Merge pull request #1450 from bgyori/proc_improve

Processor improvements

indra/assemblers/indranet/net.py

@@ -327,8 +327,8 @@ def _simple_scorer_update(G, edge):
     # Catch underflow
     except FloatingPointError as err:
         # Numpy precision
-        NP_PRECISION = 10 ** -np.finfo(np.longfloat).precision
-        logger.warning('%s: Resetting ag_belief to 10*np.longfloat precision '
+        NP_PRECISION = 10 ** -np.finfo(np.longdouble).precision
+        logger.warning('%s: Resetting ag_belief to 10*np.longdouble precision '
                        '(%.0e)' % (err, Decimal(NP_PRECISION * 10)))
         ag_belief = NP_PRECISION * 10
     return ag_belief
@@ -337,14 +337,14 @@ def _simple_scorer_update(G, edge):
 def _complementary_belief(G, edge):
     # Aggregate belief score: 1-prod(1-belief_i)
     np.seterr(all='raise')
-    NP_PRECISION = 10 ** -np.finfo(np.longfloat).precision  # Numpy precision
+    NP_PRECISION = 10 ** -np.finfo(np.longdouble).precision  # Numpy precision
     belief_list = [s['belief'] for s in G.edges[edge]['statements']]
     try:
-        ag_belief = np.longfloat(1.0) - np.prod(np.fromiter(
-            map(lambda belief: np.longfloat(1.0) - belief, belief_list),
-            dtype=np.longfloat))
+        ag_belief = np.longdouble(1.0) - np.prod(np.fromiter(
+            map(lambda belief: np.longdouble(1.0) - belief, belief_list),
+            dtype=np.longdouble))
     except FloatingPointError as err:
-        logger.warning('%s: Resetting ag_belief to 10*np.longfloat precision '
+        logger.warning('%s: Resetting ag_belief to 10*np.longdouble precision '
                        '(%.0e)' % (err, Decimal(NP_PRECISION * 10)))
         ag_belief = NP_PRECISION * 10
     return ag_belief

indra/databases/hgnc_client.py

@@ -422,6 +422,10 @@ def get_hgnc_name_from_mgi_name(mgi_name: str) -> Union[str, None]:
 def _read_hgnc_maps():
     hgnc_file = get_resource_path("hgnc_entries.tsv")
     csv_rows = read_unicode_csv(hgnc_file, delimiter='\t', encoding='utf-8')
+    hgnc_uniprot_preferred = get_resource_path("hgnc_uniprot_preferred.csv")
+    csv_rows_uniprot_preferred = \
+        read_unicode_csv(hgnc_uniprot_preferred, delimiter=',',
+                         encoding='utf-8')
     hgnc_names = {}
     hgnc_ids = {}
     hgnc_withdrawn = []
@@ -515,19 +519,26 @@ def _read_hgnc_maps():
     for old_id, new_id in hgnc_withdrawn_new_ids.items():
         hgnc_names[old_id] = hgnc_names[new_id]
 
+    uniprot_ids_preferred = {}
+    for row in csv_rows_uniprot_preferred:
+        hgnc_id = row[0]
+        uniprot_id = row[1]
+        uniprot_ids_preferred[hgnc_id] = uniprot_id
+
     return (
         hgnc_names, hgnc_ids, hgnc_withdrawn,
         uniprot_ids, entrez_ids, entrez_ids_reverse, mouse_map, rat_map,
         prev_sym_map, ensembl_ids, ensembl_ids_reverse, gene_types,
         dict(hgnc_to_enzymes), dict(enzyme_to_hgncs),
+        uniprot_ids_preferred
     )
 
 
 (
     hgnc_names, hgnc_ids, hgnc_withdrawn, uniprot_ids, entrez_ids,
     entrez_ids_reverse, mouse_map, rat_map, prev_sym_map, ensembl_ids,
     ensembl_ids_reverse, gene_type,
-    hgnc_to_enzymes, enzyme_to_hgncs,
+    hgnc_to_enzymes, enzyme_to_hgncs, uniprot_ids_preferred
 ) = _read_hgnc_maps()
 
