fix 3-star processing

Author: sfluegel05

chebai/preprocessing/datasets/chebi.py

@@ -253,7 +253,12 @@ def _extract_class_hierarchy(self, data_path: str) -> "nx.DiGraph":
             ):
                 term_dict = term_callback(term_doc)
                 if term_dict and (
-                    not self.subset or term_dict["subset"] == self.subset
+                    not self.subset
+                    or (
+                        "subset" in term_dict
+                        and term_dict["subset"] is not None
+                        and term_dict["subset"][0] == self.subset[0]
+                    )  # match 3:STAR to 3_STAR, 3star, 3_star, etc.
                 ):
                     elements.append(term_dict)
 
@@ -1068,6 +1073,6 @@ def term_callback(doc: "fastobo.term.TermFrame") -> Union[Dict, bool]:
 
 
 if __name__ == "__main__":
-    dataset = ChEBIOver50(chebi_version=246)
-    # dataset.prepare_data()
+    dataset = ChEBIOver50(chebi_version=248, subset="3_STAR")
+    dataset.prepare_data()
     dataset.setup()