def test_load_like(tmp_path):
gene_file = str(tmp_path / 'genes.txt')
# make a permutation
perm = np.random.choice(NGENES, NGENES - 10, replace=False)
# load data to make reference and permute
umis, genes = prep.load_txt(TXT)
umis = umis.A[:, perm]
genes = genes.loc[perm]
# write permuted/subsampled reference file
genes.to_csv(gene_file, header=None, sep='\t', index=None)
# load like permuted reference
ll_umi, ll_genes = prep.load_like(TXT, reference=gene_file)
assert_equal(len(ll_genes), len(perm))
assert_array_equal(umis, ll_umi.A)
# repeat with no_split_on_dot
ll_umi, ll_genes = prep.load_like(TXT,
reference=gene_file,
no_split_on_dot=True)
assert_equal(len(ll_genes), len(perm))
assert_array_equal(umis, ll_umi.A)
# by gene name
ll_umi, ll_genes = prep.load_like(TXT,
reference=gene_file,
by_gene_name=True)
assert_equal(len(ll_genes), len(perm))
assert_array_equal(umis, ll_umi.A)
# corrupt the permuted reference
bad_genes = genes.copy()
bad_genes.loc[5, 0] = 'random'
bad_genes.to_csv(gene_file, header=None, sep='\t', index=None)
with pytest.raises(ValueError):
ll_umi, ll_genes = prep.load_like(TXT, reference=gene_file)
def test_load_txt(ngene_cols):
coo, genes = prep.load_txt(TXT, ngene_cols)
assert genes.shape[1] == ngene_cols
assert coo.shape[1] == NGENES
assert genes.shape[0] == NGENES
assert coo.shape[0] == NCELLS + 2 - ngene_cols