Add histopheno bar graph to gene pages (#630)

Use UPHENO to drive the Histopheno bar chart, then show the visualization on gene pages (likely limited to a subset of species, due to both the content of the list and the presence or absense of upheno connectivity)
monarch-initiative · Apr 19, 2024 · fbdda4c · fbdda4c
1 parent accf783
commit fbdda4c
Show file tree

Hide file tree

Showing 17 changed files with 363 additions and 359 deletions.
diff --git a/backend/src/monarch_py/datamodels/solr.py b/backend/src/monarch_py/datamodels/solr.py
@@ -13,26 +13,26 @@ class core(Enum):
 
 
 class HistoPhenoKeys(Enum):
-    skeletal_system = "HP:0000924"
-    nervous_system = "HP:0000707"
-    head_neck = "HP:0000152"
-    integument = "HP:0001574"
-    eye = "HP:0000478"
-    cardiovascular_system = "HP:0001626"
-    metabolism_homeostasis = "HP:0001939"
-    genitourinary_system = "HP:0000119"
-    digestive_system = "HP:0025031"
-    neoplasm = "HP:0002664"
-    blood = "HP:0001871"
-    immune_system = "HP:0002715"
-    endocrine = "HP:0000818"
-    musculature = "HP:0003011"
-    respiratory = "HP:0002086"
-    ear = "HP:0000598"
-    connective_tissue = "HP:0003549"
-    prenatal_or_birth = "HP:0001197"
-    growth = "HP:0001507"
-    breast = "HP:0000769"
+    skeletal_system = "UPHENO:0002964"  # "HP:0000924"
+    nervous_system = "UPHENO:0004523"  # "HP:0000707"
+    head_neck = "UPHENO:0002764"  # "HP:0000152"
+    integument = "UPHENO:0002635"  # "HP:0001574"
+    eye = "UPHENO:0003020"  # "HP:0000478"
+    cardiovascular_system = "UPHENO:0080362"  # "HP:0001626"
+    metabolism_homeostasis = "HP:0001939"  # ??? No uPheno parent
+    genitourinary_system = "UPHENO:0002642"  # "HP:0000119"
+    digestive_system = "UPHENO:0002833"  # "HP:0025031"
+    neoplasm = "HP:0002664"  # ??? No uPheno parent
+    blood = "UPHENO:0004459"  # "HP:0001871"
+    immune_system = "UPHENO:0002948"  # "HP:0002715"
+    endocrine = "UPHENO:0003116"  # "HP:0000818"
+    musculature = "UPHENO:0002816"  # "HP:0003011"
+    respiratory = "UPHENO:0004536"  # "HP:0002086"
+    ear = "HP:0000598"  # UPHENO:0002903
+    connective_tissue = "UPHENO:0002712"  # "HP:0003549"
+    prenatal_or_birth = "UPHENO:0075949"  # "HP:0001197"
+    growth = "UPHENO:0049874"  # "HP:0001507"
+    breast = "UPHENO:0003013"  # "HP:0000769"
 
 
 class SolrQuery(BaseModel):

diff --git a/backend/src/monarch_py/implementations/solr/solr_implementation.py b/backend/src/monarch_py/implementations/solr/solr_implementation.py
@@ -289,12 +289,12 @@ def get_associations(
         associations = parse_associations(query_result, compact, offset, limit)
         return associations
 
-    def get_histopheno(self, subject_closure: Optional[str] = None) -> HistoPheno:
+    def get_histopheno(self, subject: Optional[str] = None) -> HistoPheno:
         """Get histopheno counts for a given subject_closure"""
         solr = SolrService(base_url=self.base_url, core=core.ASSOCIATION)
-        query = build_histopheno_query(subject_closure)
+        query = build_histopheno_query(subject)
         query_result = solr.query(query)
-        histopheno = parse_histopheno(query_result, subject_closure)
+        histopheno = parse_histopheno(query_result, subject)
         return histopheno
 
     def get_multi_entity_associations(

diff --git a/backend/src/monarch_py/implementations/solr/solr_query_utils.py b/backend/src/monarch_py/implementations/solr/solr_query_utils.py
@@ -115,9 +115,9 @@ def build_association_counts_query(entity: str) -> SolrQuery:
     return query
 
