def create_countgraph(args, ksize=None, multiplier=1.0, fp_rate=0.1):
"""Create and return a countgraph."""
args = _check_fp_rate(args, fp_rate)
if hasattr(args, 'force'):
if args.n_tables > 20:
if not args.force:
print_error("\n** ERROR: khmer only supports number "
"of tables <= 20.\n")
sys.exit(1)
else:
if args.n_tables > 20:
log_warn("\n*** Warning: Maximum recommended number of "
"tables is 20, discarded by force nonetheless!\n")
if ksize is None:
ksize = args.ksize
if ksize > 32:
print_error("\n** ERROR: khmer only supports k-mer sizes <= 32.\n")
sys.exit(1)
if args.small_count:
tablesize = calculate_graphsize(args,
'smallcountgraph',
multiplier=multiplier)
return khmer.SmallCountgraph(ksize, tablesize, args.n_tables)
else:
tablesize = calculate_graphsize(args,
'countgraph',
multiplier=multiplier)
cg = khmer.Countgraph(ksize, tablesize, args.n_tables)
if hasattr(args, 'bigcount'):
cg.set_use_bigcount(args.bigcount)
return cg
def test_smallcountgraph_vs_table():
x = khmer.SmallCounttable(4, 21, 3)
y = khmer.SmallCountgraph(4, 21, 3)
assert hasattr(x, 'add')
assert hasattr(y, 'add')
assert not hasattr(x, 'consume_and_tag')
assert hasattr(y, 'consume_and_tag')
def test_get_raw_tables_smallcountgraph():
# for the same number of entries a SmallCountgraph uses ~half the memory
# of a normal Countgraph
ht = khmer.SmallCountgraph(20, 1e5, 4)
tables = ht.get_raw_tables()
for size, table in zip(ht.hashsizes(), tables):
assert isinstance(table, memoryview)
assert size // 2 + 1 == len(table)
def test_get_raw_tables_view_smallcountgraph():
ht = khmer.SmallCountgraph(4, 1e5, 4)
tables = ht.get_raw_tables()
for tab in tables:
assert sum(tab.tolist()) == 0
ht.consume('AAAA')
# the actual count is 1 but stored in the first 4bits of a Byte
# and so becomes 16
for tab in tables:
assert sum(tab.tolist()) == int('00010000', 2)
def test_save_load_occupied_small(ctfile):
print('working with', ctfile)
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename(ctfile)
orig = khmer.SmallCountgraph(12, 1e5, 4)
orig.consume_seqfile(inpath)
orig.save(savepath)
loaded = khmer.load_countgraph(savepath, small=True)
orig_count = orig.n_occupied()
loaded_count = loaded.n_occupied()
assert orig_count == 3886, orig_count
assert loaded_count == orig_count, loaded_count