Variant search config. (#995)

* Attribute isRepeated flag. ValidateDataTypes indexing job. * CreateEntityMainn indexing job. * WriteEntityLevelDisplayHints indexing job. * WriteTextSearchField indexing job. * Select repeated attr field. Result parsing test list query. * Fetch string-only hints. Result parsing test hint query. * Group by repeated attribute field. Sql building and results parsing tests count query. * Filter on repeated attribute field. Sql building test list query. * Only use FLATTENED alias for repeated attribute field. * OpenAPI repeated type in ValueDisplay. Bug in returning hint query results. * Remove variant config files. * Regenerate docs. * Check actual data rows in result parsing hint & count query tests. * Fix overlapping conditional in eldh indexing job. * Variant search config -- first pass. * Use t_item_count in criteria selector. * Put back rest of underlay. * Add multi-select for variant entity. * Revert whitespace changes. * Always populate repeatedValue property of ValueDisplay in OpenAPI. * Move variant config to testonly version of aouSC underlay. * Move variant config files into separate directories. --------- Co-authored-by: Dexter Amundsen <dexamundsen@verily.com>
DataBiosphere · Sep 13, 2024 · 3b9ed7a · 3b9ed7a
1 parent 6401fc0
commit 3b9ed7a
Show file tree

Hide file tree

Showing 10 changed files with 475 additions and 2 deletions.
diff --git a/...n/resources/config/criteria/aouSC2023Q3R2_testonly/criteriaselector/variant/selector.json b/...n/resources/config/criteria/aouSC2023Q3R2_testonly/criteriaselector/variant/selector.json
@@ -0,0 +1,15 @@
+{
+  "name": "tanagra-variant",
+  "displayName": "SNP/Indel Variant",
+  "isEnabledForCohorts": true,
+  "isEnabledForDataFeatureSets": false,
+  "display": {
+    "category": "Genomics",
+    "tags": null
+  },
+  "filterBuilder": "core.EntityGroupFilterBuilder",
+  "plugin": "entityGroup",
+  "pluginConfig": null,
+  "pluginConfigFile": "variant.json",
+  "modifiers": null
+}
diff --git a/...in/resources/config/criteria/aouSC2023Q3R2_testonly/criteriaselector/variant/variant.json b/...in/resources/config/criteria/aouSC2023Q3R2_testonly/criteriaselector/variant/variant.json
@@ -0,0 +1,55 @@
+{
+  "columns": [
+    {
+      "key": "variant_id",
+      "widthString": "20%",
+      "title": "Variant id"
+    },
+    {
+      "key": "gene",
+      "widthDouble": 100,
+      "title": "Gene"
+    },
+    {
+      "key": "rs_number",
+      "widthDouble": 100,
+      "title": "RS Number"
+    },
+    {
+      "key": "genomic_location",
+      "widthDouble": 100,
+      "title": "Genomic Location"
+    },
+    {
+      "key": "protein_change",
+      "widthDouble": 100,
+      "title": "Protein Change"
+    },
+    {
+      "key": "allele_count",
+      "widthDouble": 100,
+      "title": "Allele Count"
+    },
+    {
+      "key": "allele_number",
+      "widthDouble": 100,
+      "title": "Allele Number"
+    },
+    {
+      "key": "allele_frequency",
+      "widthDouble": 100,
+      "title": "Allele Frequency"
+    },
+    {
+      "key": "t_item_count",
+      "widthDouble": 150,
+      "title": "Participant count"
+    }
+  ],
+  "classificationEntityGroups": [
+    {
+      "id": "variantPerson"
+    }
+  ],
+  "multiSelect": true
+}
diff --git a/underlay/src/main/resources/config/datamapping/aouCT_testonly/entity/variant/all.sql b/underlay/src/main/resources/config/datamapping/aouCT_testonly/entity/variant/all.sql
@@ -0,0 +1,12 @@
+SELECT ROW_NUMBER() OVER (ORDER BY vid) AS row_num,
+       vid,
+       gene_symbol,
+       dbsnp_rsid,