 
-def build_histopheno_query(subject_closure: str) -> SolrQuery:
+def build_histopheno_query(subject: str) -> SolrQuery:
     query = build_association_query(
-        subject_closure=subject_closure,
+        subject=[subject],
         offset=0,
         limit=0,
     )

diff --git a/backend/tests/fixtures/association_counts_response.py b/backend/tests/fixtures/association_counts_response.py
@@ -5,7 +5,7 @@
 def association_counts_response():
     return {
         "responseHeader": {
-            "QTime": 1,
+            "QTime": 0,
             "params": {
                 "facet.query": [
                     '(category:"biolink:DiseaseToPhenotypicFeatureAssociation") AND (subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121")',

diff --git a/backend/tests/fixtures/association_table_response.py b/backend/tests/fixtures/association_table_response.py
@@ -5,7 +5,7 @@
 def association_table_response():
     return {
         "responseHeader": {
-            "QTime": 1,
+            "QTime": 0,
             "params": {
                 "mm": "100%",
                 "q": "*:*",

diff --git a/backend/tests/fixtures/histopheno.py b/backend/tests/fixtures/histopheno.py
@@ -6,25 +6,25 @@ def histopheno():
     return {
         "id": "MONDO:0020121",
         "items": [
-            {"label": "musculature", "count": 1677, "id": "HP:0003011"},
-            {"label": "nervous_system", "count": 1071, "id": "HP:0000707"},
-            {"label": "head_neck", "count": 577, "id": "HP:0000152"},
-            {"label": "skeletal_system", "count": 471, "id": "HP:0000924"},
-            {"label": "eye", "count": 287, "id": "HP:0000478"},
+            {"label": "musculature", "count": 1677, "id": "UPHENO:0002816"},
+            {"label": "nervous_system", "count": 1071, "id": "UPHENO:0004523"},
+            {"label": "head_neck", "count": 577, "id": "UPHENO:0002764"},
+            {"label": "skeletal_system", "count": 471, "id": "UPHENO:0002964"},
+            {"label": "eye", "count": 287, "id": "UPHENO:0003020"},
             {"label": "metabolism_homeostasis", "count": 213, "id": "HP:0001939"},
-            {"label": "cardiovascular_system", "count": 177, "id": "HP:0001626"},
-            {"label": "blood", "count": 175, "id": "HP:0001871"},
-            {"label": "connective_tissue", "count": 161, "id": "HP:0003549"},
-            {"label": "respiratory", "count": 150, "id": "HP:0002086"},
+            {"label": "cardiovascular_system", "count": 177, "id": "UPHENO:0080362"},
+            {"label": "blood", "count": 175, "id": "UPHENO:0004459"},
+            {"label": "connective_tissue", "count": 161, "id": "UPHENO:0002712"},
+            {"label": "respiratory", "count": 150, "id": "UPHENO:0004536"},
             {"label": "neoplasm", "count": 148, "id": "HP:0002664"},
-            {"label": "digestive_system", "count": 142, "id": "HP:0025031"},
-            {"label": "integument", "count": 47, "id": "HP:0001574"},
-            {"label": "genitourinary_system", "count": 44, "id": "HP:0000119"},
-            {"label": "growth", "count": 32, "id": "HP:0001507"},
