Adapted blast json document to the signalp document structure

bin/convert_blast_swissprot.py

@@ -5,7 +5,8 @@ import json
 filename = sys.argv[1]
 documents = {}
 with open(filename) as f:
-    header = {}
+    header = None
+    tool = None
     for line in f:
         line = line.strip()
         if line.startswith('#'):
@@ -14,10 +15,19 @@ with open(filename) as f:
                 header = {}
                 for idx, key in enumerate(header_entries):
                     header[key] = idx
+            if line.startswith('# BLASTP'):
+                split = line.split()
+                tool = {'name': 'blastp',
+                        'version': split[2]}
+            if line.startswith('# Database'):
+                split = line.split()
+                tool['database'] = split[2]
+
         else:
             split = line.split("\t")
             if not split[0] in documents:
-                documents[split[0]] = {"id": split[0], "results": []}
+                documents[split[0]] = {"id": split[0], "computations": [{'tool':tool, 'results':[]}]}
+            results = documents[split[0]]['computations'][0]['results']
 
             result = {}
             result["dbxref"] = "UniProtKB/Swiss-Prot:"+split[header['subject id']].split("|")[1]
@@ -32,7 +42,7 @@ with open(filename) as f:
             if 'BTOP' in header:
                 result['btop'] = split[header['BTOP']]
 
-            documents[split[0]]["results"].append(result)
+            results.append(result)
 
 
 for key in documents: