def __repr__(self):
if self.name is None:
return "ChromosomeInterval('{}', {}, {}, '{}')".format(self.chromosome, self.start, self.stop,
convert_strand(self.strand))
else:
return "ChromosomeInterval('{}', {}, {}, '{}', '{}')".format(self.chromosome, self.start, self.stop,
convert_strand(self.strand), self.name)
def __init__(self, gene_pred_tokens):
# Text genePred fields
self.name = gene_pred_tokens[0]
self.chromosome = gene_pred_tokens[1]
self.strand = convert_strand(gene_pred_tokens[2])
# Integer genePred fields
self.score = 0 # no score in genePred files
self.thick_start = int(gene_pred_tokens[5])
self.thick_stop = int(gene_pred_tokens[6])
self.start = int(gene_pred_tokens[3])
self.stop = int(gene_pred_tokens[4])
self.rgb = "128,0,0" # no RGB in genePred files
# genePred specific fields
self.id = gene_pred_tokens[10]
self.name2 = gene_pred_tokens[11]
self.cds_start_stat = gene_pred_tokens[12]
self.cds_end_stat = gene_pred_tokens[13]
self.exon_frames = [int(x) for x in gene_pred_tokens[14].split(",") if x != ""]
# convert genePred format coordinates to BED-like coordinates to make intervals
self.block_count = gene_pred_tokens[7]
block_starts = [int(x) for x in gene_pred_tokens[8].split(",") if x != ""]
block_ends = [int(x) for x in gene_pred_tokens[9].split(",") if x != ""]
self.block_sizes = ",".join(map(str, [e - s for e, s in izip(block_ends, block_starts)]))
self.block_starts = ",".join(map(str, [x - self.start for x in block_starts]))
bed_tokens = [gene_pred_tokens[1], self.start, self.stop, self.name, self.score, gene_pred_tokens[2],
self.thick_start, self.thick_stop, self.rgb, self.block_count,
self.block_sizes, self.block_starts]
# build chromosome intervals for exons and introns
self.exon_intervals = self._get_exon_intervals(bed_tokens)
self.intron_intervals = self._get_intron_intervals()
# build Exons mapping transcript space coordinates to chromosome
self.exons = self._get_exons(bed_tokens)
# calculate sizes
self._get_cds_size()
self._get_size()
def __init__(self, chromosome, start, stop, strand, name=None):
self.chromosome = str(chromosome)
assert start <= stop
self.start = int(start) # 0 based
self.stop = int(stop) # exclusive
if strand not in [True, False, None]:
strand = convert_strand(strand)
self.strand = strand # True or False
self.name = name
def chromosome_region_to_bed(t, start, stop, rgb, name):
"""
This is different from chromosome_coordinate_to_bed - this function will not resize the BED information
for the input transcript, but instead be any coordinate on the chromosome.
"""
strand = convert_strand(t.strand)
chrom = t.chromosome
assert start is not None and stop is not None, (t.name, start, stop, name)
assert stop >= start, (t.name, start, stop, name)
return [chrom, start, stop, name + "/" + t.name, 0, strand, start, stop, rgb, 1, stop - start, 0]
def splice_intron_interval_to_bed(t, intron_interval, rgb, name):
"""
Specific case of turning an intron interval into the first and last two bases (splice sites)
"""
interval = intron_interval
assert interval.stop >= interval.start, (t.name, t.chromosome)
assert interval.stop - interval.start - 2 > 2, (t.name, t.chromosome)
block_starts = "0,{}".format(interval.stop - interval.start - 2)
return [interval.chromosome, interval.start, interval.stop, "/".join([name, t.name]), 0,
convert_strand(interval.strand), interval.start, interval.stop, rgb, 2, "2,2", block_starts]
def _get_exon_intervals(self, bed_tokens):
"""
Gets a list of exon intervals in chromosome coordinate space.
These exons are on (+) strand ordering regardless of transcript strand.
This means (-) strand genes will be represented backwards
"""
exons = []
start, stop = int(bed_tokens[1]), int(bed_tokens[2])
chrom, strand = bed_tokens[0], convert_strand(bed_tokens[5])
block_sizes = [int(x) for x in bed_tokens[10].split(",") if x != ""]
block_starts = [int(x) for x in bed_tokens[11].split(",") if x != ""]
for block_size, block_start in izip(block_sizes, block_starts):
exons.append(ChromosomeInterval(chrom, start + block_start, start + block_start + block_size, strand))
return exons
def __init__(self, bed_tokens):
self.chromosome = bed_tokens[0]
self.start = int(bed_tokens[1])
self.stop = int(bed_tokens[2])
self.name = bed_tokens[3]
self.score = int(bed_tokens[4])
self.strand = convert_strand(bed_tokens[5])
self.thick_start = int(bed_tokens[6])
self.thick_stop = int(bed_tokens[7])
self.rgb = bed_tokens[8]
self.block_count = bed_tokens[9]
self.block_sizes = bed_tokens[10]
self.block_starts = bed_tokens[11]
# build chromosome intervals for exons and introns
self.exon_intervals = self._get_exon_intervals(bed_tokens)
self.intron_intervals = self._get_intron_intervals()
# build Exons mapping transcript space coordinates to chromosome
self.exons = self._get_exons(bed_tokens)
# calculate sizes
self._get_cds_size()
self._get_size()
def get_bed(self, rgb, name):
"""
Returns BED tokens representing this interval. Requires a name and a rgb value. BED is BED12.
