Merge pull request #506 from monarch-initiative/exclude-phenio-edges-…

…with-nulls

Exclude phenio edges with null subject/object

pyproject.toml

@@ -18,17 +18,15 @@ packages = [
 python = ">=3.10,<3.12"
 kghub-downloader = "^0.3.2"
 koza = "^0.3.0"
-cat-merge = "^0.1.19"
+cat-merge = "0.1.20"
 closurizer = "0.3.0"
-#monarch-py = "^0.13.2"
-monarch-py = {git = "https://github.com/monarch-initiative/monarch-app.git", branch = "taxon-labels", subdirectory = "backend"}
+monarch-py = "0.15.1"
 kgx = "^2.1.0"
 multi-indexer = "*"
 biolink-model = "3.5.0"
-linkml = "^1.5.5"
+linkml = "^1.5.7"
 #linkml-solr = "^0.1.3"
 linkml-solr = {git = "https://github.com/linkml/linkml-solr.git", branch = "top_class"}
-greenlet = "^1.1.2"
 importlib-metadata = "^4.6.1"
 sh = "^1.14.3"
 typer = "^0.7"

src/monarch_ingest/cli_utils.py

@@ -203,6 +203,9 @@ def transform_phenio(
     edges_df = edges_df[~edges_df["subject"].str.startswith(tuple(exclude_prefixes))
                         & ~edges_df["object"].str.startswith(tuple(exclude_prefixes))]
 
+    # Remove edges where the subject or object is NA
+    edges_df = edges_df[~edges_df["subject"].isna() & ~edges_df["object"].isna()]
+
     valid_predicates = {f"biolink:{convert_to_snake_case(pred)}" for pred in biolink_model_schema.slot_descendants("related to")}
     phenio_predicates = set(edges_df['predicate'].unique())
     invalid_predicates = phenio_predicates - valid_predicates