SRJointCallGVCFsWithGenomicsDB.wdl now prefers dbsnp_vcf to `know…

…n_sites_vcf` for genotyping (#480)

- Added ResolveMapKeysInPriorityOrder to resolve keys in a WDL Map object in priority order. This enables fallbacks / preferences for different entries in a map (such as the refmap).
- For genotyping with the gnarly genotyper, the dbsnp resource now prefers dbsnp_vcf from the RefMap. If this does not exist, it will fall back to known_sites_vcf. This allows different resources to be used for BQSR (which prefers known_sites_vcf in SRFlowcell) and the gnarly genotyper, which now prefers dbsnp_vcf.

wdl/pipelines/TechAgnostic/VariantCalling/SRJointCallGVCFsWithGenomicsDB.wdl

@@ -115,6 +115,14 @@ workflow SRJointCallGVCFsWithGenomicsDB {
 
     Map[String, String] ref_map = read_map(ref_map_file)
 
+    # Resolve the db_snp_vcf file, with preference to the db_snp_vcf file if it exists:
+    call UTILS.ResolveMapKeysInPriorityOrder as ResolveMapKeysInPriorityOrder {
+        input:
+            map = ref_map,
+            keys = ["test_bad_key_should_not_be_found", "dbsnp_vcf", "known_sites_vcf"]
+    }
+    File db_snp_vcf = ref_map[ResolveMapKeysInPriorityOrder.key]
+
     # Create sample-name map:
     call SRJOINT.CreateSampleNameMap as CreateSampleNameMap {
         input:
@@ -178,7 +186,7 @@ workflow SRJointCallGVCFsWithGenomicsDB {
                     ref_fasta       = ref_map['fasta'],
                     ref_fasta_fai   = ref_map['fai'],
                     ref_dict        = ref_map['dict'],
-                    dbsnp_vcf       = ref_map["known_sites_vcf"],
+                    dbsnp_vcf       = db_snp_vcf,
                     prefix          = prefix + "." + interval_name + ".gnarly_genotyper.raw",
                     heterozygosity = heterozygosity,
                     heterozygosity_stdev = heterozygosity_stdev,
@@ -194,7 +202,7 @@ workflow SRJointCallGVCFsWithGenomicsDB {
                     ref_fasta       = ref_map['fasta'],
                     ref_fasta_fai   = ref_map['fai'],
                     ref_dict        = ref_map['dict'],
-                    dbsnp_vcf       = ref_map["known_sites_vcf"],
+                    dbsnp_vcf       = db_snp_vcf,
                     prefix          = prefix + "." + interval_name + ".genotype_gvcfs.raw",
                     heterozygosity = heterozygosity,
                     heterozygosity_stdev = heterozygosity_stdev,

wdl/tasks/Utility/Utils.wdl

@@ -2013,3 +2013,50 @@ task SplitContigToIntervals {
         docker:                 select_first([runtime_attr.docker,            default_attr.docker])
     }
 }
+
+task ResolveMapKeysInPriorityOrder {
+    meta {
+        description: "Gets the first key in the map that exists.  If no keys exist, returns an empty string."
+    }
+
+    parameter_meta {
+        map: "Map[String, String] to resolve." 
+        keys: "Array[String] of keys to check in order of priority"
+    }
+
+    input {
+        Map[String, String] map
+        Array[String] keys
+    }
+
+    String out_file = "key.txt"
+
+    command <<<
+        touch ~{out_file}
+
+        f=~{write_map(map)}
+        awk '{print $1}' < ${f} > keys_in_map.txt
+        while read key ; do 
+            grep -q "^${key}$" keys_in_map.txt
+            if [[ $? -eq 0 ]] ; then
+                echo "${key}" > ~{out_file}
+                exit 0
+            fi
+        done < ~{write_lines(keys)}
+    >>>
+
+    output {
+        String key = read_string(out_file)
+    }
+
+    ###################
+    runtime {
+        cpu: 1
+        memory:  "512 MiB"
+        disks: "local-disk 50 HDD"
+        bootDiskSizeGb: 25
+        preemptible:     3
+        maxRetries:      2
+        docker:"gcr.io/cloud-marketplace/google/ubuntu2004:latest"
+    }
+}