def test_bam_file_to_region_fasta(self):
'''Test bam_file_to_region_fasta'''
tmp = 'tmp.to_fasta.fa'
bam = os.path.join(data_dir, 'mapping_test.smalt.out.sorted.bam')
mapping.bam_file_to_region_fasta(bam, tmp, 'ref', start=42, end=142)
self.assertTrue(filecmp.cmp(os.path.join(data_dir, 'mapping_test.bam_to_region.fasta'), tmp))
os.unlink(tmp)
def _make_new_seed(self, seed_name):
if self.verbose:
print('Making new seed for', seed_name, ' ... start')
tmp_prefix = os.path.join(self.tmpdir, 'out')
seed_reads = tmp_prefix + '.' + seed_name + '.reads_1.fa'
if len(self.original_seeds[seed_name]) > self.seed_stop_length:
start = int(0.5 * len(self.original_seeds[seed_name]) -
0.5 * self.seed_stop_length)
end = int(0.5 * len(self.original_seeds[seed_name]) +
0.5 * self.seed_stop_length)
else:
start = None
end = None
if self.verbose:
print('Making new seed for', seed_name, ' ... getting reads')
mapping.bam_file_to_region_fasta(self.bam_file, seed_reads, seed_name,
start, end)
if self.verbose:
print('Making new seed for', seed_name,
' ... finding most common kmer')
new_seed = seed.Seed(extend_length=self.extend_length,
overlap_length=self.overlap_length,
reads1=seed_reads,
ext_min_cov=self.ext_min_cov,
ext_min_ratio=self.ext_min_ratio,
verbose=self.verbose,
seed_length=self.seed_length,
seed_min_count=self.seed_min_count,
seed_max_count=self.seed_max_count,
kmc_threads=self.kmc_threads,
map_threads=self.threads)
if len(new_seed) == 0:
print('Warning: could not get most common kmer for', seed_name)
return
if self.verbose:
print('Making new seed for', seed_name,
' ... extending most common kmer')
new_seed.extend(self.reads1, self.reads2, self.seed_stop_length)
f = pyfastaq.utils.open_file_write(tmp_prefix + '.' + seed_name +
'.fa')
print(pyfastaq.sequences.Fasta('seed.' + seed_name,
new_seed.seq[10:-10]),
file=f)
pyfastaq.utils.close(f)
if self.verbose:
print('Making new seed for', seed_name, ' ... finished')
def _make_new_seed(self, seed_name):
if self.verbose:
print('Making new seed for', seed_name, ' ... start')
tmp_prefix = os.path.join(self.tmpdir, 'out')
seed_reads = tmp_prefix + '.' + seed_name + '.reads_1.fa'
if len(self.original_seeds[seed_name]) > self.seed_stop_length:
start = int(0.5 * len(self.original_seeds[seed_name]) - 0.5 * self.seed_stop_length)
end = int(0.5 * len(self.original_seeds[seed_name]) + 0.5 * self.seed_stop_length)
else:
start = None
end = None
if self.verbose:
print('Making new seed for', seed_name, ' ... getting reads')
mapping.bam_file_to_region_fasta(self.bam_file, seed_reads, seed_name, start, end)
if self.verbose:
print('Making new seed for', seed_name, ' ... finding most common kmer')
new_seed = seed.Seed(
extend_length = self.extend_length,
overlap_length = self.overlap_length,
reads1 = seed_reads,
ext_min_cov = self.ext_min_cov,
ext_min_ratio = self.ext_min_ratio,
verbose = self.verbose,
seed_length = self.seed_length,
seed_min_count = self.seed_min_count,
seed_max_count = self.seed_max_count,
kmc_threads = self.kmc_threads,
map_threads = self.threads
)
if len(new_seed) == 0:
print('Warning: could not get most common kmer for', seed_name)
return
if self.verbose:
print('Making new seed for', seed_name, ' ... extending most common kmer')
new_seed.extend(self.reads1, self.reads2, self.seed_stop_length)
f = pyfastaq.utils.open_file_write(tmp_prefix + '.' + seed_name + '.fa')
print(pyfastaq.sequences.Fasta('seed.' + seed_name, new_seed.seq[10:-10]), file=f)
pyfastaq.utils.close(f)
if self.verbose:
print('Making new seed for', seed_name, ' ... finished')
