Templates

Scripts/Validation/Validate_Templates.py

@@ -8,7 +8,24 @@
 
 def get_id_set(tsv_path):
     with open(tsv_path) as infile:
-        return set([line.split('\t')[0] for line in infile])
+        ids = set()
+        obs = {}
+        for line in DictReader(infile, dialect='excel-tab'):
+            if 'is_obsolete' in line and int(line['is_obsolete']):
+                obs[line['id']] = line.get('linked_reaction', "").split(';')[0]
+                continue
+            if 'status' in line and ("MI" in line['status']):
+                continue
+            ids.add(line['id'])
+    return ids, obs
+
+
+def get_id_dict(tsv_path):
+    with open(tsv_path) as infile:
+        id_dict = {}
+        for line in DictReader(infile, dialect='excel-tab'):
+            id_dict[line['id']] = line
+    return id_dict
 
 
 def validate_biomass_compounds(path, comp_set):
@@ -32,6 +49,24 @@ def validate_reaction_list(path, rxn_set, complex_set):
             template_complexes.update(line['complexes'].strip().split("|"))
         return template_rxns - rxn_set, template_complexes - complex_set
 
+
+def remove_ids(path, ids):
+    ids = set(ids)
+    txt = open(path).readlines()
+    with open(path, 'w') as outfile:
+        outfile.writelines([x for x in txt if (x.split('\t')[0] not in ids)])
+
+def update_obsolete(path, obs_rxns):
+    txt = open(path).readlines()
+    with open(path, 'w') as outfile:
+        for line in txt:
+            rid = line.split('\t')[0]
+            if rid in obs_rxns and obs_rxns[rid]:
+                outfile.write(line.replace(rid, obs_rxns[rid]))
+            else:
+                outfile.write(line)
+
+
 if __name__ == '__main__':
     script_dir = os.path.dirname(__file__)
     parser = argparse.ArgumentParser(
@@ -48,10 +83,16 @@ def validate_reaction_list(path, rxn_set, complex_set):
     parser.add_argument('-t', dest='template_dir',
                         default=script_dir+'/../../Templates',
                         help='Path to the templates directory')
+    parser.add_argument('-d', dest='delete',
+                        default=False, action='store_true',
+                        help='Delete reactions which are invalid')
+    parser.add_argument('-u', dest='update',
+                        default=False, action='store_true',
+                        help='Update reactions which are invalid')
     args = parser.parse_args()
-    comp_ids = get_id_set(args.comp_tsv)
-    rxn_ids = get_id_set(args.rxn_tsv)
-    complex_ids = get_id_set(args.complex_tsv) | {'universal', 'null'}
+    comp_ids, obs_comps = get_id_set(args.comp_tsv)
+    rxn_ids, obs_rxns = get_id_set(args.rxn_tsv)
+    complex_ids = get_id_set(args.complex_tsv)[0] | {'universal', 'null'}
     exit_code = 0
     for template in os.listdir(args.template_dir):
         if not os.path.isdir(os.path.join(args.template_dir, template)):
@@ -65,16 +106,22 @@ def validate_reaction_list(path, rxn_set, complex_set):
             complex_ids)
 
         if undef_comps:
-            print("ERROR-Undefined Compounds: " + ", ".join(undef_comps),
-                  file=sys.stderr)
+            print("ERROR-%s Invalid Compounds: %s"
+                  % (len(undef_comps), ", ".join(undef_comps)))
             exit_code = 1
         if undef_rxns:
-            print("ERROR-Undefined Reactions: " + ", ".join(undef_rxns),
-                  file=sys.stderr)
+            print("ERROR-%s Invalid Reactions: %s"
+                  % (len(undef_rxns), ", ".join(undef_rxns)))
             exit_code = 1
         if undef_complex:
-            print("ERROR-Undefined Complexes: " + ", ".join(undef_complex),
-                  file=sys.stderr)
+            print("ERROR-%s Invalid Complexes: %s"
+                  % (len(undef_complex), ", ".join(undef_complex)))
             exit_code = 1
+        if args.delete and undef_comps:
+            remove_ids('%s/%s/BiomassCompounds.tsv' % (args.template_dir, template), undef_comps)
+        if args.delete and undef_rxns:
+            remove_ids('%s/%s/Reactions.tsv' % (args.template_dir, template), undef_rxns)
+        if args.update and undef_rxns:
+            update_obsolete('%s/%s/Reactions.tsv' % (args.template_dir, template), obs_rxns)
 
     exit(exit_code)