first verison

diseaseLabelGenerator.py

+from pymongo import MongoClient
+import re
+
+## init 
+client = MongoClient()
+db = client['arraymap']
+samples = db['samples_copy']
+
+
+## disease keywords 
+diseases = {'fanconi': ['DOID:13636','Fanconi anemia'],
+			'down': ['DOID:14250', 'Down syndrome'],
+			'ollier': ['DOID:4624', 'Ollier disease'],
+			'fraumeni': ['DOID:3012', 'Li-Fraumeni syndrome'],
+			'mafucci': ['DOID:0060221', 'Maffuccis syndrome'],
+			'richter': ['DOID:1703', 'Richter’s syndrome'],
+			'lynch': ['DOID:3883', 'Lynch syndrome'],
+			'fap': ['DOID:0050424', 'Familial Adenomatous Polyposis'],
+			'neurofibromatosis': ['DOID:8712', 'Neurofibromatosis']}
+
+
+## update the database
+for k,v in diseases.items():
+	keyword = k
+	ontology_term = v[0]
+	disease_name = v[1] 
+	samples.update_many({'DIAGNOSISTEXT':{'$regex': re.compile(keyword, re.IGNORECASE) }}, 
+		{'$set': {'DISEASE':[{'ONTOLOGY':ontology_term, 'PHENOTYPE':disease_name}]}})
+
+
+
+
+
+
+#print(samples.find({'DIAGNOSISTEXT':{'$regex': re.compile('lynch', re.IGNORECASE)}}).count())
+
+
+