makeSynoyms.py

makeSynoyms.py

entrezIds = {}
inFile = open('mart_export.txt','r')
inFile.readline() # Get rid of header
while 1:
    line = inFile.readline()
    if not line:
        break
    splitUp = line.strip().split(',')
    if not splitUp[3]=='':
        if not splitUp[3] in entrezIds:
            entrezIds[splitUp[3]] = []
        entrezIds[splitUp[3]] += [i for i in splitUp if i and not entrezIds[splitUp[3]].count(i)]
inFile.close()

outFile = open('synonymThesaurus.csv','w')
outFile.write('\n'.join([i+','+';'.join(['"'+j+'"' for j in entrezIds[i]]) for i in entrezIds]))
outFile.close()