add check for duplicate cram paths

Author: mmorgantaylor

pipelines/wdl/glimpse/low_pass_imputation/input_qc/LowPassImputationQC.wdl

@@ -192,6 +192,16 @@ task ValidateCramsAndIndices {
             echo "All CRAM index files have the correct .crai extension."
         fi
 
+        # validate that cram paths are unique
+        unique_crams=$(cat crams_list.txt | sort -u | wc -l)
+        if [ $unique_crams -ne ~{num_crams} ]; then
+            # find duplicate CRAM paths
+            duplicate_crams=$(cat crams_list.txt | sort | uniq -d | paste -sd, | sed 's/,/, /g')
+            echo "Duplicate CRAM paths found: ${duplicate_crams}" >> qc_messages.txt
+        else
+            echo "CRAM paths are unique."
+        fi
+
         # ensure that all CRAM files are less than the maximum file size allowed by the service (currently 10GB)
         # this also serves as an access check, which should already have been performed by the service
         crams_exceeding_max_size=$(cat crams_list.txt | while read cram; do

pipelines/wdl/glimpse/low_pass_imputation/input_qc/test_inputs/Plumbing/fail_cram_duplicate_cram_paths.json

@@ -0,0 +1,11 @@
+{
+  "LowPassImputationQC.reference_panel_prefix": "gs://broad-dsde-methods-bge-resources-public/GlimpseImputation/ReferencePanels/1000G_HGDP_with_trio_information_old_chrX",
+  "LowPassImputationQC.contigs": ["chr1","chr2","chr3","chr4","chr5","chr6","chr7","chr8","chr9","chr10","chr11","chr12","chr13","chr14","chr15","chr16","chr17","chr18","chr19","chr20","chr21","chr22"],
+  "LowPassImputationQC.fasta": "gs://gcp-public-data--broad-references/hg38/v0/Homo_sapiens_assembly38.fasta",
+  "LowPassImputationQC.fasta_index": "gs://gcp-public-data--broad-references/hg38/v0/Homo_sapiens_assembly38.fasta.fai",
+  "LowPassImputationQC.output_basename": "plumbing_test",
+  "LowPassImputationQC.ref_dict": "gs://gcp-public-data--broad-references/hg38/v0/Homo_sapiens_assembly38.dict",
+  "LowPassImputationQC.crams": ["gs://fc-cddd72b5-323c-495c-9557-5057fff0275a/morgan_test/fakeCram0.cram", "gs://fc-cddd72b5-323c-495c-9557-5057fff0275a/morgan_test/fakeCram0.cram"],
+  "LowPassImputationQC.cram_indices": ["gs://fc-cddd72b5-323c-495c-9557-5057fff0275a/morgan_test/fakeCram0.cram.crai", "gs://fc-cddd72b5-323c-495c-9557-5057fff0275a/morgan_test/fakeCram0.cram.crai"],
+  "LowPassImputationQC.sample_ids": ["sample0", "sample1"]
+}