"""
return [self.chromosome, self.start, self.stop, name, 0, convert_strand(self.strand), self.start, self.stop,
rgb, 1, len(self), 0]
def interval_to_bed(t, interval, rgb, name):
"""
If you are turning interval objects into BED records, look here. t is a transcript object.
Interval objects should always have start <= stop (+ strand chromosome ordering)
"""
assert interval.stop >= interval.start, (t.name, t.chromosome)
return [interval.chromosome, interval.start, interval.stop, name + "/" + t.name, 0, convert_strand(interval.strand),
interval.start, interval.stop, rgb, 1, interval.stop - interval.start, 0]
def get_bed(self, rgb=None, name=None, start_offset=None, stop_offset=None):
"""
Returns this transcript as a BED record with optional changes to rgb and name.
If start_offset or stop_offset are set (chromosome coordinates), then this record will be changed to only
show results within that region, which is defined in chromosome coordinates.
"""
if start_offset is not None and stop_offset is not None:
assert start_offset <= stop_offset
if start_offset is not None:
assert start_offset >= self.start
if stop_offset is not None:
assert stop_offset <= self.stop
if rgb is None:
rgb = self.rgb
if name is not None:
name += "/" + self.name
else:
name = self.name
if start_offset is None and stop_offset is None:
return [self.chromosome, self.start, self.stop, name, self.score, convert_strand(self.strand),
self.thick_start, self.thick_stop, rgb, self.block_count, self.block_sizes, self.block_starts]
elif start_offset == stop_offset:
assert self.chromosome_coordinate_to_transcript(start_offset) is not None # no intron records
return [self.chromosome, start_offset, stop_offset, name, self.score, convert_strand(self.strand),
start_offset, stop_offset, rgb, 1, 0, 0]
def _move_start(exon_intervals, block_count, block_starts, block_sizes, start, start_offset):
to_remove = len([x for x in exon_intervals if x.start <= start_offset and x.stop <= start_offset])
assert to_remove < len(exon_intervals)
if to_remove > 0:
block_count -= to_remove
block_sizes = block_sizes[to_remove:]
start += block_starts[to_remove]
new_block_starts = [0]
for i in xrange(to_remove, len(block_starts) - 1):
new_block_starts.append(block_starts[i + 1] - block_starts[i] + new_block_starts[-1])
block_starts = new_block_starts
if start_offset > start:
block_sizes[0] += start - start_offset
block_starts[1:] = [x + start - start_offset for x in block_starts[1:]]
start = start_offset
return start, block_count, block_starts, block_sizes
def _move_stop(exon_intervals, block_count, block_starts, block_sizes, stop, start, stop_offset):
to_remove = len([x for x in exon_intervals if x.stop >= stop_offset and x.start >= stop_offset])
assert to_remove < len(exon_intervals)
if to_remove > 0:
block_count -= to_remove
block_sizes = block_sizes[:-to_remove]
block_starts = block_starts[:-to_remove]
assert len(block_sizes) == len(block_starts)
if len(block_sizes) == 0:
block_sizes = block_starts = [0]
block_count = 1
stop = start + block_sizes[-1] + block_starts[-1]
if start + block_starts[-1] < stop_offset < stop:
block_sizes[-1] = stop_offset - start - block_starts[-1]
stop = stop_offset
return stop, block_count, block_starts, block_sizes
block_count = int(self.block_count)
block_starts = map(int, self.block_starts.split(","))
block_sizes = map(int, self.block_sizes.split(","))
start = self.start
stop = self.stop
thick_start = self.thick_start
thick_stop = self.thick_stop
if start_offset is not None and start_offset > start:
start, block_count, block_starts, block_sizes = _move_start(self.exon_intervals, block_count, block_starts,
block_sizes, start, start_offset)
if stop_offset is not None and stop_offset < stop:
stop, block_count, block_starts, block_sizes = _move_stop(self.exon_intervals, block_count, block_starts,
block_sizes, stop, start, stop_offset)
if start > thick_start:
thick_start = start
if stop < thick_stop:
thick_stop = stop
if (start > thick_stop and stop > thick_stop) or (start < thick_start and stop < thick_start):
thick_start = 0
thick_stop = 0
block_starts = ",".join(map(str, block_starts))
block_sizes = ",".join(map(str, block_sizes))
return [self.chromosome, start, stop, name, self.score, convert_strand(self.strand), thick_start, thick_stop, rgb,
block_count, block_sizes, block_starts]
def get_interval(self):
"""
Returns a ChromosomeInterval object representing the full span of this transcript.
