VS-1113 Update VQSR and VETS naming (#8948)

* Refactor:
VQSR Classic -> VQSR
VQSR Lite -> VETS
* Point to new truth

scripts/variantstore/repo/generate_git_filter_repo_command.sh

@@ -33,7 +33,7 @@ ah_var_store_branch_point() {
 files_added_on_ah_var_store() {
   # Look for files added to ah_var_store since the branch point. Note that these files were not necessarily *uniquely*
   # added to ah_var_store and might represent cherry picks from master (e.g. scripts and build files for the migration
-  # from Travis to GitHub Actions, VQSR Lite work, etc.)
+  # from Travis to GitHub Actions, VETS work, etc.)
   git diff "$(ah_var_store_branch_point)" $variants_branch --name-status | grep -E '^A' | cut -f 2-
 }
 
@@ -46,8 +46,8 @@ files_added_on_ah_var_store_not_on_master() {
 
 files_deleted_from_master() {
   # This intentionally does not use `git diff` as is used in `files_added_on_ah_var_store` since that would only show
-  # files deleted from the branch point to the head of master. There are numerous files here (mostly related to VQSR
-  # Lite) where files added to master after the branch point were cherry picked onto ah_var_store and subsequently
+  # files deleted from the branch point to the head of master. There are numerous files here (mostly related to VETS)
+  # where files added to master after the branch point were cherry picked onto ah_var_store and subsequently
   # deleted from master. This `git log` finds these while the `git diff` does not.
   #
   # https://waylonwalker.com/git-find-deleted-files/#git-log-diff-filter

scripts/variantstore/tieout/legacy_wdl/JointGenotyping.wdl

@@ -269,7 +269,7 @@ workflow JointGenotyping {
   }
 
   if (num_gvcfs <= snps_variant_recalibration_threshold) {
-    call Tasks.SNPsVariantRecalibrator as SNPsVariantRecalibratorClassic {
+    call Tasks.SNPsVariantRecalibrator {
       input:
         sites_only_variant_filtered_vcf = SitesOnlyGatherVcf.output_vcf,
         sites_only_variant_filtered_vcf_index = SitesOnlyGatherVcf.output_vcf_index,
@@ -300,9 +300,9 @@ workflow JointGenotyping {
         indels_recalibration = IndelsVariantRecalibrator.recalibration,
         indels_recalibration_index = IndelsVariantRecalibrator.recalibration_index,
         indels_tranches = IndelsVariantRecalibrator.tranches,
-        snps_recalibration = if defined(SNPsVariantRecalibratorScattered.recalibration) then select_first([SNPsVariantRecalibratorScattered.recalibration])[idx] else select_first([SNPsVariantRecalibratorClassic.recalibration]),
-        snps_recalibration_index = if defined(SNPsVariantRecalibratorScattered.recalibration_index) then select_first([SNPsVariantRecalibratorScattered.recalibration_index])[idx] else select_first([SNPsVariantRecalibratorClassic.recalibration_index]),
-        snps_tranches = select_first([SNPGatherTranches.tranches, SNPsVariantRecalibratorClassic.tranches]),
+        snps_recalibration = if defined(SNPsVariantRecalibratorScattered.recalibration) then select_first([SNPsVariantRecalibratorScattered.recalibration])[idx] else select_first([SNPsVariantRecalibrator.recalibration]),
+        snps_recalibration_index = if defined(SNPsVariantRecalibratorScattered.recalibration_index) then select_first([SNPsVariantRecalibratorScattered.recalibration_index])[idx] else select_first([SNPsVariantRecalibrator.recalibration_index]),
+        snps_tranches = select_first([SNPGatherTranches.tranches, SNPsVariantRecalibrator.tranches]),
         indel_filter_level = indel_filter_level,
         snp_filter_level = snp_filter_level,
         use_allele_specific_annotations = allele_specific_annotations,

scripts/variantstore/wdl/GvsCalculatePrecisionAndSensitivity.wdl

@@ -119,7 +119,7 @@ workflow GvsCalculatePrecisionAndSensitivity {
         variants_docker = effective_variants_docker,
     }
 
