cleanup

generate.py

 from pymongo import MongoClient
 import re
-# import argparse, sys
+# import argparse, sys, json
 
 client = MongoClient()
 db = client.arraymap
@@ -16,28 +16,23 @@ for sample in samples.find({}, {'UID': 1, 'BIOSAMPLEID': 1, 'SEGMENTS_HG18': 1})
     if ('SEGMENTS_HG18' in sample) and (sample['SEGMENTS_HG18'] is not None) and (len(sample['SEGMENTS_HG18']) > 1):
         callset_id = sample['UID']
         biosample_id = sample['BIOSAMPLEID']
+
         # check if BIOSAMPLEID has string & use this as 'biosample_id';
         # if not, create biosample_id as 'AM_BS__' + callset_id
         matchObj = re.search('^\w+.$', sample['BIOSAMPLEID'])
         if not matchObj:
             biosample_id = 'AM_BS__'+callset_id
+
         for seg in sample['SEGMENTS_HG18']:
             alternate_bases = ''
             if seg['SEGTYPE'] < 0:
                 alternate_bases = 'DEL'
             elif seg['SEGTYPE'] > 0:
                 alternate_bases = 'DUP'
+
             tag = str(seg['CHRO'])+'_'+str(seg['SEGSTART'])+'_'+str(seg['SEGSTOP'])+'_'+alternate_bases
-            # seg_type = int(seg['SEGTYPE'])
-            # start = int(seg['SEGSTART'])
-            # end = int(seg['SEGSTOP'])
-            # value = seg['SEGVALUE']
-            # chrom = int(seg['CHRO'])
-            # arraymap_id = sample['_id']
-            # call = {'type': seg_type, 'start': start, 'end': end, 'value': value,
-            #         'arraymap_id': arraymap_id, 'callset_id': callset_id}
             call = {'call_set_id': sample['UID'], 'biosample_id': biosample_id, 'VALUE': seg['SEGVALUE']}
-            #print(str(sample['_id']))
+
             if tag in variants:
                 variants[tag]['CALLS'].append(call)
                 callno += 1
@@ -65,3 +60,6 @@ for k,v in variants.items():
 	insert_id = db_variants.insert(v)
 
 print str(callno)+' calls were found for '+str(varid)+' variants'
+
+# with open('variants.json', 'w') as outfile:
+#     json.dump(variants, outfile, indent=4, sort_keys=True, separators=(',', ':'))

variants.json

+{
+    "10_46499590_47154881_DUP":{
+        "CALLS":[
+            {
+                "VALUE":"0.4956",
+                "biosample_id":"AM_BS__GSM253289",
+                "call_set_id":"GSM253289"
+            }
+        ],
+        "_id":
\ No newline at end of file