Update interpretation of DrugBank actions

indra/sources/drugbank/api.py

@@ -1,6 +1,7 @@
 import logging
 from typing import Optional, Sequence, Union
 from xml.etree import ElementTree
+
 from .processor import DrugbankProcessor
 
 logger = logging.getLogger(__name__)

indra/sources/drugbank/processor.py

@@ -1,16 +1,19 @@
 import logging
 from xml.etree import ElementTree
-from indra.statements import *
-from indra.databases.identifiers import ensure_chebi_prefix, \
+
+from indra.databases.identifiers import ensure_chebi_prefix,\
     ensure_chembl_prefix
+from indra.ontology.standardize import get_standard_agent
+from indra.statements import Activation, Complex, DecreaseAmount, Evidence,\
+    IncreaseAmount, Inhibition
 from indra.statements.validate import assert_valid_db_refs
-from indra.ontology.standardize import standardize_name_db_refs, \
-    get_standard_agent
 
 logger = logging.getLogger(__name__)
 
 drugbank_ns = {'db': 'http://www.drugbank.ca'}
 
+_UNHANDLED = set()
+
 
 class DrugbankProcessor:
     """Processor to extract INDRA Statements from DrugBank content.
@@ -28,6 +31,7 @@ class DrugbankProcessor:
     statements : list of indra.statements.Statement
         A list of INDRA Statements that were extracted from DrugBank content.
     """
+
     def __init__(self, xml_tree: ElementTree.ElementTree):
         self.xml_tree = xml_tree
         self.statements = []
@@ -59,9 +63,7 @@ def _extract_statements_for_drug(drug_element):
 
     @staticmethod
     def _get_statement_type(action):
-        if action in neutral_actions:
-            return None
-        elif action in activation_actions:
+        if action in activation_actions:
             return Activation
         elif action in inhibition_actions:
             return Inhibition
@@ -71,8 +73,13 @@ def _get_statement_type(action):
             return IncreaseAmount
         elif action == 'N/A':
             return Inhibition
-        else:
+        elif action in neutral_actions:
+            return _complex
+        elif action in skip_actions:
             return None
+        elif action not in _UNHANDLED:
+            _UNHANDLED.add(action)
+            logger.warning('unhandled DrugBank action: %s', action)
 
     @staticmethod
     def _get_target_agent(target_element):
@@ -160,22 +167,64 @@ def db_findall(element, path):
     return element.findall(path, namespaces=drugbank_ns)
 
 
-activation_actions = {'substrate', 'agonist', 'inducer', 'potentiator',
-                      'stimulator', 'cofactor', 'activator', 'ligand',
-                      'chaperone', 'partial agonist', 'protector',
-                      'positive allosteric modulator', 'positive modulator'}
-
-inhibition_actions = {'antagonist', 'inhibitor', 'binder', 'antibody',
-                      'inactivator', 'binding', 'blocker', 'negative modulator',
-                      'inverse agonist', 'neutralizer', 'weak inhibitor',
-                      'suppressor', 'disruptor',
-                      'inhibitory allosteric modulator'}
+def _complex(a, b, evidence):
+    return Complex([a, b], evidence=evidence)
 
-decrease_amount_actions = {'downregulator', 'metabolizer', 'chelator',
-                           'degradation',
-                           'incorporation into and destabilization'}
 
-increase_amount_actions = {'stabilization'}
+activation_actions = {'inducer', 'potentiator',
+                      'stimulator', 'cofactor', 'activator',
+                      'protector',
+                      'positive allosteric modulator', 'positive modulator'}
 
-neutral_actions = {'modulator', 'other/unknown', 'unknown', 'other',
-                   'regulator'}
+inhibition_actions = {'inhibitor', 'binder', 'antibody',
+                      'inactivator', 'binding', 'blocker', 'negative modulator',
+                      'neutralizer', 'weak inhibitor',
+                      'suppressor', 'disruptor', 'chelator',
+                      'inhibitory allosteric modulator',
+                      'translocation inhibitor', 'nucleotide exchange blocker',
+                      }
+
+decrease_amount_actions = {
+    'downregulator',
+    'metabolizer',
+    'degradation',
+    'incorporation into and destabilization',
+    'cleavage',
+    'inhibition of synthesis',
+}
+
+increase_amount_actions = {'stabilization', 'chaperone'}
+
+neutral_actions = {
+    'modulator',
+    'regulator',
+    'antagonist',
+    'substrate',
+    'agonist',
+    'ligand',
+    # e.g., Doxorubicin intercalates DNA to prevent transcription
+    'intercalation',
+    'inverse agonist',
+    # e.g., inhibits process on a protein's aggregation (like APP or LRRK)
+    'aggregation inhibitor',
+    'partial agonist',
+    'partial antagonist',
+    'antisense oligonucleotide',
+    'adduct',
+    'component of',
+    'product of',
+    'reducer',
+    'oxidizer',
+    # map to Ac INDRA statement?, but I'm not convinced by the idea of
+    # splitting up actions
+    'acetylation',
+    'allosteric modulator',
+    'deoxidizer',
+    'cross-linking/alkylation', # e.g. Busulfan (DB01008) alkalytes DNA
+}
+
+skip_actions = {
+    'other/unknown',
+    'unknown',
+    'other',
+}