# Read in alignments of intrachromosomal chimeras and proceed with ordering and orientation
log('-- Reading intrachromosomal chimera broken alignments')
alns = read_paf_alignments('chimera_break/intra_contigs_against_ref.paf')
alns = clean_alignments(alns, l=1000, in_exclude_file=exclude_file)
contigs_file = '/ragoo_output/chimera_break/' + out_intra_fasta
log('-- The total number of interchromasomally chimeric contigs broken is %r' % total_inter_broken)
log('-- The total number of intrachromasomally chimeric contigs broken is %r' % total_intra_broken)
# Assign each contig to a corresponding reference chromosome.
log('-- Assigning contigs')
all_unique_contigs = dict()
for i in alns.keys():
all_unique_contigs[i] = UniqueContigAlignment(alns[i])
# Add to this the list of headers that did not make it
write_contig_clusters(all_unique_contigs, group_score_thresh, skip_ctg)
log('-- Ordering and orienting contigs')
order_orient_contigs(all_unique_contigs, alns)
log('-- Creating Pseudomolecules')
create_pseudomolecules(contigs_file, all_unique_contigs, g)
if call_svs:
log('-- Aligning pseudomolecules to reference')
align_pms(minimap_path, t, reference_file)
log('-- Getting structural variants')
def order_orient_contigs(in_unique_contigs, in_alns):
current_path = os.getcwd()
output_path = current_path + '/orderings'
if not os.path.exists(output_path):
os.makedirs(output_path)
# Get longest alignments
longest_contigs = dict()
for i in in_alns.keys():
# Only consider alignments to the assigned chromosome
uniq_aln = UniqueContigAlignment(in_alns[i])
best_header = uniq_aln.ref_chrom
ctg_alns = copy.deepcopy(in_alns[i])
ctg_alns.filter_ref_chroms([best_header])
longest_contigs[i] = LongestContigAlignment(ctg_alns)
# Save the orientations
final_orientations = dict()
for i in longest_contigs.keys():
final_orientations[i] = longest_contigs[i].strand
# Get the location and orientation confidence scores
orientation_confidence = dict()
location_confidence = dict()
forward_bp = 0
reverse_bp = 0
for i in in_alns.keys():
uniq_aln = UniqueContigAlignment(in_alns[i])
best_header = uniq_aln.ref_chrom
ctg_alns = copy.deepcopy(in_alns[i])
ctg_alns.filter_ref_chroms([best_header])
# Orientation confidence scores
# Every base pair votes for the orientation of the alignment in which it belongs
# Score is # votes for the assigned orientation over all votes
for j in range(len(ctg_alns.ref_headers)):
if ctg_alns.strands[j] == '+':
forward_bp += ctg_alns.aln_lens[j]
else:
reverse_bp += ctg_alns.aln_lens[j]
if final_orientations[i] == '+':
orientation_confidence[i] = forward_bp / (forward_bp + reverse_bp)
else:
orientation_confidence[i] = reverse_bp / (forward_bp + reverse_bp)
forward_bp = 0
reverse_bp = 0
# Location confidence
location_confidence[i] = get_location_confidence(ctg_alns)
all_chroms = set([in_unique_contigs[i].ref_chrom for i in in_unique_contigs.keys()])
for this_chrom in all_chroms:
# Intialize the list of start and end positions w.r.t the query
ref_pos = []
groupings_file = 'groupings/' + this_chrom + '_contigs.txt'
contigs_list = get_contigs_from_groupings(groupings_file)
for i in range(len(contigs_list)):
# There is a scope issue here. Pass this (longest_contigs) to the method explicitly.
ref_pos.append((longest_contigs[contigs_list[i]].ref_start, longest_contigs[contigs_list[i]].ref_end, i))
final_order = [contigs_list[i[2]] for i in sorted(ref_pos)]
# Get ordering confidence
# To do this, get the max and min alignments to this reference chromosome
# Then within that region, what percent of bp are covered
with open('orderings/' + this_chrom + '_orderings.txt', 'w') as out_file:
for i in final_order:
# Also have a scope issue here.
out_file.write(i + '\t' + final_orientations[i] + '\t' + str(location_confidence[i]) + '\t' + str(orientation_confidence[i]) + '\n')
def get_intra_contigs(alns, l, d, c):
"""
Flag contigs as being intrachromosomal chimeras
:param alns:
:param l: Minimum alignment length to consider
:param d: Distance between consecutive adjacent alignments with respect to the reference. If larger than this, flag
:param c: Distance between consecutive adjacent alignments with respect to the query. If larger than this, flag
:return: dict of contigs and break points.
