def _score_splice_site(seq, splicetype="donor"):
"""
Return PSSM splice site score
@type seq: string
@param seq: DNA sequence of EXACT length of the PSSM
@type splicetype: string
@param splicetype: 'donor' or 'acceptor'
@rtype: float
@return: PSSM score for this splice site
"""
if splicetype == 'acceptor':
return pssmscore(seq, IC_ACCEPTOR)
else:
return pssmscore(seq, IC_DONOR)
def scan_pssm_splice_site(seq,
splicetype="donor",
override_pattern_offset=(),
min_pssm_score=None,
allow_non_canonical=False,
non_canonical_min_pssm_score=0.0):
"""
Find splice sites by a PSSM on input sequence
@type seq: string
@param seq: DNA sequence of EXACT length of the PSSM
@type splicetype: string
@param splicetype: 'donor' or 'acceptor'
@type min_pssm_score: float
@param min_pssm_score:
@type allow_non_canonical: boolean
@param allow_non_canonical: True of False
@type non_canonical_min_pssm_score: float
@param non_canonical_min_pssm_score:
@type override_pattern_offset: tuple
@param override_pattern_offset: tuple with 2 integers; use cautiously!!
@rtype: list
@return: list with SpliceDonors or SpliceAcceptors
"""
if splicetype == 'acceptor':
PSSM_MATRIX = IC_ACCEPTOR
pattern_offset = IC_ACCEPTOR_PATTERN_OFFSET
canonical = "AG"
else:
PSSM_MATRIX = IC_DONOR
pattern_offset = IC_DONOR_PATTERN_OFFSET
canonical = "GT"
if allow_non_canonical:
# obtain PSSM_IC for non-canonical (GC) donors
IC_NCGC_DONOR = parse_ic_file(IC_DONOR_NCGC_DATA_FILE)
# hmm... somebody knows what he or she is doing ;-)
if override_pattern_offset:
pattern_offset = override_pattern_offset
pssmlength = len(PSSM_MATRIX)
sites = []
for offset in range(0, len(seq) - pssmlength + 1):
# get sequence slice of pattern and actual splice site
seqpart = seq[offset:offset + pssmlength].upper()
splicesite = seqpart[pattern_offset[0]:-pattern_offset[1]]
# continue if non-canonical sites if not requested for
if not allow_non_canonical and splicesite != canonical:
continue
elif splicesite == canonical:
# score this splicesite
score = _score_splice_site(seqpart, splicetype=splicetype)
# check if site must be stored
if min_pssm_score or min_pssm_score == 0.0:
if score < min_pssm_score:
continue
elif splicesite != canonical and splicetype == 'donor':
# score non-canonical donor site
score = pssmscore(seqpart, IC_NCGC_DONOR)
# check if site must be stored
if non_canonical_min_pssm_score or non_canonical_min_pssm_score == 0.0:
if score < non_canonical_min_pssm_score:
continue
else:
continue
if splicetype == 'acceptor':
a = SpliceAcceptor(offset,
seqpart,
acceptor=splicesite,
pssm_score=score)
sites.append(a)
else:
d = SpliceDonor(offset,
seqpart,
donor=splicesite,
pssm_score=score)
sites.append(d)
# return sites for Donor
if splicetype == 'donor':
sites.reverse()
# and return
return sites