachRG.aligned_pbi
-        Array[String] readgroup_meta_movies = eachRG.movie
-        Array[Float]  readgroup_metrics_contam = select_all(eachRG.contamination_est)
+        Array[File] readgroup_aligned_bams = select_all(eachRG.aligned_bam)
+        Array[File] readgroup_aligned_bais = select_all(eachRG.aligned_bai)
+        Array[File] readgroup_aligned_pbis = select_all(eachRG.aligned_pbi)
+        Array[String] readgroup_meta_movies = select_all(eachRG.movie)
+        Array[Float]  readgroup_metrics_contam = select_all(select_all(eachRG.contamination_est))
 
-        Array[Map[String, String]?] readgroup_metrics_optional_fingerprint_check          = eachRG.fingerprint_check
-        Array[Map[String, String]?] readgroup_metrics_optional_inferred_sex_info          = eachRG.inferred_sex_info
-        Array[Map[String, String]?] readgroup_metrics_optional_methyl_tag_simple_stats    = eachRG.methyl_tag_simple_stats
-        Array[Map[String, String]]  readgroup_metrics_optional_aBAM_metrics_files         = eachRG.aBAM_metrics_files
+        Array[Map[String, String]?] readgroup_metrics_optional_fingerprint_check          = select_all(eachRG.fingerprint_check)
+        Array[Map[String, String]?] readgroup_metrics_optional_inferred_sex_info          = select_all(eachRG.inferred_sex_info)
+        Array[Map[String, String]?] readgroup_metrics_optional_methyl_tag_simple_stats    = select_all(eachRG.methyl_tag_simple_stats)
+        Array[Map[String, String]]  readgroup_metrics_optional_aBAM_metrics_files         = select_all(eachRG.aBAM_metrics_files)
     }
 
     # split by read group
@@ -77,23 +78,31 @@ workflow ProcessOnDownsampledMultiReadgroupUBAM {
     # align by each read group
     Array[Pair[String, File]] readgroup_2_bam = zip(SplitByRG.rg_ids, SplitByRG.split_bam)
     scatter (pair in readgroup_2_bam) {
-        call TreatPerReadGroup.ProcessOnInstrumentDemuxedChunk as eachRG { input:
-            gcs_out_root_dir = gcs_out_root_dir,
+        call U.CountBamRecords as PreventEmptyBam { input:
+            bam = pair.right
+        }
+        if (defined(PreventEmptyBam.samools_error)) {
+            call U.StopWorkflow { input: reason = "samtools count error occurred" }
+        }
+        if (PreventEmptyBam.num_records != 0) {
+            call TreatPerReadGroup.ProcessOnInstrumentDemuxedChunk as eachRG { input:
+                gcs_out_root_dir = gcs_out_root_dir,
 
-            uBAM = pair.right,
+                uBAM = pair.right,
 
-            readgroup_id = pair.left,
-            bam_SM_field = bam_SM_field,
+                readgroup_id = pair.left,
+                bam_SM_field = bam_SM_field,
 
-            platform = 'Revio',  # non-critical lie, this parameter was used for setting bam size lower bound threshold for failing QC
+                platform = 'Revio',  # non-critical lie, this parameter was used for setting bam size lower bound threshold for failing QC
 
-            # args for optional QC subworkflows
-            qc_metrics_config_json = qc_metrics_config_json,
-            fingerprint_sample_id = fingerprint_sample_id,
-            expected_sex_type = expected_sex_type,
+                # args for optional QC subworkflows
+                qc_metrics_config_json = qc_metrics_config_json,
+                fingerprint_sample_id = fingerprint_sample_id,
+                expected_sex_type = expected_sex_type,
 
-            ref_map_file = ref_map_file,
-            disk_type = disk_type
+                ref_map_file = ref_map_file,
+                disk_type = disk_type
+            }
         }
     }