+       genomic_location,
+       consequence,
+       aa_change,
+       clinvar_classification,
+       gvs_all_ac,
+       gvs_all_an,
+       gvs_all_af
+FROM `${omopDataset}.prep_vat`
diff --git a/underlay/src/main/resources/config/datamapping/aouCT_testonly/entity/variant/entity.json b/underlay/src/main/resources/config/datamapping/aouCT_testonly/entity/variant/entity.json
@@ -0,0 +1,22 @@
+{
+  "name": "variant",
+  "allInstancesSqlFile": "all.sql",
+  "attributes": [
+    { "name": "id", "dataType": "INT64", "valueFieldName": "row_num" },
+    { "name": "variant_id", "dataType": "STRING", "valueFieldName": "vid" },
+    { "name": "gene", "dataType": "STRING", "valueFieldName": "gene_symbol", "isComputeDisplayHint": true },
+    { "name": "rs_number", "dataType": "STRING", "isDataTypeRepeated": true, "valueFieldName": "dbsnp_rsid" },
+    { "name": "genomic_location", "dataType": "STRING" },
+    { "name": "consequence", "dataType": "STRING", "isDataTypeRepeated":  true, "isComputeDisplayHint": true },
+    { "name": "protein_change", "dataType": "STRING", "valueFieldName": "aa_change" },
+    { "name": "clinvar_significance", "dataType": "STRING", "isDataTypeRepeated":  true, "valueFieldName": "clinvar_classification", "isComputeDisplayHint": true },
+    { "name": "allele_count", "dataType": "INT64", "valueFieldName": "gvs_all_ac", "isComputeDisplayHint": true },
+    { "name": "allele_number", "dataType": "INT64", "valueFieldName": "gvs_all_an", "isComputeDisplayHint": true },
+    { "name": "allele_frequency", "dataType": "DOUBLE", "valueFieldName": "gvs_all_af", "isComputeDisplayHint": true }
+  ],
+  "idAttribute": "id",
+  "textSearch": {
+    "attributes": [ "variant_id", "gene", "rs_number", "genomic_location" ]
+  },
+  "optimizeGroupByAttributes": [ "variant_id" ]
+}
diff --git a/...in/resources/config/datamapping/aouCT_testonly/entitygroup/variantPerson/entityGroup.json b/...in/resources/config/datamapping/aouCT_testonly/entitygroup/variantPerson/entityGroup.json
@@ -0,0 +1,8 @@
+{
+  "name": "variantPerson",
+  "groupEntity": "variant",
+  "itemsEntity": "person",
+  "idPairsSqlFile":  "idPairs.sql",
+  "groupEntityIdFieldName": "variant_row_num",
+  "itemsEntityIdFieldName": "flattened_person_id"
+}
diff --git a/...rc/main/resources/config/datamapping/aouCT_testonly/entitygroup/variantPerson/idPairs.sql b/...rc/main/resources/config/datamapping/aouCT_testonly/entitygroup/variantPerson/idPairs.sql
@@ -0,0 +1,6 @@
+SELECT v.variant_row_num, flattened_person_id
+FROM `${omopDataset}.variant_to_person` AS vtop
+JOIN
+  (SELECT ROW_NUMBER() OVER (ORDER BY vid) AS variant_row_num, vid FROM `${omopDataset}.prep_vat`)
+  AS v ON v.vid = vtop.vid
+CROSS JOIN UNNEST(vtop.person_ids) AS flattened_person_id
diff --git a/underlay/src/main/resources/config/indexer/aouSC2023Q3R2_verily.json b/underlay/src/main/resources/config/indexer/aouSC2023Q3R2_verily.json
@@ -1,5 +1,5 @@
 {
-  "underlay": "aouSC2023Q3R2",
+  "underlay": "aouSC2023Q3R2_testonly",
   "bigQuery": {
     "sourceData": {
       "projectId": "verily-tanagra-dev",

diff --git a/underlay/src/main/resources/config/service/aouSC2023Q3R2_verily.json b/underlay/src/main/resources/config/service/aouSC2023Q3R2_verily.json
@@ -1,5 +1,5 @@
 {
-  "underlay": "aouSC2023Q3R2",
+  "underlay": "aouSC2023Q3R2_testonly",
   "bigQuery": {
     "sourceData": {
       "projectId": "verily-tanagra-dev",

diff --git a/underlay/src/main/resources/config/underlay/aouSC2023Q3R2_testonly/ui.json b/underlay/src/main/resources/config/underlay/aouSC2023Q3R2_testonly/ui.json
