def test_build_chunks():
N = 3
bloomfilter_filepaths = ['bigsi/tests/data/test_kmers.bloom'] * N
sample_names = generate_sample_names(len(bloomfilter_filepaths))
bigsi1 = BIGSI.create(db="./db-bigsi-no-max-mem/",
m=10,
k=9,
h=1,
force=True)
build(bloomfilter_filepaths, sample_names, bigsi1)
bigsi2 = BIGSI.create(db="./db-bigsi-max-mem/", m=10, k=9, h=1, force=True)
build(bloomfilter_filepaths, sample_names, bigsi2,
max_memory=20) # 20bytes
# Reload and test equal
bigsi1 = BIGSI("./db-bigsi-no-max-mem/")
bigsi2 = BIGSI("./db-bigsi-max-mem")
for i in range(10):
assert bigsi1.graph[i] == bigsi2.graph[i]
for k, v in bigsi2.metadata.items():
assert bigsi1.metadata[k] == v
bigsi1.delete_all()
bigsi2.delete_all()
def test_merge():
kmers1 = ['AAAAAAAAA'] * 3
kmers2 = ['AAAAAAAAT'] * 9
bigsi1 = BIGSI.create(db="./db-bigsi1/", m=10, k=9, h=1, force=True)
blooms1 = []
for s in kmers1:
blooms1.append(bigsi1.bloom([s]))
samples1 = [str(i) for i in range(len(kmers1))]
bigsi1.build(blooms1, samples1)
bigsi2 = BIGSI.create(db="./db-bigsi2/", m=10, k=9, h=1, force=True)
blooms2 = []
for s in kmers2:
blooms2.append(bigsi2.bloom([s]))
samples2 = [str(i) for i in range(len(kmers2))]
bigsi2.build(blooms2, samples2)
combined_samples = combine_samples(samples1, samples2)
bigsicombined = BIGSI.create(db="./db-bigsi-c/",
m=10,
k=9,
h=1,
force=True)
bigsicombined = BIGSI(db="./db-bigsi-c/", mode="c")
bigsicombined.build(blooms1 + blooms2, combined_samples)
bigsi1.merge(bigsi2)
bigsi1 = BIGSI(db="./db-bigsi1/")
for i in range(10):
assert bigsi1.graph[i] == bigsicombined.graph[i]
for k, v in bigsicombined.metadata.items():
assert bigsi1.metadata[k] == v
bigsi1.delete_all()
bigsi2.delete_all()
bigsicombined.delete_all()