retrieve go identifiers for molecular functions from Uniprot

ippisite/ippidb/ws.py

@@ -42,10 +42,17 @@ def get_uniprot_info(uniprot_id):
     organism = resp.root.findall('{http://uniprot.org/uniprot}entry/{http://uniprot.org/uniprot}organism/{http://uniprot.org/uniprot}dbReference[@type="NCBI Taxonomy"]')[0].attrib['id']
     gene = resp.root.findall('{http://uniprot.org/uniprot}entry/{http://uniprot.org/uniprot}gene/{http://uniprot.org/uniprot}name[@type="primary"]')[0].text
     entry_name = resp.root.findall('{http://uniprot.org/uniprot}entry/{http://uniprot.org/uniprot}name')[0].text
+    go_els = resp.root.findall('{http://uniprot.org/uniprot}entry/{http://uniprot.org/uniprot}dbReference[@type="GO"]')
+    molecular_functions = []
+    for go_el in go_els:
+        term_property_value = go_el.findall('{http://uniprot.org/uniprot}property[@type="term"]')[0].attrib['value']
+        if term_property_value[0:2]=='F:':
+            molecular_functions.append('GO_'+go_el.attrib['id'][3:])
     return {'recommended_name': recommended_name,
             'organism': int(organism),
             'gene': gene,
-            'entry_name': entry_name
+            'entry_name': entry_name,
+            'molecular_functions': molecular_functions
            }
 
 def get_taxonomy_info(taxonomy_id):