WIP nxontology_ml_train_predict

nxontology_data/efo/efo.py

@@ -13,6 +13,7 @@
 import pandas as pd
 import rdflib
 from nxontology import NXOntology
+from nxontology_ml.model.predict import train_predict as nxontology_ml_train_predict
 
 from nxontology_data.utils import (
     get_source_output_dir,
@@ -319,6 +320,18 @@ def write_outputs(self) -> None:
         )
         if nxo.name == "efo_otar_profile":
             nxo_slim = self.create_slim_nxo(nxo)
+            # classify EFO node/disease precision using nxontology-ml
+            nxo_slim.freeze()
+            logger.info("Beginning nxontology-ml disease precision classification.")
+            precision_df = nxontology_ml_train_predict(nxo=nxo_slim)
+            write_dataframe(
+                precision_df,
+                output_dir.joinpath(f"{self.name}_precision_classifications.json.gz"),
+            )
+            for row in precision_df.itertuples():
+                nxo_slim.graph.nodes[row.identifier][
+                    "disease_precision"
+                ] = row.precision
             write_ontology(nxo_slim, output_dir, compression_threshold_mb=30.0)
 
     @staticmethod