Cb/vda 1902 agd subject queue (#1112)

* VDA-1902 Adding new selector for agd_queue and indexer for new dataset * VDA-1902 Adding config files, and fixing agd_queue selector * VDA-1902 Adding new selector for agd_queue and indexer for new dataset * VDA-1902 Adding config files, and fixing agd_queue selector
DataBiosphere · Dec 31, 2024 · bc80134 · bc80134
1 parent a6027d4
commit bc80134
Show file tree

Hide file tree

Showing 9 changed files with 109 additions and 2 deletions.
diff --git a/...c/main/resources/config/criteria/sd/criteriaselector/agdSubjectQueue/agdSubjectQueue.json b/...c/main/resources/config/criteria/sd/criteriaselector/agdSubjectQueue/agdSubjectQueue.json
@@ -0,0 +1,3 @@
+{
+  "attribute": "has_agd_queue"
+}
diff --git a/...rlay/src/main/resources/config/criteria/sd/criteriaselector/agdSubjectQueue/selector.json b/...rlay/src/main/resources/config/criteria/sd/criteriaselector/agdSubjectQueue/selector.json
@@ -0,0 +1,15 @@
+{
+  "name": "agdSubjectQueue",
+  "displayName": "AGD Subject Queue",
+  "isEnabledForCohorts": true,
+  "isEnabledForDataFeatureSets": false,
+  "display": {
+    "category": "BioVU",
+    "tags": null
+  },
+  "filterBuilder": "core.PrimaryEntityFilterBuilder",
+  "plugin": "attribute",
+  "pluginConfig": null,
+  "pluginConfigFile": "agdSubjectQueue.json",
+  "modifiers": null
+}
diff --git a/underlay/src/main/resources/config/datamapping/sd/entity/person/all.sql b/underlay/src/main/resources/config/datamapping/sd/entity/person/all.sql
@@ -13,7 +13,9 @@ SELECT
     /* As a courtesy, convert string fields to boolean: 0 -> No, 1 -> Yes */
     CASE WHEN x.compromise_ind = '1' THEN true WHEN x.compromise_ind = '0' THEN false ELSE null END AS biovu_sample_is_compromised,
     CASE WHEN x.nonshippable_ind = '1' THEN true WHEN x.nonshippable_ind = '0' THEN false ELSE null END AS biovu_sample_is_nonshippable,
-    CASE WHEN x.plasma_ind = '1' THEN true WHEN x.plasma_ind = '0' THEN false ELSE null END AS biovu_sample_has_plasma
+    CASE WHEN x.plasma_ind = '1' THEN true WHEN x.plasma_ind = '0' THEN false ELSE null END AS biovu_sample_has_plasma,
+    EXISTS
+        (SELECT 1 FROM `${omopDataset}.x_agd_queue` aq WHERE p.person_id = aq.person_id) AS has_agd_queue
 
 FROM `${omopDataset}.person` p
 

diff --git a/underlay/src/main/resources/config/datamapping/sd/entity/person/entity.json b/underlay/src/main/resources/config/datamapping/sd/entity/person/entity.json
@@ -16,7 +16,8 @@
     { "name": "biovu_sample_dna_yield", "dataType": "INT64" },
     { "name": "biovu_sample_is_compromised", "dataType": "BOOLEAN" },
     { "name": "biovu_sample_is_nonshippable", "dataType": "BOOLEAN" },
-    { "name": "biovu_sample_has_plasma", "dataType": "BOOLEAN" }
+    { "name": "biovu_sample_has_plasma", "dataType": "BOOLEAN" },
+    { "name": "has_agd_queue", "dataType": "BOOLEAN" }
   ],
   "idAttribute": "id",
   "optimizeGroupByAttributes": [ "gender", "race", "age" ]

