comparison baseline/script_imgt.py @ 63:8728284105ee draft

Uploaded

62:aa8d37bd1930

parser.add_argument("--output", help="Output file")
parser.add_argument("--id", help="ID to be used at the '>>>' line in the output")

args = parser.parse_args()

refdic = dict()
with open(args.ref, 'rU') as ref:
	currentSeq = ""
	currentId = ""
	for line in ref:
		if line[0] is ">":
			if currentSeq is not "" and currentId is not "":
				refdic[currentId[1:]] = currentSeq
			currentId = line.rstrip()
			currentSeq = ""
		else:
			currentSeq += line.rstrip()
	refdic[currentId[1:]] = currentSeq
	

vPattern = [r"(IGHV[0-9]-[0-9ab]+-?[0-9]?D?\*\d{1,2})"]#,
#						r"(TRBV[0-9]{1,2}-?[0-9]?-?[123]?)",
#						r"(IGKV[0-3]D?-[0-9]{1,2})",
#						r"(IGLV[0-9]-[0-9]{1,2})",

#vPattern = re.compile(r"|".join(vPattern))
vPattern = re.compile("|".join(vPattern))

def filterGene(s, pattern):
	s1 = s[s.find(" ") + 1:]
	return s1[:s1.find(" ")]
	"""
    if type(s) is not str:
        return None
    res = pattern.search(s)
    if res:
        return res.group(0)
    return None
	"""


currentSeq = ""
currentId = ""
first=True

63:8728284105ee

parser.add_argument("--output", help="Output file")
parser.add_argument("--id", help="ID to be used at the '>>>' line in the output")

args = parser.parse_args()

print "script_imgt.py"
print "input:", args.input
print "ref:", args.ref
print "output:", args.output
print "id:", args.id

refdic = dict()
with open(args.ref, 'rU') as ref:
	currentSeq = ""
	currentId = ""
	for line in ref:
		if line.startswith(">"):
			if currentSeq is not "" and currentId is not "":
				refdic[currentId[1:]] = currentSeq
			currentId = line.rstrip()
			currentSeq = ""
		else:
			currentSeq += line.rstrip()
	refdic[currentId[1:]] = currentSeq

print "Have", str(len(refdic)), "reference sequences"

vPattern = [r"(IGHV[0-9]-[0-9ab]+-?[0-9]?D?\*\d{1,2})"]#,
#						r"(TRBV[0-9]{1,2}-?[0-9]?-?[123]?)",
#						r"(IGKV[0-3]D?-[0-9]{1,2})",
#						r"(IGLV[0-9]-[0-9]{1,2})",

#vPattern = re.compile(r"|".join(vPattern))
vPattern = re.compile("|".join(vPattern))

def filterGene(s, pattern):
    if type(s) is not str:
        return None
    res = pattern.search(s)
    if res:
        return res.group(0)
    return None



currentSeq = ""
currentId = ""
first=True