def run_unit_reconstructor(self, ncrf_fn):
unit_star_fn = os.path.join(self.params.outdir, 'cons_unit',
'unit_star.fasta')
unit_star_cmd = [
"python", "-u", self.unit_reconstructor, "--reads-ncrf", ncrf_fn,
"--unit", self.params.unit, "-k", self.params.cons_k_mer_len,
"--output", unit_star_fn
]
unit_star_cmd = listEls2str(unit_star_cmd)
print("Running unit star reconstruction")
print(list2str(unit_star_cmd))
subprocess.call(unit_star_cmd)
return unit_star_fn
def run_NCRF(self):
ncrf_outdir = os.path.join(self.params.outdir, "NCRF")
ncrf_cmd = [
"python", "-u", self.ncrf_script, "--reads", self.params.reads,
"--threads", self.params.threads, "--outdir", ncrf_outdir,
"--repeat", self.params.unit
]
ncrf_cmd = listEls2str(ncrf_cmd)
print('Running NCRF:')
print(list2str(ncrf_cmd))
subprocess.call(ncrf_cmd)
ncrf_fn = os.path.join(ncrf_outdir, "report.ncrf")
return ncrf_fn
def run_tandemPolisher(self, assembly_fn):
polisher_outdir = os.path.join(self.params.outdir, "polishing2")
polisher_cmd = [
"python", "-u", self.tandemPolisher, "--nano", self.params.reads,
"-o", polisher_outdir, "--only-polish", "-t", self.params.threads,
assembly_fn
]
polisher_cmd = listEls2str(polisher_cmd)
print("Running polishing with tandemMapper")
print(list2str(polisher_cmd))
subprocess.call(polisher_cmd)
polished_assembly_fn = \
os.path.join(polisher_outdir, 'polished', f'polished_{self.params.num_polish_iters}.fasta')
return polished_assembly_fn
def run_read_placer(self, ncrf_fn, kmers_fn):
placer_outdir = os.path.join(self.params.outdir, "tr_resolution")
placer_cmd = [
"python", "-u", self.placer_script, "--ncrf", ncrf_fn,
"--genomic-kmers", kmers_fn, "--outdir", placer_outdir,
"--n-motif", self.params.n_motif, "--min-cloud-kmer-freq",
self.params.min_cloud_kmer_freq, "--min-kmer-mult",
self.params.min_kmer_mult, "--min-unit", self.params.min_unit,
"--min-inters", self.params.min_inters
]
placer_cmd = listEls2str(placer_cmd)
print("Running tandem repeat resolution")
print(list2str(placer_cmd))
subprocess.call(placer_cmd)
read_pos_fn = os.path.join(placer_outdir, "read_positions.csv")
return read_pos_fn
def run_polisher(self, ncrf_fn, read_pos_fn, unit_star_fn):
polisher_outdir = os.path.join(self.params.outdir, "polishing1")
polisher_cmd = [
"python", "-u", self.polisher_script, "--read-placement",
read_pos_fn, "--outdir", polisher_outdir, "--ncrf", ncrf_fn,
"--error-mode", self.params.error_mode, "--num-iters",
self.params.num_polish_iters, "--num-threads", self.params.threads,
"--unit", unit_star_fn, "--min-pos", self.params.min_pos
]
if self.params.max_pos != math.inf:
polisher_cmd.append("--max-pos")
polisher_cmd.append(self.params.max_pos)
polisher_cmd = listEls2str(polisher_cmd)
print("Running polishing -- first stage")
print(list2str(polisher_cmd))
subprocess.call(polisher_cmd)
assembly_fn = \
os.path.join(
polisher_outdir,
f'final_sequence_{self.params.num_polish_iters}.fasta'
)
return assembly_fn
def run_kmer_recr(self, ncrf_fn):
recr_outdir = os.path.join(self.params.outdir,
"recruited_unique_kmers")
recr_cmd = [
"python", "-u", self.recr_script, "--ncrf", ncrf_fn, "--coverage",
self.params.coverage, "--min-coverage", self.params.min_coverage,
"--outdir", recr_outdir, "--min-nreads", self.params.min_nreads,
"--max-nreads", self.params.max_nreads, "--min-distance",
self.params.min_distance, "--max-distance",
self.params.max_distance, "--bottom", self.params.bottom, "--top",
self.params.top, "--kmer-survival-rate",
self.params.kmer_survival_rate, "--max-nonuniq",
self.params.max_nonuniq
]
recr_cmd = listEls2str(recr_cmd)
print("Running unique kmer recruitment")
print(list2str(recr_cmd))
subprocess.call(recr_cmd)
kmers_fn = os.path.join(
recr_outdir,
f"unique_kmers_min_edge_cov_{self.params.min_coverage}.txt")
return kmers_fn