@@ -0,0 +1,164 @@
+{
+  "featureConfig": {
+    "overrideExportButton": true
+  },
+  "demographicChartConfigs": {
+    "groupByAttributes": ["gender", "race", "age"],
+    "chartConfigs": [
+      {
+        "title": "Gender identity",
+        "primaryProperties": [{ "key": "gender" }]
+      },
+      {
+        "title": "Gender identity, Current age, Race",
+        "primaryProperties": [
+          { "key": "gender" },
+          {
+            "key": "age",
+            "buckets": [
+              {
+                "min": 18,
+                "max": 45,
+                "displayName": "18-44"
+              },
+              {
+                "min": 45,
+                "max": 65,
+                "displayName": "45-64"
+              },
+              {
+                "min": 65,
+                "displayName": "65+"
+              }
+            ]
+          }
+        ],
+        "stackedProperty": { "key": "race" }
+      }
+    ]
+  },
+  "criteriaSearchConfig": {
+    "criteriaTypeWidth": 120,
+    "columns": [
+      { "key": "name", "width": "100%", "title": "Name" },
+      { "key": "vocabulary_t_value", "width": 120, "title": "Vocab" },
+      { "key": "concept_code", "width": 120, "title": "Code" },
+      { "key": "t_rollup_count", "width": 150, "title": "Roll-up Count" }
+    ]
+  },
+  "cohortReviewConfig": {
+    "participantsListColumns": [
+      { "key": "id", "width": 35, "title": "Id" },
+      { "key": "age", "width": 25, "title": "Age" },
+      { "key": "gender", "width": 30, "title": "Gender" },
+      { "key": "sex_at_birth", "width": 40, "title": "Sex at Birth" },
+      { "key": "race", "width": 60, "title": "Race" },
+      { "key": "ethnicity", "width": 60, "title": "Ethnicity" }
+    ],
+    "attributes": [
+      {
+        "title": "Gender",
+        "key": "gender"
+      },
+      {
+        "title": "Sex at Birth",
+        "key": "sex_at_birth"
+      },
+      {
+        "title": "Age",
+        "key": "age"
+      },
+      {
+        "title": "Race",
+        "key": "race"
+      },
+      {
+        "title": "Ethnicity",
+        "key": "ethnicity"
+      }
+    ],
+    "pages": [
+      {
+        "type": "entityTable",
+        "id": "condition",
+        "title": "Conditions",
+        "plugin": {
+          "entity": "conditionOccurrence",
+          "columns": [
+            { "key": "condition_start_datetime", "width": "10%", "title": "Start date", "sortable": true },
+            { "key": "standard_concept_code", "width": "15%", "title": "Standard Code", "sortable": true, "filterable": true },
+            { "key": "standard_vocabulary", "width": "15%", "title": "Standard Vocabulary", "sortable": true, "filterable": true },
+            { "key": "standard_concept_name", "width": "30%", "title": "Condition name", "sortable": true, "filterable": true },
+            { "key": "age_at_occurrence", "width": "15%", "title": "Age at Occurrence", "sortable": true },
+            { "key": "visit_type", "width": "15%", "title": "Visit Type", "sortable": true }
+          ]
+        }
+      },
+      {
+        "type": "entityTable",
+        "id": "procedure",
+        "title": "Procedures",
+        "plugin": {
+          "entity": "procedureOccurrence",
+          "columns": [
+            { "key": "procedure_datetime", "width": "10%", "title": "Date", "sortable": true, "filterable": true },
+            { "key": "standard_concept_code", "width": "15%", "title": "Standard Code", "sortable": true, "filterable": true },
