Include mature sequence in json output if available.

scripts/convert_signalp.py

@@ -3,13 +3,26 @@
 import sys
 import json
 import argparse
+from os import path
 
 parser = argparse.ArgumentParser(description='Convert signalp results to json document')
 parser.add_argument('--result', '-r', required=True, help='The signalp result directory')
 parser.add_argument('--output', '-o', required=True, help='The converted results json file')
 args = parser.parse_args()
 
-# Check wether args.result + '/mature.fas' exists and if so read into id:seq dict
+mature_file = args.result + '/mature.fas'
+include_mature = path.isfile(mature_file)
+mature = {}
+
+if include_mature:
+    with open(mature_file) as m:
+        id = None
+        for line in m:
+            if line.startswith(">"):
+                id = line.split()[0][1:]
+                mature[id] = ''
+            elif id is not None:
+                mature[id] += line.strip()
 
 filename = args.result + '/results.txt'
 documents = {}
@@ -21,15 +34,23 @@ with open(filename) as f:
             split = line.split()
             if not split[0] in documents:
                 documents[split[0]] = {"id": split[0], "computations": [{'tool': tool, 'results':[]}]}
-            results =  documents[split[0]]['computations'][0]["results"]
+            results = documents[split[0]]['computations'][0]["results"]
 
             if split[9] == "Y":
                 results.append({
                     'signalpeptide': True,
                     'score': float(split[8]),
-                    'start':1,
+                    'start': 1,
                     'end': int(split[2])-1
                     })
+
+                if include_mature:
+                    mature_seq = ''
+                    try:
+                        mature_seq = mature[split[0]]
+                    except ValueError:
+                        print('SignalP Converter: No mature sequence received for protein: {}, although signal peptide present!'.format(split[0]))
+                    results[-1]['mature sequence'] = mature_seq
             else:
                 results.append({'signalpeptide': False})
         else: