def makeDNAannotation(variant, transcript, reference):
# Returning DNA level annotation if variant is a base substitution
if variant.isSubstitution():
if transcript.strand == 1:
return variant.ref + '>' + variant.alt, ''
else:
return variant.ref.reverseComplement(
) + '>' + variant.alt.reverseComplement(), ''
# Returning DNA level annotation if variant is an insertion
if variant.isInsertion():
insert = core.Sequence(variant.alt)
if transcript.strand == 1:
before = reference.getReference(variant.chrom,
variant.pos - len(insert),
variant.pos - 1)
# Checking if variant is a duplication
if insert == before and (variant.pos - len(insert) >=
transcript.transcriptStart):
if len(insert) > 4:
return 'dup' + str(len(insert)), 'ins' + insert
return 'dup' + insert, 'ins' + insert
else:
return 'ins' + insert, ''
else:
before = reference.getReference(variant.chrom, variant.pos,
variant.pos + len(insert) - 1)
# Checking if variant is a duplication
if insert == before and (variant.pos + len(insert) - 1 <=
transcript.transcriptEnd):
if len(insert) > 4:
return 'dup' + str(
len(insert)), 'ins' + insert.reverseComplement()
return 'dup' + insert.reverseComplement(
), 'ins' + insert.reverseComplement()
else:
return 'ins' + insert.reverseComplement(), ''
# Returning DNA level annotation if variant is a deletion
if variant.isDeletion():
if len(variant.ref) > 4:
return 'del' + str(len(variant.ref)), ''
if transcript.strand == 1:
return 'del' + variant.ref, ''
else:
return 'del' + core.Sequence(variant.ref).reverseComplement(), ''
# Returning DNA level annotation if variant is a complex indel
if variant.isComplex():
if transcript.strand == 1:
return 'delins' + variant.alt, ''
else:
return 'delins' + variant.alt.reverseComplement(), ''
def getReference(self, chrom, start, end):
# Checking if chromosome name exists
goodchrom = chrom
if not goodchrom in self.fastafile.references:
goodchrom = 'chr' + chrom
if not goodchrom in self.fastafile.references:
if chrom == 'MT':
goodchrom = 'chrM'
if not goodchrom in self.fastafile.references: return None
else:
return None
# Fetching data from reference genome
if end < start: return core.Sequence('')
if start < 1: start = 1
if pysam.__version__ in ['0.7.7', '0.7.8', '0.8.0']:
last = self.fastafile.getReferenceLength(goodchrom)
else:
last = self.fastafile.get_reference_length(goodchrom)
if end > last: end = last
seq = self.fastafile.fetch(goodchrom, start - 1, end)
return core.Sequence(seq.upper())