fix #28, missing attribute names and descriptions

GeneMANIA · Jul 17, 2015 · d1b948e · d1b948e
1 parent 0eb5db5
commit d1b948e
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 5 deletions.
diff --git a/builder/extract_attributes.py b/builder/extract_attributes.py
@@ -43,12 +43,11 @@ def extract_attributes(input_files, groups_file, output_file, organism_id,
         group_id = matched[0]
 
         attributes = pd.read_csv(input_file, sep='\t', header=False,
-                            names=['NAME', 'DESCRIPTION'],
+                            names=['EXTERNAL_ID', 'NAME', 'DESCRIPTION'],
                             dtype='str', na_filter=False)
 
         attributes['ATTRIBUTE_GROUP_ID'] = group_id
         attributes['ORGANISM_ID'] = organism_id
-        attributes['EXTERNAL_ID'] = attributes['NAME']
 
         all_attributes.append(attributes)
 
@@ -89,4 +88,4 @@ def extract_attributes(input_files, groups_file, output_file, organism_id,
         extract_attributes(args.inputs, args.groups, args.output, args.organism_id,
                            args.key_lstrip, args.key_rstrip)
     else:
-        raise Exception('unexpected command')
+        raise Exception('unexpected command')
diff --git a/builder/update_attribute_descriptions.py b/builder/update_attribute_descriptions.py
@@ -16,7 +16,7 @@ def main(attribute_file, description_file, output_file):
                           names=['GENE', 'ATTRIBUTE'])
 
     descs = pd.read_csv(description_file, sep='\t', header=None, na_filter=False,
-                        names=['ATTRIBUTE', 'DESCRIPTION'])
+                        names=['ATTRIBUTE', 'NAME', 'DESCRIPTION'])
 
     # remove duplicates in the descriptions
     descs.drop_duplicates(subset=['ATTRIBUTE'], inplace=True)
@@ -27,7 +27,7 @@ def main(attribute_file, description_file, output_file):
     attribs.drop_duplicates(inplace=True)
 
     # left merge, to grab only the needed descriptions
-    output = pd.merge(attribs[['ATTRIBUTE']], descs[['ATTRIBUTE', 'DESCRIPTION']],
+    output = pd.merge(attribs[['ATTRIBUTE']], descs[['ATTRIBUTE', 'NAME', 'DESCRIPTION']],
                       left_on='ATTRIBUTE', right_on='ATTRIBUTE', how='left')
 
     # empty string for missing descriptions