def __init__(self, block_assignments, **kwargs):
"""Initialise CustomAssingmentQueryNameFlagger
Args:
block_assignments(list) : list of two letter codes to assign blocks to
"""
self.block_assignments = block_assignments
self.origin_colons = 7 # amount of ':' in original read name
# Verify if all of the assignments are 2 letters:
if not all((len(b) == 2 for b in block_assignments)):
for b in block_assignments:
if len(b) != 2:
raise ValueError(f'Tag {b} is not two letters long')
DigestFlagger.__init__(self, **kwargs)
def __init__(self, reference, **kwargs):
DigestFlagger.__init__(self, **kwargs)
self.overlap_tag = 'XM'
self.reference = reference
if self.reference is None:
raise ValueError('A reference fasta file is required!')
"""
z unmethylated C in CpG context (CG)
Z methylated C in CpG context (CG)
x unmethylated C in CHG context ( C[ACT]G )
X methylated C in CHG context ( C[ACT]G )
h unmethylated C in CHH context ( C[ACT][ACT] )
H methylated C in CHH context ( C[ACT][ACT] )
"""
self.context_mapping = {}
self.context_mapping[False] = {
'CGA': 'z',
'CGT': 'z',
'CGC': 'z',
'CGG': 'z',
# CHG:
'CAG': 'x',
'CCG': 'x',
'CTG': 'x',
}
self.context_mapping[True] = {
context: letter.upper()
for context, letter in self.context_mapping[False].items()
}
# CHH:
# h unmethylated C in CHH context ( C[ACT][ACT] )
for x in list(itertools.product('ACT', repeat=2)):
self.context_mapping[True][''.join(['C'] + list(x))] = 'H'
self.context_mapping[False][''.join(['C'] + list(x))] = 'h'
def __init__(self, **kwargs):
DigestFlagger.__init__(self, **kwargs)
def __init__(self, **kwargs):
self.assignedReadGroups = set()
DigestFlagger.__init__(self, **kwargs)
def __init__(self, **kwargs):
DigestFlagger.__init__(self, **kwargs)
if self.reference is None:
raise ValueError(
'The MSPJI digest flagger requires a reference genome file (FASTA)'
)
def __init__(self, tag=None, **kwargs):
DigestFlagger.__init__(self, **kwargs)
if tag is None or (tag != 'chrom' and len(tag) != 2):
raise ValueError(f'Invalid tag:{tag}')
self.tag = tag