got resolve_dbxrefs.py working, still unstable

scripts/resolve_dbxrefs.py

@@ -4,24 +4,46 @@ import json
 import argparse
 import os
 import sys
-sys.path.insert(0, '../../dbxref/dbxref')
-sys.path.insert(0, '../../dbxref')
-import retriever
+import subprocess
 
 parser = argparse.ArgumentParser(description='Resolve dbxrefs from json file')
 parser.add_argument('--input', '-i', required=True, help='The json input document')
-#parser.add_argument('--output', '-o', required=True, help='The name of the output document')
+parser.add_argument('--scripts', '-s', required=True, help='Location of the retriever scripts')
+parser.add_argument('--output', '-o', required=True, help='Name of the output document')
 args = parser.parse_args()
 
-joined = {}
 json_file = open(args.input).read()
-json = json.loads(json_file)
+j = json.loads(json_file)
 
-for c in json['computations']:
-	li = c['results']['informations']
-	di = retriever.retrieve(li)
+def sort_refs(ref_list):
+	refs = {}
+	for ref in ref_list:
+		ble = ref.split(':')
+		if ble[0] in refs:
+			refs[ble[0]].append(ble[0] + ":" + ble[1])
+		else:
+			refs[ble[0]] = [ble[0] + ":" + ble[1]]
+	return (refs)
 
-#output_filename = args.output
-#with open(output_filename, 'w') as o:
-#	json.dump(joined, o)
+script_dir = args.scripts
+di = []
+
+for c in j['computations']:
+	for r in c['results']:
+		refs = sort_refs(r['informations'])
+		if 'EC' in refs:
+			di.append(json.loads(str(subprocess.check_output(script_dir + "retrieve_enzyme.py " + " ".join(refs['EC']), shell=True).rstrip())[2:-1]))
+		if 'GO' in refs:
+			di.append(json.loads(str(subprocess.check_output(script_dir + "retrieve_gene_ontology.py " + " ".join(refs['GO']), shell=True).rstrip())[2:-1]))
+#		if 'PFAM' in refs:
+#			di = json.loads(str(subprocess.check_output(script_dir + "retrieve_pfam.py " + " ".join(refs['PFAM']), shell=True).rstrip())[2:-1])
+		if 'SO' in refs:
+			di.append(json.loads(str(subprocess.check_output(script_dir + "retrieve_sequence_ontology.py " + " ".join(refs['SO']), shell=True).rstrip())[2:-1]))
+		if 'UniProtKB/Swiss-Prot' in refs:
+			di.append(json.loads(str(subprocess.check_output(script_dir + "retrieve_uniprot.py " + " ".join(refs['UniProtKB/Swiss-Prot']), shell=True).rstrip())[2:-1]))
+#		print (json.dumps(di))
+
+output_filename = args.output
+with open(output_filename, 'w') as o:
+	json.dump(di, o)