-    call IsVQSRLite {
+    call IsVETS {
       input:
         input_vcf = Add_AS_MAX_VQS_SCORE_ToVcf.output_vcf,
         basic_docker = effective_basic_docker,
@@ -153,7 +153,7 @@ workflow GvsCalculatePrecisionAndSensitivity {
         vcf_eval_bed_file = vcf_eval_bed_file,
         chromosomes = chromosomes,
         output_basename = sample_name + "-bq_roc_filtered",
-        is_vqsr_lite = IsVQSRLite.is_vqsr_lite,
+        is_vets = IsVETS.is_vets,
         ref_fasta = ref_fasta,
         real_time_genomics_docker = effective_real_time_genomics_docker,
     }
@@ -169,7 +169,7 @@ workflow GvsCalculatePrecisionAndSensitivity {
         chromosomes = chromosomes,
         all_records = true,
         output_basename = sample_name + "-bq_all",
-        is_vqsr_lite = IsVQSRLite.is_vqsr_lite,
+        is_vets = IsVETS.is_vets,
         ref_fasta = ref_fasta,
         real_time_genomics_docker = effective_real_time_genomics_docker,
     }
@@ -320,13 +320,13 @@ task Add_AS_MAX_VQS_SCORE_ToVcf {
   }
 }
 
-task IsVQSRLite {
+task IsVETS {
   input {
     File input_vcf
     String basic_docker
   }
 
-  String is_vqsr_lite_file = "is_vqsr_lite_file.txt"
+  String is_vets_file = "is_vets_file.txt"
 
   command <<<
     # Prepend date, time and pwd to xtrace log entries.
@@ -337,9 +337,9 @@ task IsVQSRLite {
     set +o errexit
     grep -v '^#' ~{input_vcf} | grep CALIBRATION_SENSITIVITY > /dev/null
     if [[ $? -eq 0 ]]; then
-      echo "true" > ~{is_vqsr_lite_file}
+      echo "true" > ~{is_vets_file}
     else
-      echo "false" > ~{is_vqsr_lite_file}
+      echo "false" > ~{is_vets_file}
     fi
     set -o errexit
   >>>
@@ -351,7 +351,7 @@ task IsVQSRLite {
     preemptible: 3
   }
   output {
-    Boolean is_vqsr_lite = read_boolean(is_vqsr_lite_file)
+    Boolean is_vets = read_boolean(is_vets_file)
   }
 }
 
@@ -406,15 +406,15 @@ task EvaluateVcf {
 
     String output_basename
 
-    Boolean is_vqsr_lite
+    Boolean is_vets
 
     String real_time_genomics_docker
     Int cpu = 1
     Int memory_mb = 3500
     Int disk_size_gb = ceil(2 * size(ref_fasta, "GiB")) + 500
   }
 
-  String max_score_field_tag = if (is_vqsr_lite == true) then 'MAX_CALIBRATION_SENSITIVITY' else 'MAX_AS_VQSLOD'
+  String max_score_field_tag = if (is_vets == true) then 'MAX_CALIBRATION_SENSITIVITY' else 'MAX_AS_VQSLOD'
 
   command <<<
     chromosomes=( ~{sep=' ' chromosomes} )
@@ -436,7 +436,7 @@ task EvaluateVcf {
       ~{if all_records then "--all-records" else ""} \
       --roc-subset snp,indel \
       --vcf-score-field=INFO.~{max_score_field_tag} \
-      ~{if is_vqsr_lite then "--sort-order ascending" else "--sort-order descending"} \
+      ~{if is_vets then "--sort-order ascending" else "--sort-order descending"} \
       -t human_REF_SDF \
       -b ~{truth_vcf} \
       -e ~{truth_bed}\

scripts/variantstore/wdl/GvsCreateFilterSet.wdl

@@ -1,8 +1,8 @@
 version 1.0
 
 import "GvsUtils.wdl" as Utils
-import "GvsVQSRClassic.wdl" as VQSRClassic
-import "../../vcf_site_level_filtering_wdl/JointVcfFiltering.wdl" as VQSRLite
+import "GvsVQSR.wdl" as VQSR
+import "../../vcf_site_level_filtering_wdl/JointVcfFiltering.wdl" as VETS
 
 workflow GvsCreateFilterSet {
   input {
@@ -21,16 +21,16 @@ workflow GvsCreateFilterSet {
     String? git_hash
     File? gatk_override
 