+            {"label": "digestive_system", "count": 142, "id": "UPHENO:0002833"},
+            {"label": "integument", "count": 47, "id": "UPHENO:0002635"},
+            {"label": "genitourinary_system", "count": 44, "id": "UPHENO:0002642"},
+            {"label": "growth", "count": 32, "id": "UPHENO:0049874"},
             {"label": "ear", "count": 28, "id": "HP:0000598"},
-            {"label": "endocrine", "count": 25, "id": "HP:0000818"},
-            {"label": "immune_system", "count": 22, "id": "HP:0002715"},
-            {"label": "prenatal_or_birth", "count": 21, "id": "HP:0001197"},
-            {"label": "breast", "count": 1, "id": "HP:0000769"},
+            {"label": "endocrine", "count": 25, "id": "UPHENO:0003116"},
+            {"label": "immune_system", "count": 22, "id": "UPHENO:0002948"},
+            {"label": "prenatal_or_birth", "count": 21, "id": "UPHENO:0075949"},
+            {"label": "breast", "count": 1, "id": "UPHENO:0003013"},
         ],
     }
diff --git a/backend/tests/fixtures/histopheno_query.py b/backend/tests/fixtures/histopheno_query.py
@@ -11,28 +11,28 @@ def histopheno_query():
         "facet_min_count": 1,
         "facet_fields": [],
         "facet_queries": [
-            'object_closure:"HP:0000924"',
-            'object_closure:"HP:0000707"',
-            'object_closure:"HP:0000152"',
-            'object_closure:"HP:0001574"',
-            'object_closure:"HP:0000478"',
-            'object_closure:"HP:0001626"',
+            'object_closure:"UPHENO:0002964"',
+            'object_closure:"UPHENO:0004523"',
+            'object_closure:"UPHENO:0002764"',
+            'object_closure:"UPHENO:0002635"',
+            'object_closure:"UPHENO:0003020"',
+            'object_closure:"UPHENO:0080362"',
             'object_closure:"HP:0001939"',
-            'object_closure:"HP:0000119"',
-            'object_closure:"HP:0025031"',
+            'object_closure:"UPHENO:0002642"',
+            'object_closure:"UPHENO:0002833"',
             'object_closure:"HP:0002664"',
-            'object_closure:"HP:0001871"',
-            'object_closure:"HP:0002715"',
-            'object_closure:"HP:0000818"',
-            'object_closure:"HP:0003011"',
-            'object_closure:"HP:0002086"',
+            'object_closure:"UPHENO:0004459"',
+            'object_closure:"UPHENO:0002948"',
+            'object_closure:"UPHENO:0003116"',
+            'object_closure:"UPHENO:0002816"',
+            'object_closure:"UPHENO:0004536"',
             'object_closure:"HP:0000598"',
-            'object_closure:"HP:0003549"',
-            'object_closure:"HP:0001197"',
-            'object_closure:"HP:0001507"',
-            'object_closure:"HP:0000769"',
+            'object_closure:"UPHENO:0002712"',
+            'object_closure:"UPHENO:0075949"',
+            'object_closure:"UPHENO:0049874"',
+            'object_closure:"UPHENO:0003013"',
         ],
-        "filter_queries": ["subject_closure:MONDO\\:0020121"],
+        "filter_queries": ['subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121"'],
         "query_fields": None,
         "def_type": "edismax",
         "q_op": "AND",

