def getDataFromUniprot(fasta, accs):
global fetchme
uniprot_file = fasta + ".uniprot"
print uniprot_file
lastline = tail(uniprot_file, 1)
lastline = lastline[0]
if lastline:
lastline = lastline.rstrip()
print "::::::::::: " + lastline
if lastline != "</uniprot>":
with open(uniprot_file, "a") as f:
f.write("</uniprot>\n")
handle = open(uniprot_file)
cntr = 0
proteins = {}
for record in UniprotIterator(handle):
print record.id, record.name
proteins[record.name] = Protein(xmlrecord=record)
cntr = cntr + 1
# if cntr > 1000:
# break
if accs != None:
accs2get = [acc for acc in accs if acc not in proteins.keys()]
if len(accs2get) > 0:
print "fetching: "
print accs2get
if fetchme:
fetchUniprotEntries(fasta, accs2get)
return getDataFromUniprot(fasta, None)
print "There are total of " + str(len(proteins)) + " entries mapped in uniprot"
return proteins
def fetchUniprotEntries(fasta, newaccs=None):
print "Fetching uniprot entries.."
uniprot_file = fasta + ".uniprot"
already_exists = False
if os.path.isfile(uniprot_file):
already_exists = True
lastline = tail(uniprot_file, 1)
lastline = lastline[0]
lastline = lastline.rstrip()
#TODO: change head linux use to the appropriate python command
if lastline == "</uniprot>":
lines = file(uniprot_file, 'r').readlines()
del lines[-1]
file(uniprot_file, 'w').writelines(lines)
#cmd = "head -n -1 " + uniprot_file + " > " + uniprot_file + ".2; " + "mv " + uniprot_file + ".2 " + uniprot_file
#print cmd
#os.system(cmd)
fasta_ids = newaccs
if newaccs == None:
fasta_ids = getACCsFromFastaFile(fasta)
print fasta_ids
if fasta_ids:
l = range(0, len(fasta_ids))
chunks = FetchUtils.getChunks(l, 50)
for i in range(0, len(chunks)):
print "Fetching " + str(i) + " chunk out of " + str(len(chunks))
sys.stdout.flush()
accs = [fasta_ids[j] for j in chunks[i]]
# fetchBatch <dbName> <id1,id2,...> [formatName [styleName]] [options...]
try:
uniprots = FetchUtils.returnFetchBatch('uniprot', ','.join(accs), 'xml', 'default')
with open(fasta + ".uniprot", "a") as f:
if not i == 0 or already_exists:
uniprots = re.sub(r"\<uniprot.+", '', uniprots)
uniprots = re.sub(r"\<\?xml version=.+\>", '', uniprots)
if not i == len(chunks) - 1:
uniprots = re.sub(r"\</uniprot.*>", '', uniprots)
f.write(uniprots)
except:
print "Coundnot fetch this chunk"
continue
else:
print "succeeded in fetching chunk"
continue
return (fasta_ids)