-    Boolean use_VQSR_lite = true
+    Boolean use_VETS = true
 
-    Int? INDEL_VQSR_CLASSIC_max_gaussians_override = 4
-    Int? INDEL_VQSR_CLASSIC_mem_gb_override
-    Int? SNP_VQSR_CLASSIC_max_gaussians_override = 6
-    Int? SNP_VQSR_CLASSIC_mem_gb_override
+    Int? INDEL_VQSR_max_gaussians_override = 4
+    Int? INDEL_VQSR_mem_gb_override
+    Int? SNP_VQSR_max_gaussians_override = 6
+    Int? SNP_VQSR_mem_gb_override
 
-    RuntimeAttributes? vqsr_lite_extract_runtime_attributes = {"command_mem_gb": 27}
-    RuntimeAttributes? vqsr_lite_train_runtime_attributes = {"command_mem_gb": 27}
-    RuntimeAttributes? vqsr_lite_score_runtime_attributes = {"command_mem_gb": 15}
+    RuntimeAttributes? vets_extract_runtime_attributes = {"command_mem_gb": 27}
+    RuntimeAttributes? vets_train_runtime_attributes = {"command_mem_gb": 27}
+    RuntimeAttributes? vets_score_runtime_attributes = {"command_mem_gb": 15}
 
     File? training_python_script
     File? scoring_python_script
@@ -127,10 +127,10 @@ workflow GvsCreateFilterSet {
       gatk_docker = effective_gatk_docker,
   }
 
-  # From this point, the paths diverge depending on whether they're using classic VQSR or VQSR-Lite
-  # The first branch here is VQSR-Lite, and the second is classic VQSR
-  if (use_VQSR_lite) {
-    call VQSRLite.JointVcfFiltering as JointVcfFiltering {
+  # From this point, the paths diverge depending on whether they're using VQSR or VETS
+  # The first branch here is VETS, and the second is VQSR
+  if (use_VETS) {
+    call VETS.JointVcfFiltering as JointVcfFiltering {
       input:
         input_vcfs = ExtractFilterTask.output_vcf,
         input_vcf_idxs = ExtractFilterTask.output_vcf_index,
@@ -141,9 +141,9 @@ workflow GvsCreateFilterSet {
         resource_args = "--resource:hapmap,training=true,calibration=true gs://gcp-public-data--broad-references/hg38/v0/hapmap_3.3.hg38.vcf.gz --resource:omni,training=true,calibration=true gs://gcp-public-data--broad-references/hg38/v0/1000G_omni2.5.hg38.vcf.gz --resource:1000G,training=true,calibration=false gs://gcp-public-data--broad-references/hg38/v0/1000G_phase1.snps.high_confidence.hg38.vcf.gz --resource:mills,training=true,calibration=true gs://gcp-public-data--broad-references/hg38/v0/Mills_and_1000G_gold_standard.indels.hg38.vcf.gz --resource:axiom,training=true,calibration=false gs://gcp-public-data--broad-references/hg38/v0/Axiom_Exome_Plus.genotypes.all_populations.poly.hg38.vcf.gz",
         extract_extra_args = "-L ${interval_list}",
         score_extra_args = "-L ${interval_list}",
-        extract_runtime_attributes = vqsr_lite_extract_runtime_attributes,
-        train_runtime_attributes = vqsr_lite_train_runtime_attributes,
-        score_runtime_attributes = vqsr_lite_score_runtime_attributes,
+        extract_runtime_attributes = vets_extract_runtime_attributes,
+        train_runtime_attributes = vets_train_runtime_attributes,
+        score_runtime_attributes = vets_score_runtime_attributes,
         gatk_docker = effective_gatk_docker,
         gatk_override = gatk_override,
         monitoring_script = "gs://gvs_quickstart_storage/cromwell_monitoring_script.sh",
@@ -198,12 +198,12 @@ workflow GvsCreateFilterSet {
         indel_recal_file = CreateFilteredScoredINDELsVCF.output_vcf,
         indel_recal_file_index = CreateFilteredScoredINDELsVCF.output_vcf_index,
         project_id = project_id,
-        useClassic = false
+        useVQSR = false
     }
   }
 