diff --git a/underlay/src/main/resources/config/indexer/sd/sd020240831.json b/underlay/src/main/resources/config/indexer/sd/sd020240831.json
@@ -0,0 +1,25 @@
+{
+  "underlay": "sd",
+  "bigQuery": {
+    "sourceData": {
+      "projectId": "vumc-sd-dev",
+      "datasetId": "sd_20240831",
+      "sqlSubstitutions": {
+        "omopDataset": "vumc-sd-dev.sd_20240831",
+        "staticTablesDataset": "vumc-sd-dev.sd_20240831"
+      }
+    },
+    "indexData": {
+      "projectId": "vumc-sd-dev",
+      "datasetId": "indexed_sd_20240831_1"
+    },
+    "queryProjectId": "vumc-sd-dev",
+    "dataLocation": "us-central1"
+  },
+  "dataflow": {
+    "serviceAccountEmail": "[email protected]",
+    "gcsTempDirectory": "gs://dataflow-indexing-sd-dev/temp/",
+    "workerMachineType": "n1-standard-4",
+    "usePublicIps": false
+  }
+}
diff --git a/underlay/src/main/resources/config/service/sd/sd_20240831_dev.json b/underlay/src/main/resources/config/service/sd/sd_20240831_dev.json
@@ -0,0 +1,20 @@
+{
+  "underlay": "sd",
+  "bigQuery": {
+    "sourceData": {
+      "projectId": "vumc-sd-dev",
+      "datasetId": "sd_20240831",
+      "sqlSubstitutions": {
+        "omopDataset": "vumc-sd-dev.sd_20240831",
+        "staticTablesDataset": "vumc-sd-dev.sd_20240831"
+      }
+    },
+    "indexData": {
+      "projectId": "vumc-sd-dev",
+      "datasetId": "indexed_sd_20240831"
+    },
+    "queryProjectId": "vumc-sd-dev",
+    "dataLocation": "us-central1"
+  },
+  "uiConfigFile": "ui.json"
+}
diff --git a/underlay/src/main/resources/config/service/sd/sd_20240831_prod.json b/underlay/src/main/resources/config/service/sd/sd_20240831_prod.json
@@ -0,0 +1,20 @@
+{
+  "underlay": "sd",
+  "bigQuery": {
+    "sourceData": {
+      "projectId": "vumc-sd-prod",
+      "datasetId": "sd_20240831",
+      "sqlSubstitutions": {
+        "omopDataset": "vumc-sd-prod.sd_20240831",
+        "staticTablesDataset": "vumc-sd-prod.sd_20240831"
+      }
+    },
+    "indexData": {
+      "projectId": "vumc-sd-prod",
+      "datasetId": "indexed_sd_20240831"
+    },
+    "queryProjectId": "vumc-sd-prod",
+    "dataLocation": "us-central1"
+  },
+  "uiConfigFile": "ui.json"
+}
diff --git a/underlay/src/main/resources/config/service/sd/sd_20240831_test.json b/underlay/src/main/resources/config/service/sd/sd_20240831_test.json
@@ -0,0 +1,20 @@
+{
+  "underlay": "sd",
+  "bigQuery": {
+    "sourceData": {
+      "projectId": "vumc-sd-test",
+      "datasetId": "sd_20240831",
+      "sqlSubstitutions": {
+        "omopDataset": "vumc-sd-test.sd_20240831",
+        "staticTablesDataset": "vumc-sd-test.sd_20240831"
+      }
+    },
+    "indexData": {
+      "projectId": "vumc-sd-test",
+      "datasetId": "indexed_sd_20240831"
+    },
+    "queryProjectId": "vumc-sd-test",
+    "dataLocation": "us-central1"
+  },
+  "uiConfigFile": "ui.json"
+}
diff --git a/underlay/src/main/resources/config/underlay/sd/underlay.json b/underlay/src/main/resources/config/underlay/sd/underlay.json
@@ -89,6 +89,7 @@
     "sd/age",
     "sd/bioVU",
     "sd/bioVUPlasma",
+    "sd/agdSubjectQueue",
     "sd/genotyping",
     "sd/measurement",
     "sd/documents",