def _fix_stop_codon(transcript):
"""This private function will fix the CDS and stop codons when the transcript comes from GTF2
and therefore has, incorrectly, the stop codon outside the CDS."""
if transcript.strand == "-":
# We need to check whether the stop codon is actually in the same exon.
if transcript.stop_codon[-1][1] == transcript.combined_cds[0][0] - 1:
phase = transcript.phases.pop(transcript.combined_cds[0], None)
transcript.combined_cds[0] = (transcript.stop_codon.pop(-1)[0],
transcript.combined_cds[0][1])
transcript.phases[transcript.combined_cds[0]] = phase
transcript.combined_cds = [tuple(_) for _ in transcript.stop_codon] + transcript.combined_cds
for pos, utr in enumerate(transcript.combined_utr):
if utr[0] > transcript.combined_cds[-1][1]:
continue # Skip the 5'
over = overlap(utr, transcript.combined_cds[0])
if over < 0:
continue
elif over > 3:
raise InvalidCDS("Invalid overlap between UTR and CDS found")
else:
if over == utr[1] - utr[0] + 1: # This is equivalent to a fragment. Remove.
transcript.combined_utr[pos] = None
else:
transcript.combined_utr[pos] = (utr[0], max(utr[0], transcript.combined_cds[0][0] - 1))
else:
# Expand the last CDS
if transcript.stop_codon[0][0] == transcript.combined_cds[-1][1] + 1:
phase = transcript.phases.pop(transcript.combined_cds[-1], None)
transcript.combined_cds[-1] = (transcript.combined_cds[-1][0],
transcript.stop_codon.pop(0)[1])
transcript.phases[transcript.combined_cds[-1]] = phase
transcript.combined_cds.extend([tuple(_) for _ in transcript.stop_codon])
for pos, utr in enumerate(transcript.combined_utr):
if utr[1] < transcript.combined_cds[0][0]:
continue # Skip the 5'
over = overlap(utr, transcript.combined_cds[-1])
if over < 0:
continue
elif over > 3:
raise InvalidCDS("Invalid overlap between UTR and CDS found")
else:
if over == utr[1] - utr[0] + 1: # This is equivalent to a fragment. Remove.
transcript.combined_utr[pos] = None
else:
transcript.combined_utr[pos] = (min(utr[1], transcript.combined_cds[-1][1] + 1),
utr[1])
transcript.combined_utr = [_ for _ in transcript.combined_utr if _ is not None] # Remove the deleted UTRs
return transcript
def _analyse_cDNAs(self, cdnas, beds, peps):
result, cigar = transfer.get_and_prepare_cigar(*cdnas)
t1bed, t2bed = beds
assert sorted(t1bed.blocks)[-1][1] == len(cdnas[0]), (t1bed.blocks,
len(cdnas[0]))
assert sorted(t2bed.blocks)[-1][1] == len(cdnas[1]), (t2bed.blocks,
len(cdnas[1]))
try:
c_t1_exons, c_t2_exons, common = self.transfer_exon_coordinates(
cigar, t1bed.blocks, t2bed.blocks)
except (ValueError, AssertionError) as exc:
raise ValueError(exc)
# Common: list(zip(query_array, target_array))
identical = sum(length for length, op in cigar if op in ("=", ))
if identical == 0:
identity = 0
else:
identity = round(100 * identical / len(common), 2)
result = array_compare(np.ravel(np.array(c_t1_exons)),
np.ravel(np.array(c_t2_exons)), identity)
result, ccode = result[:-1].reshape((2, 3)), int(result[-1])
# Now that we have analysed the cDNAs, it is time for the CDS
if identity > 0 and t1bed.coding and t2bed.coding and all(peps):
t1_coding_exons = [
(max(t1bed.thick_start - 1, _[0]), min(t1bed.thick_end, _[1]))
for _ in t1bed.blocks
if overlap(_, (t1bed.thick_start - 1, t1bed.thick_end)) > 0
]
assert t1_coding_exons, (t1bed.blocks, t1bed.block_starts,
t1bed.block_sizes, t1bed.thick_start,
t1bed.thick_end)
t2_coding_exons = [
(max(t2bed.thick_start - 1, _[0]), min(t2bed.thick_end, _[1]))
for _ in t2bed.blocks
if overlap(_, (t2bed.thick_start - 1, t2bed.thick_end)) > 0
]
assert t2_coding_exons
query_array, target_array = list(zip(*common))
c_t1_coding = transfer.transfer_exons(t1_coding_exons, query_array)
c_t2_coding = transfer.transfer_exons(t2_coding_exons,
target_array)
t1pep, t2pep = peps
self.log.debug("CDS: %s:%s-%s: %s", t2bed.chrom,
t2bed.thick_start - 2, t2bed.thick_end, t2pep)
# print(t2bed.chrom, t2bed.thick_start-2, t2bed.thick_end, t2_coding_exons, t2pep)
coding_result, coding_cigar = transfer.get_and_prepare_cigar(
t1pep, t2pep)
coding_common = transfer.cigar_length_in_common(coding_cigar)
coding_identical = sum(length for length, op in coding_cigar
if op in ("=", "M"))
if coding_identical == 0:
self.log.warning(
"No protein overlap at all for %s and %s.\nProtein 1: %s\nProtein 2: %s",
t1bed.name, t2bed.name, t1pep, t2pep)
c_t1_coding = "NA"
c_t2_coding = "NA"
coding_identity = 0
coding_result = np.zeros((2, 3))
coding_ccode = 0
# raise ValueError((t1bed.name, coding_cigar, t1pep, t2pep))
else:
coding_identity = round(100 * coding_identical / coding_common,
2)
coding_result = array_compare(
np.ravel(np.array(c_t1_coding, dtype=np.int)),
np.ravel(np.array(c_t2_coding, dtype=np.int)),
coding_identity)
coding_result, coding_ccode = coding_result[:-1].reshape(
(2, 3)), int(coding_result[-1])
else:
c_t1_coding = "NA"
c_t2_coding = "NA"
coding_identity = 0
coding_result = np.zeros((2, 3))
coding_ccode = 0
return (c_t1_exons, c_t2_exons, identity, result, ccode, c_t1_coding,
c_t2_coding, coding_identity, coding_result, coding_ccode)