-  if (!use_VQSR_lite) {
-    call VQSRClassic.JointVcfFiltering as VQSRClassic {
+  if (!use_VETS) {
+    call VQSR.JointVcfFiltering as VQSR {
       input:
         git_branch_or_tag = git_branch_or_tag,
         git_hash = git_hash,
@@ -218,10 +218,10 @@ workflow GvsCreateFilterSet {
         sites_only_variant_filtered_vcf_idx = MergeVCFs.output_vcf_index,
         sites_only_variant_filtered_vcfs = ExtractFilterTask.output_vcf,
         sites_only_variant_filtered_vcf_idxs = ExtractFilterTask.output_vcf_index,
-        INDEL_VQSR_max_gaussians_override = INDEL_VQSR_CLASSIC_max_gaussians_override,
-        INDEL_VQSR_mem_gb_override = INDEL_VQSR_CLASSIC_mem_gb_override,
-        SNP_VQSR_max_gaussians_override = SNP_VQSR_CLASSIC_max_gaussians_override,
-        SNP_VQSR_mem_gb_override = SNP_VQSR_CLASSIC_mem_gb_override,
+        INDEL_VQSR_max_gaussians_override = INDEL_VQSR_max_gaussians_override,
+        INDEL_VQSR_mem_gb_override = INDEL_VQSR_mem_gb_override,
+        SNP_VQSR_max_gaussians_override = SNP_VQSR_max_gaussians_override,
+        SNP_VQSR_mem_gb_override = SNP_VQSR_mem_gb_override,
         gatk_docker = effective_gatk_docker,
         gatk_override = gatk_override,
     }
@@ -255,7 +255,7 @@ workflow GvsCreateFilterSet {
                [CreateFilteredScoredSNPsVCF.monitoring_log],
                [CreateFilteredScoredINDELsVCF.monitoring_log],
                [PopulateFilterSetInfo.monitoring_log],
-               select_first([VQSRClassic.monitoring_logs, []]),
+               select_first([VQSR.monitoring_logs, []]),
                [PopulateFilterSetSites.monitoring_log]
                ]
                )

scripts/variantstore/wdl/GvsCreateVATfromVDS.wdl

@@ -27,7 +27,6 @@ workflow GvsCreateVATfromVDS {
         Int? split_intervals_disk_size_override
         Int? split_intervals_mem_override
         Int? split_intervals_scatter_count
-        Boolean use_classic_VQSR = false
         Boolean use_reference_disk = true
 
         String? cloud_sdk_docker
@@ -144,7 +143,6 @@ workflow GvsCreateVATfromVDS {
             call GenerateSitesOnlyVcf {
                 input:
                     vds_path = select_first([vds_path]),
-                    use_classic_VQSR = use_classic_VQSR,
                     workspace_project = effective_google_project,
                     hail_version = effective_hail_version,
                     hail_wheel = hail_wheel,
@@ -310,7 +308,6 @@ workflow GvsCreateVATfromVDS {
 task GenerateSitesOnlyVcf {
     input {
         String vds_path
-        Boolean use_classic_VQSR
         String workspace_project
         String workspace_bucket
         String region

scripts/variantstore/wdl/GvsCreateVDS.wdl

@@ -19,7 +19,7 @@ workflow GvsCreateVDS {
         Int? cluster_max_age_minutes
         Boolean leave_cluster_running_at_end = false
         Float? master_memory_fraction
-        Boolean use_classic_VQSR = false
+        Boolean use_VQSR = false
 
