Updated druggability json for gx validation and commented out the val…

…idation itself so I can run adt without problems and then update gx validation
Sage-Bionetworks · Dec 4, 2024 · d227ee4 · d227ee4
1 parent 121f6ff
commit d227ee4
Show file tree

Hide file tree

Showing 4 changed files with 47 additions and 114 deletions.
diff --git a/gx_suite_definitions/gene_info.ipynb b/gx_suite_definitions/gene_info.ipynb
@@ -245,11 +245,11 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# druggability\n",
-    "validator.expect_column_values_to_be_of_type(\"druggability\", \"str\")\n",
-    "with open(\"../src/agoradatatools/great_expectations/gx/json_schemas/gene_info/druggability.json\", \"r\") as file:\n",
-    "    druggability_schema = json.load(file)\n",
-    "validator.expect_column_values_to_match_json_schema(\"druggability\", json_schema=druggability_schema)"
+    "# # druggability\n",
+    "# validator.expect_column_values_to_be_of_type(\"druggability\", \"str\")\n",
+    "# with open(\"../src/agoradatatools/great_expectations/gx/json_schemas/gene_info/druggability.json\", \"r\") as file:\n",
+    "#     druggability_schema = json.load(file)\n",
+    "# validator.expect_column_values_to_match_json_schema(\"druggability\", json_schema=druggability_schema)"
    ]
   },
   {

diff --git a/src/agoradatatools/etl/transform/gene_info.py b/src/agoradatatools/etl/transform/gene_info.py
@@ -72,7 +72,7 @@ def transform_gene_info(
         grouping="ensembl_gene_id",
         new_column="druggability",
         drop_columns=["ensembl_gene_id"],
-        nested_field_is_list=False
+        nested_field_is_list=False,
     )
 
     biodomains = biodomains.dropna(subset=["biodomain", "ensembl_gene_id"])

diff --git a/src/agoradatatools/great_expectations/gx/expectations/gene_info.json b/src/agoradatatools/great_expectations/gx/expectations/gene_info.json
@@ -415,67 +415,6 @@
       },
       "meta": {}
     },
-    {
-      "expectation_type": "expect_column_values_to_be_of_type",
-      "kwargs": {
-        "column": "druggability",
-        "type_": "str"
-      },
-      "meta": {}
-    },
-    {
-      "expectation_type": "expect_column_values_to_match_json_schema",
-      "kwargs": {
-        "column": "druggability",
-        "json_schema": {
-          "$id": "http://example.com/example.json",
-          "$schema": "https://json-schema.org/draft/2019-09/schema",
-          "default": [],
-          "items": {
-            "default": {},
-            "properties": {
-              "abability_bucket": {
-                "type": "number"
-              },
-              "abability_bucket_definition": {
-                "maxLength": 1000,
-                "minLength": 44,
-                "type": "string"
-              },
-              "classification": {
-                "maxLength": 1000,
-                "minLength": 22,
-                "type": "string"
-              },
-              "pharos_class": {
-                "type": [
-                  "string",
-                  "null"
-                ]
-              },
-              "safety_bucket": {
-                "type": "number"
-              },
-              "safety_bucket_definition": {
-                "maxLength": 1000,
-                "minLength": 50,
-                "type": "string"
-              },
-              "sm_druggability_bucket": {
-                "type": "number"
-              }
-            },
-            "type": "object"
-          },
-          "title": "Druggability Schema",
-          "type": [
-            "array",
-            "null"
-          ]
-        }
-      },
-      "meta": {}
-    },
     {
       "expectation_type": "expect_column_values_to_be_of_type",
       "kwargs": {
@@ -514,25 +453,25 @@
       "kwargs": {
         "column": "biodomains",
         "list_members": [
-          "Apoptosis",
-          "Vasculature",
-          "Lipid Metabolism",
-          "Oxidative Stress",
+          "Cell Cycle",
+          "Autophagy",
           "Mitochondrial Metabolism",
-          "APP Metabolism",
           "Proteostasis",
-          "DNA Repair",
-          "Synapse",
-          "Immune Response",
+          "APP Metabolism",
+          "Endolysosome",
+          "Vasculature",
           "Tau Homeostasis",
-          "Cell Cycle",
-          "Epigenetic",
+          "RNA Spliceosome",
+          "Apoptosis",
+          "Lipid Metabolism",
+          "Myelination",
           "Metal Binding and Homeostasis",
-          "Endolysosome",
+          "Synapse",
+          "Epigenetic",
+          "Immune Response",
+          "Oxidative Stress",
           "Structural Stabilization",
-          "Myelination",
-          "RNA Spliceosome",
-          "Autophagy"
+          "DNA Repair"
         ]
       },
       "meta": {}

diff --git a/src/agoradatatools/great_expectations/gx/json_schemas/gene_info/druggability.json b/src/agoradatatools/great_expectations/gx/json_schemas/gene_info/druggability.json
@@ -1,40 +1,34 @@
 {
     "$schema": "https://json-schema.org/draft/2019-09/schema",
     "$id": "http://example.com/example.json",
-    "type": ["array", "null"],
-    "default": [],
-    "title": "Druggability Schema",
-    "items": {
-        "type": "object",
-        "default": {},
-        "properties": {
-            "sm_druggability_bucket": {
-                "type": "number"
-            },
-            "safety_bucket": {
-                "type": "number"
-            },
-            "abability_bucket": {
-                "type": "number"
-            },
-            "pharos_class": {
-                "type": ["string", "null"]
-            },
-            "classification": {
+    "type": "object",
+    "default": {},
+    "title": "Root Schema",
+    "required": [
+        "pharos_class"
+    ],
+    "properties": {
+        "pharos_class": {
+            "type": "array",
+            "default": [],
+            "title": "The pharos_class Schema",
+            "items": {
                 "type": "string",
-                "minLength": 22,
-                "maxLength": 1000
+                "default": "",
+                "title": "A Schema",
+                "examples": [
+                    "Tchem"
+                ]
             },
-            "safety_bucket_definition": {
-                "type": "string",
-                "minLength": 50,
-                "maxLength": 1000
-            },
-            "abability_bucket_definition": {
-                "type": "string",
-                "minLength": 44,
-                "maxLength": 1000
-            }
+            "examples": [
+                [
+                    "Tchem"]
+            ]
         }
-    }
+    },
+    "examples": [{
+        "pharos_class": [
+            "Tchem"
+        ]
+    }]
 }