# Now that we have primary keys for each SNP, read in loci data
with open(curLociFilePath,'r') as csvfile:
data = csv.reader(csvfile,delimiter='\t')
for row in data:
if(len(row) == 4 and row[0] in documents):
# Load loci in Mongo documents
curDoc = documents[row[0]]
if curDoc["loci"] is None:
curDoc["loci"] = [{"mrna_acc":row[1],"gene":row[2],"class":row[3]}]
else:
curDoc["loci"].append({"mrna_acc":row[1],"gene":row[2],"class":row[3]})
documents[row[0]] = curDoc
# Data for reporting
result.lociLoadEnd = time.time()
result.totalDocuments = len(documents)
print "Starting to insert " + str(result.totalDocuments) + " documents"
# Log start time for MongoDB inserts
result.documentInsertStart = time.time()
if bulk:
print "Bulk insertion starting"
mongoCollection.insert(documents.values())
elif mongoimport:
mimpfile = "jsonchr" + str(curChr) + ".json"
print "Writing json file for mongoimport"
fp = open(mimpfile,'w')
for curDoc in documents.values():
json.dump(curDoc,fp)
