def cigar2variants(cigar, sequence, tag):
"""From cigar to Variants in GFF format"""
pos = 0
iso5p = 0
logger.debug("\nISOMIRSEA:: expanded: %s" % expand_cigar(cigar))
for l in expand_cigar(cigar):
if l == "I":
iso5p -= 1
elif l == "D":
iso5p += 1
else:
break
iso3p = 0
for l in reversed(expand_cigar(cigar)):
if l == "I":
iso3p += 1
elif l == "D":
iso3p -= 1
else:
break
isosnp = []
for l in expand_cigar(cigar):
if l in ['A', 'T', 'C', 'G']:
isosnp.append([pos, sequence[pos], l])
if l in ['D']:
continue
pos += 1
iso5p = "iso_5p:%s" % _fix(iso5p) if iso5p else ""
if tag[-1] == "T" or iso3p < 0:
iso3p = "iso_3p:%s" % _fix(iso3p) if iso3p else ""
else:
iso3p = "iso_add3p:%s" % iso3p if iso3p else ""
variant = ""
for iso in [iso5p, iso3p, _define_snp(isosnp)]:
if iso:
variant += "%s," % iso
variant = "NA;" if not variant else variant
return variant[:-1]
def cigar2variants(cigar, sequence, tag):
"""From cigar to Variants in GFF format"""
pos = 0
iso5p = 0
logger.debug("\nISOMIRSEA:: expanded: %s" % expand_cigar(cigar))
for l in expand_cigar(cigar):
if l == "I":
iso5p += 1
elif l == "D":
iso5p -= 1
else:
break
iso3p = 0
for l in reversed(expand_cigar(cigar)):
if l == "I":
iso3p += 1
elif l == "D":
iso3p -= 1
else:
break
isosnp = []
for l in expand_cigar(cigar):
if l in ['A', 'T', 'C', 'G' ]:
isosnp.append([pos, sequence[pos], l])
if l in ['D']:
continue
pos += 1
iso5p = "iso_5p:%s" % _fix(iso5p) if iso5p else ""
if tag[-1] == "T" or iso3p < 0:
iso3p = "iso_3p:%s" % _fix(iso3p) if iso3p else ""
else:
iso3p = "iso_add:%s" % _fix(iso3p) if iso3p else ""
variant = ""
for iso in [iso5p, iso3p, _define_snp(isosnp)]:
if iso:
variant += "%s," % iso
variant = "NA;" if not variant else variant
return variant[:-1]
def test_cigar(self):
"""testing cigar correction function"""
cigar = [[0, 14], [1, 1], [0, 5]]
from mirtop.mirna.realign import cigar_correction, make_cigar, \
cigar2snp, expand_cigar
fixed = cigar_correction(cigar, "AAAAGCTGGGTTGAGGAGGA",
"AAAAGCTGGGTTGAGAGGA")
if not fixed[0] == "AAAAGCTGGGTTGAGGAGGA":
raise ValueError("Sequence 1 is not right.")
if not fixed[1] == "AAAAGCTGGGTTGA-GAGGA":
raise ValueError("Sequence 2 is not right.")
if not make_cigar("AAA-AAATAAA", "AGACAAA-AAA") == "MAMD3MI3M":
raise ValueError("Cigar not eq to MAMD3MI3M: %s" %
make_cigar("AAA-AAATAAA", "AGACAAA-AAA"))
# test expand cigar
if not expand_cigar("3MA3M") == "MMMAMMM":
raise ValueError("Cigar 3MA3M not eqaul to MMMAMMM but to %s" %
expand_cigar("3MA3M"))
# test cigar to snp
if not cigar2snp("3MA3M", "AAATCCC")[0] == [3, "A", "T"]:
raise ValueError("3MA3M not equal AAATCCC but %s" %
cigar2snp("3MA3M", "AAATCCC"))
def test_cigar(self):
"""testing cigar correction function"""
cigar = [[0, 14], [1, 1], [0, 5]]
from mirtop.mirna.realign import cigar_correction, make_cigar, \
cigar2snp, expand_cigar
fixed = cigar_correction(cigar, "AAAAGCTGGGTTGAGGAGGA",
"AAAAGCTGGGTTGAGAGGA")
if not fixed[0] == "AAAAGCTGGGTTGAGGAGGA":
raise ValueError("Sequence 1 is not right.")
if not fixed[1] == "AAAAGCTGGGTTGA-GAGGA":
raise ValueError("Sequence 2 is not right.")
if not make_cigar("AAA-AAATAAA", "AGACAAA-AAA") == "MAMD3MI3M":
raise ValueError("Cigar not eq to MAMD3MI3M: %s" %
make_cigar("AAA-AAATAAA", "AGACAAA-AAA"))
# test expand cigar
if not expand_cigar("3MA3M") == "MMMAMMM":
raise ValueError("Cigar 3MA3M not eqaul to MMMAMMM but to %s" %
expand_cigar("3MA3M"))
# test cigar to snp
if not cigar2snp("3MA3M", "AAATCCC")[0] == [3, "A", "T"]:
raise ValueError("3MA3M not equal AAATCCC but %s" %
cigar2snp("3MA3M", "AAATCCC"))