         String? git_branch_or_tag
         String? hail_version
@@ -111,7 +111,7 @@ workflow GvsCreateVDS {
             prefix = cluster_prefix,
             vds_path = vds_destination_path,
             avro_path = avro_path,
-            use_classic_VQSR = use_classic_VQSR,
+            use_VQSR = use_VQSR,
             hail_version = effective_hail_version,
             hail_wheel = hail_wheel,
             hail_temp_path = hail_temp_path,
@@ -158,7 +158,7 @@ task CreateVds {
         String prefix
         String vds_path
         String avro_path
-        Boolean use_classic_VQSR
+        Boolean use_VQSR
         Boolean leave_cluster_running_at_end
         File hail_gvs_import_script
         File gvs_import_script
@@ -233,7 +233,7 @@ task CreateVds {
             "temp-path": "${hail_temp_path}",
             "avro-path": "~{avro_path}"
             ~{', "intermediate-resume-point": ' + intermediate_resume_point}
-            ~{true=', "use-classic-vqsr": ""' false='' use_classic_VQSR}
+            ~{true=', "use-vqsr": ""' false='' use_VQSR}
         }
         FIN
 

scripts/variantstore/wdl/GvsExtractAvroFilesForHail.wdl

@@ -12,7 +12,7 @@ workflow GvsExtractAvroFilesForHail {
         String dataset_name
         String filter_set_name
         String call_set_identifier
-        Boolean use_VQSR_lite = true
+        Boolean use_VETS = true
         Int scatter_width = 10
         String? basic_docker
         String? cloud_sdk_docker
@@ -43,7 +43,7 @@ workflow GvsExtractAvroFilesForHail {
             cloud_sdk_docker = effective_cloud_sdk_docker,
     }
 
-    call Utils.IsVQSRLite {
+    call Utils.IsVETS {
         input:
             project_id = project_id,
             fq_filter_set_info_table = "~{project_id}.~{dataset_name}.filter_set_info",
@@ -74,7 +74,7 @@ workflow GvsExtractAvroFilesForHail {
             filter_set_name = filter_set_name,
             avro_sibling = OutputPath.out,
             call_set_identifier = call_set_identifier,
-            is_vqsr_lite = IsVQSRLite.is_vqsr_lite,
+            is_vets = IsVETS.is_vets,
             variants_docker = effective_variants_docker,
     }
 
@@ -203,7 +203,7 @@ task ExtractFromSampleInfoTable {
 
 task ExtractFromFilterTables {
     meta {
-        description: "Extracts from the tables: filter_set_sites, filter_set_info/filter_set_info_vqsr_lite, and filter_set_tranches (if using VQSR Classic)"
+        description: "Extracts from the tables: filter_set_sites, filter_set_info, and filter_set_tranches (if using VQSR)"
         # Not dealing with caching for now as that would introduce a lot of complexity.
         volatile: true
     }
@@ -214,11 +214,11 @@ task ExtractFromFilterTables {
         String filter_set_name
         String avro_sibling
         String call_set_identifier
-        Boolean is_vqsr_lite = true
+        Boolean is_vets = true
         String variants_docker
     }
 
-    String vqs_score_field = if (is_vqsr_lite == true) then 'calibration_sensitivity' else 'vqslod'
+    String vqs_score_field = if (is_vets == true) then 'calibration_sensitivity' else 'vqslod'
 
     parameter_meta {
         avro_sibling: "Cloud path to a file that will be the sibling to the 'avro' 'directory' under which output Avro files will be written."
@@ -249,7 +249,7 @@ task ExtractFromFilterTables {
             ORDER BY location
         " --call_set_identifier ~{call_set_identifier} --dataset_name ~{dataset_name} --table_name filter_set_sites --project_id=~{project_id}
 
-        if [ ~{is_vqsr_lite} = false ]; then
+        if [ ~{is_vets} = false ]; then
             python3 /app/run_avro_query.py --sql "
                 EXPORT DATA OPTIONS(
                 uri='${avro_prefix}/vqsr_tranche_data/vqsr_tranche_data_*.avro', format='AVRO', compression='SNAPPY') AS