"""
return ChromosomeInterval(self.chromosome, self.start, self.stop, convert_strand(self.strand))
def _get_exons(self, bed_tokens):
"""
Get a list of Exons representing the exons in transcript coordinate
space. This is in transcript order. See the Exon class for more.
"""
exons = []
chrom_start, chrom_stop = int(bed_tokens[1]), int(bed_tokens[2])
thick_start, thick_stop = int(bed_tokens[6]), int(bed_tokens[7])
if thick_start == thick_stop:
thick_start = thick_stop = 0
chrom, strand = bed_tokens[0], convert_strand(bed_tokens[5])
block_count = int(bed_tokens[9])
block_sizes = [int(x) for x in bed_tokens[10].split(",") if x != ""]
block_starts = [int(x) for x in bed_tokens[11].split(",") if x != ""]
##################################################################
# HERE BE DRAGONS
# this is seriously ugly code to maintain proper mapping
# between coordinate spaces. See the unit tests.
##################################################################
if strand is False:
block_sizes = reversed(block_sizes)
block_starts = reversed(block_starts)
t_pos, cds_pos = 0, None
for block_size, block_start in izip(block_sizes, block_starts):
# calculate transcript relative coordinates
this_start = t_pos
this_stop = t_pos + block_size
# calculate chromosome relative coordinates
this_chrom_start = chrom_start + block_start
this_chrom_stop = chrom_start + block_start + block_size
# calculate transcript-relative CDS positions
# cds_pos is pos of first coding base in CDS coordinates
this_cds_start, this_cds_stop, this_cds_pos = None, None, None
if strand is True:
# special case - single exon
if block_count == 1:
this_cds_pos = 0
this_cds_start = thick_start - this_chrom_start
this_cds_stop = thick_stop - this_chrom_start
# special case - entirely non-coding
elif thick_start == thick_stop == 0:
this_cds_start, this_cds_stop, this_cds_pos = None, None, None
# special case - CDS starts and stops on the same exon
elif (this_chrom_start <= thick_start < this_chrom_stop and this_chrom_start < thick_stop <=
this_chrom_stop):
this_cds_pos = 0
cds_pos = this_chrom_stop - thick_start
this_cds_start = this_start + thick_start - this_chrom_start
this_cds_stop = this_stop + thick_stop - this_chrom_stop
# is this the start codon containing exon?
elif this_chrom_start <= thick_start < this_chrom_stop:
cds_pos = this_chrom_stop - thick_start
this_cds_pos = 0
this_cds_start = this_start + thick_start - this_chrom_start
# is this the stop codon containing exon?
elif this_chrom_start < thick_stop <= this_chrom_stop:
this_cds_pos = cds_pos
cds_pos += thick_stop - this_chrom_start
this_cds_stop = this_stop + thick_stop - this_chrom_stop
# is this exon all coding?
elif (this_cds_stop is None and this_cds_start is None and thick_stop >=
this_chrom_stop and thick_start < this_chrom_start):
this_cds_pos = cds_pos
cds_pos += block_size
else:
# special case - single exon
if block_count == 1:
this_cds_pos = 0
this_cds_start = this_chrom_stop - thick_stop
this_cds_stop = thick_stop - this_chrom_start + this_cds_start
# special case - entirely non-coding
elif thick_start == thick_stop == 0:
this_cds_start, this_cds_stop, this_cds_pos = None, None, None
# special case - start and stop codons are on the same exon
elif (this_chrom_start < thick_stop <= this_chrom_stop and this_chrom_start <= thick_start <
this_chrom_stop):
cds_pos = thick_stop - this_chrom_start
this_cds_pos = 0
this_cds_start = this_start + this_chrom_stop - thick_stop
this_cds_stop = this_start + this_chrom_stop - thick_start
# is this the start codon containing exon?
elif this_chrom_start < thick_stop <= this_chrom_stop:
cds_pos = thick_stop - this_chrom_start
this_cds_pos = 0
this_cds_start = this_start + this_chrom_stop - thick_stop
# is this the stop codon containing exon?
elif this_chrom_start <= thick_start < this_chrom_stop:
this_cds_pos = cds_pos
this_cds_stop = this_start + this_chrom_stop - thick_start
# is this exon all coding?
elif (this_cds_stop is None and this_cds_start is None and thick_stop >=
this_chrom_stop and thick_start < this_chrom_start):
this_cds_pos = cds_pos
cds_pos += block_size
exons.append(Exon(this_start, this_stop, strand, this_chrom_start, this_chrom_stop, this_cds_start,
this_cds_stop, this_cds_pos))
t_pos += block_size
return exons