+            { "key": "standard_vocabulary", "width": "15%", "title": "Standard Vocabulary", "sortable": true, "filterable": true },
+            { "key": "standard_concept_name", "width": "30%", "title": "Procedure Name", "sortable": true, "filterable": true },
+            { "key": "age_at_occurrence", "width": "15%", "title": "Age at Occurrence", "sortable": true, "filterable": true },
+            { "key": "visit_type", "width": "15%", "title": "Visit Type", "sortable": true, "filterable": true }
+          ]
+        }
+      },
+      {
+        "type": "entityTable",
+        "id": "observation",
+        "title": "Observations",
+        "plugin": {
+          "entity": "observationOccurrence",
+          "columns": [
+            { "key": "observation_datetime", "width": "10%", "title": "Date", "sortable": true, "filterable": true },
+            { "key": "standard_concept_code", "width": "15%", "title": "Standard Code", "sortable": true, "filterable": true },
+            { "key": "standard_vocabulary", "width": "15%", "title": "Standard Vocabulary", "sortable": true, "filterable": true },
+            { "key": "standard_concept_name", "width": "30%", "title": "Observation Name", "sortable": true, "filterable": true },
+            { "key": "age_at_occurrence", "width": "15%", "title": "Age at Occurrence", "sortable": true, "filterable": true },
+            { "key": "visit_type", "width": "15%", "title": "Visit Type", "sortable": true, "filterable": true }
+          ]
+        }
+      },
+      {
+        "type": "entityTable",
+        "id": "ingredient",
+        "title": "Drugs",
+        "plugin": {
+          "entity": "ingredientOccurrence",
+          "columns": [
+            { "key": "drug_exposure_start_datetime", "width": "10%", "title": "Date", "sortable": true, "filterable": true },
+            { "key": "standard_concept_code", "width": "15%", "title": "Standard Code", "sortable": true, "filterable": true },
+            { "key": "standard_vocabulary", "width": "15%", "title": "Standard Vocabulary", "sortable": true, "filterable": true },
+            { "key": "standard_concept_name", "width": "30%", "title": "Drug Name", "sortable": true, "filterable": true },
+            { "key": "age_at_occurrence", "width": "15%", "title": "Age at Occurrence", "sortable": true, "filterable": true },
+            { "key": "visit_type", "width": "15%", "title": "Visit Type", "sortable": true, "filterable": true }
+          ]
+        }
+      },
+      {
+        "type": "entityTable",
+        "id": "measurements",
+        "title": "Labs and measurements",
+        "plugin": {
+          "entity": "measurementOccurrence",
+          "columns": [
+            { "key": "measurement_datetime", "width": "10%", "title": "Date", "sortable": true, "filterable": true },
+            { "key": "standard_concept_name", "width": "25%", "title": "Lab/measurement Name", "sortable": true, "filterable": true },
+            { "key": "value_enum", "width": "15%", "title": "Categorical value", "sortable": true, "filterable": true },
+            { "key": "value_numeric", "width": "15%", "title": "Numeric value", "sortable": true, "filterable": true },
+            { "key": "unit_concept_name", "width": "10%", "title": "Unit", "sortable": true, "filterable": true },
+            { "key": "age_at_occurrence", "width": "10%", "title": "Age at Occurrence", "sortable": true, "filterable": true },
+            { "key": "visit_type", "width": "15%", "title": "Visit Type", "sortable": true, "filterable": true }
+          ]
+        }
+      }
+    ]
+  }
+}