diff --git a/backend/tests/fixtures/histopheno_response.py b/backend/tests/fixtures/histopheno_response.py
@@ -8,34 +8,34 @@ def histopheno_response():
             "QTime": 1,
             "params": {
                 "facet.query": [
-                    'object_closure:"HP:0000924"',
-                    'object_closure:"HP:0000707"',
-                    'object_closure:"HP:0000152"',
-                    'object_closure:"HP:0001574"',
-                    'object_closure:"HP:0000478"',
-                    'object_closure:"HP:0001626"',
+                    'object_closure:"UPHENO:0002964"',
+                    'object_closure:"UPHENO:0004523"',
+                    'object_closure:"UPHENO:0002764"',
+                    'object_closure:"UPHENO:0002635"',
+                    'object_closure:"UPHENO:0003020"',
+                    'object_closure:"UPHENO:0080362"',
                     'object_closure:"HP:0001939"',
-                    'object_closure:"HP:0000119"',
-                    'object_closure:"HP:0025031"',
+                    'object_closure:"UPHENO:0002642"',
+                    'object_closure:"UPHENO:0002833"',
                     'object_closure:"HP:0002664"',
-                    'object_closure:"HP:0001871"',
-                    'object_closure:"HP:0002715"',
-                    'object_closure:"HP:0000818"',
-                    'object_closure:"HP:0003011"',
-                    'object_closure:"HP:0002086"',
+                    'object_closure:"UPHENO:0004459"',
+                    'object_closure:"UPHENO:0002948"',
+                    'object_closure:"UPHENO:0003116"',
+                    'object_closure:"UPHENO:0002816"',
+                    'object_closure:"UPHENO:0004536"',
                     'object_closure:"HP:0000598"',
-                    'object_closure:"HP:0003549"',
-                    'object_closure:"HP:0001197"',
-                    'object_closure:"HP:0001507"',
-                    'object_closure:"HP:0000769"',
+                    'object_closure:"UPHENO:0002712"',
+                    'object_closure:"UPHENO:0075949"',
+                    'object_closure:"UPHENO:0049874"',
+                    'object_closure:"UPHENO:0003013"',
                 ],
                 "mm": "100%",
                 "q": "*:*",
                 "defType": "edismax",
                 "facet_min_count": "1",
                 "start": "0",
                 "q.op": "AND",
-                "fq": "subject_closure:MONDO\\:0020121",
+                "fq": 'subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121"',
                 "rows": "0",
                 "facet": "true",
             },
@@ -44,26 +44,26 @@ def histopheno_response():
         "facet_counts": {
             "facet_fields": {},
             "facet_queries": {
-                'object_closure:"HP:0000924"': 471,
-                'object_closure:"HP:0000707"': 1071,
-                'object_closure:"HP:0000152"': 577,
-                'object_closure:"HP:0001574"': 47,
-                'object_closure:"HP:0000478"': 287,
-                'object_closure:"HP:0001626"': 177,
+                'object_closure:"UPHENO:0002964"': 471,
+                'object_closure:"UPHENO:0004523"': 1071,
+                'object_closure:"UPHENO:0002764"': 577,
+                'object_closure:"UPHENO:0002635"': 47,
+                'object_closure:"UPHENO:0003020"': 287,
+                'object_closure:"UPHENO:0080362"': 177,
                 'object_closure:"HP:0001939"': 213,
-                'object_closure:"HP:0000119"': 44,
-                'object_closure:"HP:0025031"': 142,
+                'object_closure:"UPHENO:0002642"': 44,
+                'object_closure:"UPHENO:0002833"': 142,
                 'object_closure:"HP:0002664"': 148,
-                'object_closure:"HP:0001871"': 175,
-                'object_closure:"HP:0002715"': 22,
-                'object_closure:"HP:0000818"': 25,
-                'object_closure:"HP:0003011"': 1677,
-                'object_closure:"HP:0002086"': 150,
+                'object_closure:"UPHENO:0004459"': 175,
+                'object_closure:"UPHENO:0002948"': 22,
+                'object_closure:"UPHENO:0003116"': 25,
+                'object_closure:"UPHENO:0002816"': 1677,
+                'object_closure:"UPHENO:0004536"': 150,
                 'object_closure:"HP:0000598"': 28,
-                'object_closure:"HP:0003549"': 161,
-                'object_closure:"HP:0001197"': 21,
-                'object_closure:"HP:0001507"': 32,
-                'object_closure:"HP:0000769"': 1,
+                'object_closure:"UPHENO:0002712"': 161,
+                'object_closure:"UPHENO:0075949"': 21,
+                'object_closure:"UPHENO:0049874"': 32,
+                'object_closure:"UPHENO:0003013"': 1,
             },
         },
     }
diff --git a/backend/tests/fixtures/search_response.py b/backend/tests/fixtures/search_response.py
@@ -5,7 +5,7 @@
 def search_response():
     return {
         "responseHeader": {
-            "QTime": 2,
+            "QTime": 0,
             "params": {
                 "mm": "100%",
                 "q": "fanconi",

diff --git a/backend/tests/integration/test_solr_histopheno.py b/backend/tests/integration/test_solr_histopheno.py
@@ -15,4 +15,4 @@ def test_histopheno():
     for k in hp.items:
         total += k.count
 
-    assert hp.items[0].id == "HP:0003011"
+    assert hp.items[0].id == "UPHENO:0002816"
diff --git a/backend/tests/unit/test_solr_queries.py b/backend/tests/unit/test_solr_queries.py
@@ -14,7 +14,8 @@
     build_search_query,
     obsolete_unboost,
     entity_predicate_boost,
-    category_boost, blank_search_boost,
+    category_boost,
+    blank_search_boost,
 )
 from monarch_py.utils.utils import compare_dicts, dict_diff
 
@@ -164,5 +165,4 @@ def test_blank_search_boost():
     boost = blank_search_boost()
     assert boost.startswith("product(")
     assert boost.endswith(")")
-    assert "if(termfreq(id,\"MONDO:0007523\"),12,1)" in boost
-
+    assert 'if(termfreq(id,"MONDO:0007523"),12,1)' in boost