def getAUIMapCUI(self):
"""
Returns:
dict: {AUI: CUI}
"""
UMLSDataList = readMRCONSO_RRF(self.MRCONSO_RRF)
return {uTermDict['AUI']:uTermDict['CUI'] for uTermDict in UMLSDataList}
def getAUIMapDict(self):
"""实际上'eng'一栏亦为中文
Returns:
dict: {AUI: {'code': code, 'eng': ENG, 'cns': CNS)}
"""
UMLSDataList = readMRCONSO_RRF(DATA_PATH + '/umlsMT/umls/MRCONSO.RRF')
mapDict = {}
for uTermDict in UMLSDataList:
if uTermDict['LAT'] == 'CHI':
AUI = uTermDict['AUI']
code = uTermDict['CODE']
ENG = uTermDict['STR']
CNS = uTermDict['STR']
assert AUI not in mapDict # 检查是否有多个AUI映射到同一个hpo
mapDict[AUI] = ({'code': code, 'eng': ENG, 'cns': CNS})
return mapDict
def getAUIMapUmlsTerm(self):
UMLSDataList = readMRCONSO_RRF(self.MRCONSO_RRF)
return {uTermDict['AUI']: uTermDict['STR'] for uTermDict in UMLSDataList}
for termDict in dataList:
if termDict['ISPREF'] == 'Y':
countPref += 1
countPrefCharacter += len(termDict['STR'])
return countPref, countPrefCharacter
if __name__ == '__main__':
pass
from config import DATA_PATH, JSON_FILE_FORMAT
from common import getSaveFunc
from read.UMLS.ReadMrConso import readMRCONSO_RRF
sourcePath = DATA_PATH + '/umlsMT/umls/MRCONSO.RRF'
print('reading', sourcePath)
dataList = readMRCONSO_RRF(sourcePath)
print('read complete, total terms:', len(dataList))
# print 'counting charactersNum'
# print 'characters number:', countCharactersNum([termDict['STR'] for termDict in dataList])
# print 'examining AUI...'
# examineAUI(dataList)
# print 'done'
# print 'counting CUI'
# print 'CUI Number:', countCUI(dataList)
#
# print 'counting isPreffer'
# prefNum, prefCharacNum = countIsPref(dataList)
# print 'isPreffer number:', prefNum