"""
# Get only the header to which this contig mostly aligns to and filter out smaller alignments.
uniq_aln = UniqueContigAlignment(alns)
best_header = uniq_aln.ref_chrom
ctg_alns = copy.deepcopy(alns)
ctg_alns.filter_ref_chroms([best_header])
ctg_alns.filter_lengths(l)
# If there are no longer any alignments after length filtering, give up
if not len(ctg_alns.ref_headers):
return
# Sort the alignments with respect to the reference start and end positions.
ctg_alns.sort_by_ref()
# Make a list of distance between alignments
# first with respect to (wrt) the reference.
distances_wrt_ref = []
for i in range(len(ctg_alns.ref_headers) - 1):
distances_wrt_ref.append(ctg_alns.ref_starts[i + 1] -
ctg_alns.ref_starts[i])
# next, with respect to (wrt) the contig.
distances_wrt_ctg = []
for i in range(len(ctg_alns.ref_headers) - 1):
distances_wrt_ctg.append(
abs(ctg_alns.query_starts[i + 1] - ctg_alns.query_starts[i]))
# Next, assign the following two identities.
# 1. When ordered by the reference, the alignments start at the beginning or the end of the query
# 2. For the alignment which will be broken on, is it on the forward or reverse strand.
is_query_start = True
if ctg_alns.query_starts[0] >= ctg_alns.query_lens[0] / 5:
is_query_start = False
# This conditional essentially checks if there are any break points for this contig.
# Returns None otherwise (no return statement)
if distances_wrt_ref:
if max(distances_wrt_ref) > d:
gap_index = distances_wrt_ref.index(max(distances_wrt_ref))
a_alns_strands = ctg_alns.strands[:gap_index]
if is_query_start:
if a_alns_strands.count('-') > a_alns_strands.count('+'):
# The first subcontig is on the reverse strand
return (ctg_alns.contig, [(0, ctg_alns.query_ends[0]),
(ctg_alns.query_ends[0],
ctg_alns.query_lens[0])])
else:
# The first subcontig is on the forward strand.
return (ctg_alns.contig, [(0,
ctg_alns.query_ends[gap_index]),
(ctg_alns.query_ends[gap_index],
ctg_alns.query_lens[0])])
else:
# The first subcontig starts at the end of the contig
if a_alns_strands.count('-') > a_alns_strands.count('+'):
# The first subcontig is on the reverse strand
return (ctg_alns.contig,
[(0, ctg_alns.query_starts[gap_index]),
(ctg_alns.query_starts[gap_index],
ctg_alns.query_lens[0])])
else:
# The first subcontig is on the forward strand.
return (ctg_alns.contig, [(0, ctg_alns.query_starts[0]),
(ctg_alns.query_starts[0],
ctg_alns.query_lens[0])])
if max(distances_wrt_ctg) > c:
gap_index = distances_wrt_ctg.index(max(distances_wrt_ctg)) + 1
a_alns_strands = ctg_alns.strands[:gap_index]
if is_query_start:
if a_alns_strands.count('-') > a_alns_strands.count('+'):
# The first subcontig is on the reverse strand
return (ctg_alns.contig, [(0, ctg_alns.query_ends[0]),
(ctg_alns.query_ends[0],
ctg_alns.query_lens[0])])
else:
# The first subcontig is on the forward strand.
return (ctg_alns.contig, [(0,
ctg_alns.query_ends[gap_index]),
(ctg_alns.query_ends[gap_index],
ctg_alns.query_lens[0])])
else:
# The first subcontig starts at the end of the contig
if a_alns_strands.count('-') > a_alns_strands.count('+'):
# The first subcontig is on the reverse strand
return (ctg_alns.contig,
[(0, ctg_alns.query_starts[gap_index - 1]),
(ctg_alns.query_starts[gap_index - 1],
ctg_alns.query_lens[0])])
else:
# The first subcontig is on the forward strand.
return (ctg_alns.contig, [(0, ctg_alns.query_starts[0]),
(ctg_alns.query_starts[0],
ctg_alns.query_lens[0])])