def build_grid_cover_fasta(radius, dimension, threads, outfile, predb_list):
environ[
'OPENBLAS_NUM_THREADS'] = f'{threads}' # numpy uses one of these two libraries
environ['MKL_NUM_THREADS'] = f'{threads}'
predb_list = [line.strip() for line in predb_list]
logger = lambda n: stderr.write(f'\rAdded {n:,} k-mers to db')
start = time()
predb = PreDB.load_from_filepath(predb_list[0])
grid = GridCoverBuilder.build_from_predb(outfile,
predb,
radius,
logger=logger)
with click.progressbar(predb_list) as predbs:
for i, predb_filename in enumerate(predbs):
if i == 0:
continue
predb = PreDB.load_from_filepath(predb_filename)
n_added = grid.add_kmers_from_predb(predb, logger=logger)
grid.db._build_indices() # indices are disabled by `GCB.build_from_predb`
n_centers = grid.db.centroids().shape[0]
grid.close()
add_time = time() - start
click.echo((f'Added {n_added:,} kmers to {outfile} in {add_time:.5}s. '
f'{n_centers:,} clusters.'),
err=True)
def test_build_grid_cover_from_pre(self):
predb = PreDB.load_from_filepath(PRE_DB)
grid = GridCoverBuilder.build_from_predb(':memory:', predb, 0.5)
grid.db._build_indices()
grid.commit()
n_centers = grid.db.centroids().shape[0]
n_points = len(grid.db.get_kmers())
self.assertGreater(n_centers, 0)
self.assertLess(n_centers, 98)
self.assertEqual(n_points, 98)
def build_grid_cover(radius, dimension, threads, num_kmers, start_offset,
outfile, preload, rotation, kmer_table):
environ[
'OPENBLAS_NUM_THREADS'] = f'{threads}' # numpy uses one of these two libraries
environ['MKL_NUM_THREADS'] = f'{threads}'
ramifier = RotatingRamifier.from_file(dimension, rotation)
grid = GridCoverBuilder.from_filepath(outfile, ramifier, radius)
start = time()
n_added = grid.fast_add_kmers_from_file(kmer_table, num_to_add=num_kmers)
grid.commit()
n_centers = grid.db.centroids().shape[0]
grid.close()
add_time = time() - start
click.echo(
f'Added {n_added:,} kmers to {outfile} in {add_time:.5}s. {n_centers:,} clusters.',
err=True)
def build_grid_cover_fasta(radius, dimension, threads, outfile, rotation,
fasta_list):
environ[
'OPENBLAS_NUM_THREADS'] = f'{threads}' # numpy uses one of these two libraries
environ['MKL_NUM_THREADS'] = f'{threads}'
fasta_list = [line.strip() for line in fasta_list]
ramifier = RotatingRamifier.from_file(dimension, rotation)
grid = GridCoverBuilder.from_filepath(outfile, ramifier, radius)
start = time()
with click.progressbar(fasta_list) as fastas:
for fasta_filename in fastas:
n_added = grid.fast_add_kmers_from_fasta(fasta_filename)
n_centers = grid.db.centroids().shape[0]
grid.close()
add_time = time() - start
click.echo((f'Added {n_added:,} kmers to {outfile} in {add_time:.5}s. '
f'{n_centers:,} clusters.'),
err=True)
def test_build_grid_cover_from_fasta(self):
ramifier = RotatingRamifier.from_file(4, KMER_ROTATION)
db = GridCoverDB(sqlite3.connect(':memory:'),
ramifier=ramifier,
box_side_len=0.5)
grid = GridCoverBuilder(db)
grid.fast_add_kmers_from_fasta(KMER_FASTA)
grid.commit()
n_centers = grid.db.centroids().shape[0]
n_points = len(grid.db.get_kmers())
self.assertGreater(n_centers, 0)
self.assertLess(n_centers, 98)
self.assertEqual(n_points, 98)