Improved validation widget

SteampunkIslande · Jul 17, 2024 · cfa0b07 · cfa0b07
1 parent d80644c
commit cfa0b07
Show file tree

Hide file tree

Showing 3 changed files with 222 additions and 80 deletions.
diff --git a/config_folder/validation_methods/validation_ppi.json b/config_folder/validation_methods/validation_ppi.json
@@ -69,8 +69,8 @@
 	},
 	"steps": [
 		{
-			"title": "Etape 1",
-			"description": "Première étape de la validation.\nLes filtres appliqués montrent uniquement les variants dont la fréquence allélique en population générale est inférieure à 1%.",
+			"title": "Presets d'Alix",
+			"description": "<body>Première étape de la validation.<br>Les filtres appliqués montrent uniquement les variants dont la fréquence allélique en population générale est inférieure à <a style=\"color:red;background:yellow\">1%</a>.</body>",
 			"query": {
 				"select": {
 					"fields": [
@@ -166,6 +166,105 @@
 					]
 				}
 			}
+		},
+		{
+			"title": "Presets de Pascale",
+			"description": "<body>Première étape de la validation.\nLes filtres appliqués montrent uniquement les variants dont la fréquence allélique en population générale est inférieure à <a style=\"color:red;background:yellow\">1%</a>.</body>",
+			"query": {
+				"select": {
+					"fields": [
+						"main_table.run_name",
+						"main_table.sample_name",
+						"main_table.snpeff_Gene_Name",
+						"main_table.snpeff_Feature_ID",
+						"main_table.chromosome",
+						"main_table.position",
+						"main_table.reference",
+						"main_table.alternate",
+						"main_table.cv_AF",
+						"main_table.snpeff_Annotation",
+						"main_table.cv_GT",
+						"main_table.snpeff_Annotation_Impact",
+						"agg.ref_count",
+						"agg.var_count",
+						"recur.count",
+						"clinvar.CLNACC",
+						"clinvar.CLNSIG",
+						"main_table.\"snpeff_HGVS.c\" AS 'c. de l anomalie'"
+					],
+					"tables": [
+						{
+							"expression": "{main_table}",
+							"alias": "main_table"
+						},
+						{
+							"expression": "read_parquet('{pwd}/aggregates/variants.parquet')",
+							"alias": "agg",
+							"on": "main_table.variant_hash = agg.variant_hash",
+							"how": ""
+						},
+						{
+							"select": {
+								"fields": [
+									"variant_hash",
+									"COUNT(*) as count"
+								],
+								"tables": [
+									{
+										"select": {
+											"fields": [
+												"DISTINCT(sample_name)",
+												"variant_hash"
+											],
+											"tables": [
+												{
+													"expression": "{main_table}",
+													"alias": ""
+												}
+											]
+										},
+										"alias": ""
+									}
+								],
+								"group_by": [
+									"variant_hash"
+								]
+							},
+							"alias": "recur",
+							"on": "main_table.variant_hash = recur.variant_hash",
+							"how": ""
+						},
+						{
+							"select": {
+								"fields": [
+									"variant_hash",
+									"rcvAcc AS CLNACC",
+									"clinSign AS CLNSIG"
+								],
+								"tables": [
+									{
+										"expression": "read_parquet('{pwd}/annotations/clinvarMain.parquet')",
+										"alias": ""
+									}
+								]
+							},
+							"alias": "clinvar",
+							"on": "main_table.variant_hash = clinvar.variant_hash",
+							"how": ""
+						}
+					],
+					"order_by": [
+						{
+							"field": "main_table.chromosome",
+							"order": "DESC"
+						},
+						{
+							"field": "main_table.position",
+							"order": "DESC"
+						}
+					]
+				}
+			}
 		}
 	],
 	"final": {
@@ -212,25 +311,13 @@
 						"how": ""
 					},
 					{
-						"expression": "{pwd}/annotations/hgnc.hg19.parquet",
+						"expression": "read_parquet('{pwd}/annotations/hgnc.hg19.parquet')",
 						"alias": "hgnc",
 						"on": "main_table.chromosome = hgnc.chrom AND main_table.position >= hgnc.chromStart AND main_table.position <= hgnc.chromEnd",
 						"how": ""
 					},
 					{
-						"select": {
-							"fields": [
-								"variant_hash",
-								"rcvAcc AS CLNACC",
-								"clinSign AS CLNSIG"
-							],
-							"tables": [
-								{
-									"expression": "read_parquet('{pwd}/clinvarMain.parquet')",
-									"alias": ""
-								}
-							]
-						},
+						"expression": "read_parquet('{pwd}/annotations/clinvarMain.parquet')",
 						"alias": "clinvar",
 						"on": "main_table.variant_hash = clinvar.variant_hash",
 						"how": ""

diff --git a/query.py b/query.py
@@ -231,15 +231,18 @@ def generate_query_template_from_json(self, data: dict) -> "Query":
         self.internal_changed.emit()
         return self
 
-    def select_query(self):
+    def select_query(self, paginated=True):
+        """Generates the select query to run on the database. Set paginated to False if you need a query that returns all rows."""
         if not self.readonly_table:
             return ""
 
+        pagination = f" LIMIT {self.limit} OFFSET {self.offset}" if paginated else ""
+
         additional_where = (
-            f" WHERE {str(self.filter_model.root)} " if self.filter_model.root else ""
+            f" WHERE {str(self.filter_model.root)}" if self.filter_model.root else ""
         )
 
-        return f"SELECT * FROM ({self.query_template}){additional_where}LIMIT {self.limit} OFFSET {self.offset}".format(
+        return f"SELECT * FROM ({self.query_template}){additional_where}{pagination}".format(
             **{
                 "main_table": self.readonly_table,
                 "user_table": f'"{self.editable_table_name}"',
@@ -273,7 +276,6 @@ def to_do(self):
             return "Please select a validation table"
 
     def update_data(self):
-        print("Updating data")
         # Empty data before updating
         self.header = []
         self.data = []