 

indra/ontology/bio/ontology.py

@@ -26,7 +26,7 @@ class BioOntology(IndraOntology):
     # should be incremented to "force" rebuilding the ontology to be consistent
     # with the underlying resource files.
     name = 'bio'
-    version = '1.33'
+    version = '1.34'
     ontology_namespaces = [
         'go', 'efo', 'hp', 'doid', 'chebi', 'ido', 'mondo', 'eccode',
     ]
@@ -147,11 +147,15 @@ def add_hgnc_uniprot_entrez_xrefs(self):
         from indra.databases import hgnc_client
         from indra.databases import uniprot_client
         edges = []
-        for hid, uid in hgnc_client.uniprot_ids.items():
-            uids = uid.split(', ')
+        for hid, upid in hgnc_client.uniprot_ids.items():
+            uids = upid.split(', ')
+            preferred = hgnc_client.uniprot_ids_preferred.get(hid)
+            if preferred:
+                uids = [preferred]
             for uid in uids:
+                edge_data = {'type': 'xref', 'source': 'hgnc'}
                 edges.append((self.label('HGNC', hid), self.label('UP', uid),
-                              {'type': 'xref', 'source': 'hgnc'}))
+                              edge_data))
         self.add_edges_from(edges)
 
         edges = [(self.label('UP', uid), self.label('HGNC', hid),

indra/resources/hgnc_uniprot_preferred.csv

@@ -0,0 +1,3 @@
+hgnc_id,uniprot_id
+17868,Q9BXH1
+30377,Q14160

indra/sources/bel/processor.py

@@ -546,28 +546,17 @@ def get_db_refs_by_name(ns, name, node_data):
         if up_id:
             db_refs = {'UP': up_id}
     # Map Selventa families and complexes to FamPlex
-    elif ns == 'SFAM':
+    elif ns in {'SFAM', 'SCOMP'}:
         try:
-            sfam_id, xrefs = selventa_lookup[('SFAM', name)]
-            db_refs = {"SFAM": sfam_id}
+            selventa_id, xrefs = selventa_lookup[(ns, name)]
+            db_refs = {ns: selventa_id}
             indra_name = bel_to_indra.get(name)
         except KeyError:
             indra_name = None
             db_refs = None
-
-        if indra_name is None:
-            logger.info('Could not find mapping for BEL/SFAM family: '
-                        '%s (%s)' % (name, node_data))
-        else:
-            db_refs['FPLX'] = indra_name
-            name = indra_name
-    elif ns == 'SCOMP':
-        scomp_id, xrefs = selventa_lookup[('SCOMP', name)]
-        db_refs = {'SCOMP': scomp_id}
-        indra_name = bel_to_indra.get(name)
         if indra_name is None:
-            logger.info('Could not find mapping for BEL/SCOMP complex: '
-                        '%s (%s)' % (name, node_data))
+            logger.info('Could not find mapping for BEL/%s family: '
+                        '%s (%s)' % (ns, name, node_data))
         else:
             db_refs['FPLX'] = indra_name
             name = indra_name

indra/tests/test_indranet_assembler.py

@@ -175,10 +175,10 @@ def test_to_digraph():
             'Activation', 'Phosphorylation', 'Inhibition', 'IncreaseAmount'}
     assert all(digraph.edges[e].get('belief', False) for e in digraph.edges)
     assert all(isinstance(digraph.edges[e]['belief'],
-                          (float, np.longfloat)) for e in digraph.edges)
+                          (float, np.longdouble)) for e in digraph.edges)
     assert all(digraph.edges[e].get('weight', False) for e in digraph.edges)
     assert all(isinstance(digraph.edges[e]['weight'],
-                          (float, np.longfloat)) for e in digraph.edges)
+                          (float, np.longdouble)) for e in digraph.edges)
     digraph_from_df = IndraNet.digraph_from_df(df)
     assert nx.is_isomorphic(digraph, digraph_from_df)
 
@@ -206,11 +206,11 @@ def test_to_signed_graph():
     assert all(signed_graph.edges[e].get('belief', False) for e in
                signed_graph.edges)
     assert all(isinstance(signed_graph.edges[e]['belief'],
-                          (float, np.longfloat)) for e in signed_graph.edges)
+                          (float, np.longdouble)) for e in signed_graph.edges)
     assert all(signed_graph.edges[e].get('weight', False) for e in
                signed_graph.edges)
     assert all(isinstance(signed_graph.edges[e]['weight'],
-                          (float, np.longfloat)) for e in signed_graph.edges)
+                          (float, np.longdouble)) for e in signed_graph.edges)
 
 
 def _weight_mapping(G):

indra/tests/test_pathfinding.py

@@ -71,7 +71,7 @@ def _setup_unsigned_graph():
     # Add belief
     for e in dg.edges:
         dg.edges[e]['belief'] = edge_beliefs[e]
-        dg.edges[e]['weight'] = -np.log(edge_beliefs[e], dtype=np.longfloat)
+        dg.edges[e]['weight'] = -np.log(edge_beliefs[e], dtype=np.longdouble)
 
     # Add edge_by_hash
     dg.graph['hashes'] = hashes