Reannotating halobacterium with GO annotations from: ftp://ftp.geneontology.org/pub/go/ontology/gene_ontology.obo

reannotate.py

goTerms = {}
inFile = open('gene_ontology.obo','r')
while 1:
    line = inFile.readline()
    if not line:
        break
    if line.strip()=='[Term]':
        id1 = inFile.readline().strip().split('id: ')[1]
        name = inFile.readline().strip().split('name: ')[1]
        namespace = inFile.readline().strip().split('namespace: ')[1]
        goTerms[id1] = { 'name': name, 'namespace': namespace}
    if line.strip()[0:6]=='alt_id':
        goTerms[line.strip().split('alt_id: ')[1]] = { 'name': name, 'namespace': namespace}

inFile.close()

inFile = open('halo_GO.csv','r')
outFile = open('halo_GO_annotations.csv','w')
line = inFile.readline() # Get rid of header
writeMe = ['geneId,goId,annotation,type']
while 1:
    line = inFile.readline()
    if not line:
        break
    line = line.strip()
    geneId = line.split(',')[0]
    goId = 'GO:'+line.split(',')[1]
    writeMe += [geneId+','+goId+','+goTerms[goId]['name'].replace(',',';')+','+goTerms[goId]['namespace']]
inFile.close()
outFile.write('\n'.join(writeMe))
outFile.close()