def run(self):
reads_order = os.path.join(self.work_dir, "reads_order.fasta")
asm.assemble(self.args, reads_order, self.log_file)
contigs_fasta = aln.concatenate_contigs(reads_order)
fp.write_fasta_dict(contigs_fasta, self.out_assembly)
Job.run_description["stage_name"] = self.name
def run(self):
logger.info("Computing rough consensus")
contigs_info = aln.get_contigs_info(self.in_contigs)
consensus_fasta = cons.get_consensus(self.in_alignment, self.in_contigs,
contigs_info, self.args.threads,
self.args.platform,
self.args.min_overlap)
fp.write_fasta_dict(consensus_fasta, self.out_consensus)
def run(self):
logger.info("Computing rough consensus")
contigs_info = aln.get_contigs_info(self.in_contigs)
consensus_fasta = cons.get_consensus(self.in_alignment, contigs_info,
self.args.threads)
fp.write_fasta_dict(consensus_fasta, self.out_consensus)
Job.run_description["stage_name"] = self.name
def make_blasr_reference(contigs_fasta, out_file):
"""
Outputs 'reference' for BLASR run, appends a suffix to circular contigs
"""
circular_window = config.vals["circular_window"]
for contig_id in contigs_fasta:
contig_type = contig_id.split("_")[0]
if (contig_type == "circular"
and len(contigs_fasta[contig_id]) > circular_window):
contigs_fasta[contig_id] += \
contigs_fasta[contig_id][:circular_window]
fp.write_fasta_dict(contigs_fasta, out_file)
def run(self):
logger.info("Polishing genome ({0}/{1})".format(self.stage_id,
self.args.num_iters))
contigs_info = aln.get_contigs_info(self.in_contigs)
logger.info("Separating alignment into bubbles")
bubbles = bbl.get_bubbles(self.in_alignment, contigs_info, self.in_contigs,
self.seq_platform, self.args.threads,
self.args.min_overlap)
logger.info("Correcting bubbles")
polished_fasta = pol.polish(bubbles, self.args.threads,
self.seq_platform, self.work_dir,
self.stage_id)
fp.write_fasta_dict(polished_fasta, self.out_consensus)
def run(self):
alignment, mean_error = aln.parse_alignment(self.in_alignment)
#if Job.run_description["error_profile"] == "":
# Job.run_description["error_profile"] = \
# aln.choose_error_mode(mean_error)
bubbles = bbl.get_bubbles(alignment, self.seq_platform)
polished_fasta = pol.polish(bubbles, self.args.threads,
self.seq_platform, self.work_dir,
self.stage_id)
fp.write_fasta_dict(polished_fasta, self.out_consensus)
Job.run_description["stage_name"] = self.name
Job.run_description["stage_id"] = self.stage_id
def patch_genome(alignment, reference_file, out_patched):
aln_by_ctg = defaultdict(list)
for aln in alignment:
aln_by_ctg[aln.trg_id].append(aln)
ref_fasta = fp.read_fasta_dict(reference_file)
fixed_fasta = {}
for ctg_id, ctg_aln in aln_by_ctg.iteritems():
patches = _get_patching_alignmemnts(ctg_aln)
fixed_sequence = _apply_patches(patches, ref_fasta[ctg_id])
fixed_fasta[ctg_id] = fixed_sequence
fp.write_fasta_dict(fixed_fasta, out_patched)