adds is_valid_predicate_name to ensure that positive examples are val…

…idkl

agatha/ml/hypothesis_predictor/hypothesis_predictor.py

@@ -208,7 +208,10 @@ def prepare_for_training(self)->None:
     entities = self.embeddings.keys()
     assert len(entities) > 0, "Failed to find embedding entities."
     self.coded_terms = list(filter(is_umls_term_type, entities))
-    self.predicates = list(filter(is_predicate_type, entities))
+    self.predicates = list(filter(
+      predicate_util.is_valid_predicate_name,
+      entities
+    ))
     self._vprint("Splitting train/validation")
     validation_size = int(
         len(self.predicates) * self.hparams.validation_fraction

agatha/ml/hypothesis_predictor/predicate_util.py

@@ -32,6 +32,15 @@ def clean_coded_term(term:str)->str:
     return f"{UMLS_TERM_TYPE}:{term}".lower()
 
 
+def is_valid_predicate_name(predicate_name:str)->bool:
+  if not is_predicate_type(predicate_name):
+    return False
+  try:
+    typ, sub, vrb, obj = predicate_name.lower().split(":")
+  except Exception:
+    return False
+  return (len(sub) > 0) and (len(obj) > 0)
+
 def parse_predicate_name(predicate_name:str)->Tuple[str, str]:
   """Parses subject and object from predicate name strings.
 

agatha/ml/hypothesis_predictor/test_predicate_util.py

@@ -23,3 +23,15 @@ def test_clean_coded_term_passthrough_lower():
   expected = "m:c0444567"
   actual = predicate_util.clean_coded_term(term)
   assert actual == expected
+
+def test_is_valid_predicate_name():
+  assert predicate_util.is_valid_predicate_name("p:c123:CAUSES:c456")
+  assert predicate_util.is_valid_predicate_name("p:123:CAUSES:456")
+  assert predicate_util.is_valid_predicate_name("p:1:FOO:4")
+  assert not predicate_util.is_valid_predicate_name("p:c123:CAUSES:")
+  assert not predicate_util.is_valid_predicate_name("p::CAUSES:c456")
+  assert not predicate_util.is_valid_predicate_name("c123:CAUSES:c456")
+  assert not predicate_util.is_valid_predicate_name("p:")
+  assert not predicate_util.is_valid_predicate_name("p:1:2")
+  assert not predicate_util.is_valid_predicate_name("p:::")
+  assert not predicate_util.is_valid_predicate_name("")