Load phenio.db from /data if available, use pystow to fetch from mona…

…rch data bucket if not (#356)

This should stop our multiple pystow download problem (and mean no
download on startup at all, happily)

backend/src/monarch_py/api/config.py

@@ -11,6 +11,7 @@ class Settings(BaseSettings):
     solr_host = os.getenv("SOLR_HOST") if os.getenv("SOLR_HOST") else "127.0.0.1"
     solr_port = os.getenv("SOLR_PORT") if os.getenv("SOLR_PORT") else 8983
     solr_url = os.getenv("SOLR_URL") if os.getenv("SOLR_URL") else f"http://{solr_host}:{solr_port}/solr"
+    phenio_db_path = os.getenv("PHENIO_DB_PATH") if os.getenv("PHENIO_DB_PATH") else "/data/phenio.db"
 
 
 settings = Settings()
@@ -23,4 +24,5 @@ def solr():
 
 @lru_cache(maxsize=1)
 def oak():
-    return OakImplementation().init_semsim()
+    phenio_db_path = settings.phenio_db_path if os.path.exists(settings.phenio_db_path) else None
+    return OakImplementation().init_semsim(phenio_path=phenio_db_path)

backend/src/monarch_py/implementations/oak/oak_implementation.py

@@ -8,6 +8,7 @@
 from oaklib.interfaces.semsim_interface import SemanticSimilarityInterface
 from oaklib.selector import get_adapter
 from linkml_runtime.dumpers.json_dumper import JSONDumper
+import pystow
 
 
 @dataclass
@@ -18,16 +19,28 @@ class OakImplementation(SemanticSimilarityInterface):
     json_dumper = JSONDumper()
     default_predicates = ["rdfs:subClassOf", "BFO:0000050", "UPHENO:0000001"]
 
-    def init_semsim(self):
+    default_phenio_db_url = "https://data.monarchinitiative.org/monarch-kg-dev/latest/phenio.db.gz"
+
+    def init_semsim(self, phenio_path: str = None, force_update: bool = False):
         if self.semsim is None:
             logger.info("Warming up semsimian")
             start = time.time()
             # self.semsim = get_adapter(f"sqlite:obo:phenio")
-            logger.debug("Getting semsimian adapter")
-            self.semsim = get_adapter(f"semsimian:sqlite:obo:phenio")
 
-            # for some reason, we need to run a query to get the adapter
-            # to initialize properly
+
+            if phenio_path:
+                logger.debug(f"Creating semsimian adapter using phenio_path at {phenio_path}")
+                self.semsim = get_adapter(f"semsimian:sqlite:{phenio_path}")
+            else:
+                monarchstow = pystow.module("monarch")
+
+                with monarchstow.ensure_gunzip(
+                    "phenio", url=self.default_phenio_db_url, force=force_update
+                ) as stowed_phenio_path:
+                    logger.debug(f"Creating semsimian adapter using pystow at {stowed_phenio_path}")
+                    self.semsim = get_adapter(f"semsimian:sqlite:{stowed_phenio_path}")
+
+            # run a query to get the adapter to initialize properly
             logger.debug("Running query to initialize adapter")
             self.semsim.termset_pairwise_similarity(
                 subjects=["MP:0010771"],