Add dbxref to ghostx converter. Adapt to blastp and hmmscan json format (Fixes #19)

bin/convert_ghostx.py

@@ -6,6 +6,9 @@ import argparse
 parser = argparse.ArgumentParser(description='Convert ghostx results to json documents')
 parser.add_argument('--result', '-r', required=True, help='The ghostx result directory')
 parser.add_argument('--output', '-o', required=True, help='The converted results json file')
+parser.add_argument('--dbxref', '-d', required=True, help='The dbxref prefix that will be prepended to the accession or id')
+parser.add_argument('--acc_split', '-s', required=False, help='The delimiter used to split the subject id (optional)')
+parser.add_argument('--acc_position', '-p', type=int, required=False, help='The position of the accession after splitting the subject_id (optional)')
 args = parser.parse_args()
 
 result_directory = args.result
@@ -27,11 +30,22 @@ with open(result_filename) as f:
         results = documents[split[0]]['computations'][0]['results']
 
         result = {}
-        result["dbxref"] = "UniProtKB/Swiss-Prot:"+split[1].split("|")[1]
-        result["percent_identity"] = float(split[2])
-        result['qloc'] = split[6] + '-' + split[7]
-        result['sloc'] = split[8] + '-' + split[9]
-        result['evalue'] = float(split[10])
+        accession = split[1]
+        if args.acc_split and args.acc_position:
+            accession = accession.split(args.acc_split)[args.acc_position]
+        elif args.acc_split:
+            accession = acession.split(args.acc_split)[0]
+        result['query'] = {
+                'start': split[6],
+                'end': split[7]
+                }
+        result['target'] = {
+                'dbxref': args.dbxref + ':' + accession,
+                'start': split[8],
+                'end': split[9],
+                'evalue': float(split[10]),
+                'percent_identity': float(split[2])
+                }
 
         results.append(result)
 

modules/ghostx_vs_swissprot.yaml

@@ -17,4 +17,7 @@ analysis:
 # result file from the analysis_script
 converter:
     script: 'convert_ghostx.py'
-
+    parameters:
+        dbxref: 'UniProtKB/Swiss-Prot'
+        acc_split: '|'
+        acc_position: '1'