def __init__(self, inFile):
# print inFile
self.isSeql = False
self.charEncoding = 'not known'
try:
with open(inFile, 'r') as f:
self.charEncoding = chardet.detect(f.read())['encoding']
self.seqlGenerator = su.generateChunks(inFile, pa)
generalInformationString = self.seqlGenerator.next()['chunk']
generalInformationUnicode = generalInformationString.decode(self.charEncoding)
# self.generalInformation = GeneralInformation(generalInformationString)
self.generalInformation = GeneralInformation(generalInformationUnicode)
if self.generalInformation.genInfoPatternFound:
self.isSeql = True
# print 'SequenceListing initialised for file', inFile
except IOError:
# self.logger.exception("Invalid input file: %s" % self.in_file_name)
print 'Invalid file name', inFile
self.quantity_nuc = 0
self.quantity_prt = 0
self.quantity_mix = 0
def __init__(self, inFile):
# print inFile
self.isSeql = False
try:
self.seqlGenerator = su.generateChunks(inFile, pa)
generalInformationString = self.seqlGenerator.next()['chunk']
self.generalInformation = GeneralInformation(generalInformationString)
if self.generalInformation.genInfoPatternFound:
self.isSeql = True
# print 'SequenceListing initialised for file', inFile
except IOError:
# self.logger.exception("Invalid input file: %s" % self.in_file_name)
print 'Invalid file name', inFile
def getSequenceFromFile(cls, aFile, aSeqIdNo):
'''
Return the Sequence instance corresponding to aSeqIdNo in aFile.
The assumption is that seqIdNo = actualSeqIdNo.
:param aFile: a sequence listing file
:param aSeqIdNo: int representing the seq id no
:return: Sequence
'''
# TODO: test it
listOfChunks = []
for c in su.generateChunks(aFile, pa):
listOfChunks.append(c)
inputString = listOfChunks[aSeqIdNo]['chunk']
seq = Sequence(inputString)
seq.actualSeqIdNo = aSeqIdNo#TODO: to see if this is OK
return seq
def getSequenceFromFile(cls, aFile, aSeqIdNo):
'''
Return the Sequence instance corresponding to aSeqIdNo in aFile.
The assumption is that seqIdNo = actualSeqIdNo.
:param aFile: a sequence listing file
:param aSeqIdNo: int representing the seq id no
:return: Sequence
'''
# TODO: test it, unicode it!
listOfChunks = []
for c in su.generateChunks(aFile, pa):
listOfChunks.append(c)
inputString = listOfChunks[aSeqIdNo]['chunk']
seq = Sequence(inputString)
seq.actualSeqIdNo = aSeqIdNo#TODO: to see if this is OK
return seq
