Skip to content

fix: avoid dtype overflow in top_segment_proportions for sparse small ints #4461

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
pujitha24 wants to merge 1 commit into
base: main
Choose a base branch
from
+22 −2
Open

fix: avoid dtype overflow in top_segment_proportions for sparse small ints #4461

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions docs/release-notes/4440.fix.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
Fix {func}`~scanpy.pp.calculate_qc_metrics` for sparse matrices of small integer dtypes, which gave negative `pct_counts_in_top_N_genes` (e.g. `int16`) or raised (e.g. `uint16`) {smaller}`P Paladugu`
4 changes: 2 additions & 2 deletions src/scanpy/preprocessing/_qc.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -434,7 +434,7 @@ def top_segment_proportions_sparse_csr(data, indptr, ns):
ns = ns.astype(np.int64)
ns = np.sort(ns)
maxidx = ns[-1]
sums = np.zeros((indptr.size - 1), dtype=data.dtype)
sums = np.zeros((indptr.size - 1), dtype=np.float64)
values = np.zeros((indptr.size - 1, len(ns)), dtype=np.float64)
# Just to keep it simple, as a dense matrix
partitioned = np.zeros((indptr.size - 1, maxidx), dtype=data.dtype)
Expand All @@ -447,7 +447,7 @@ def top_segment_proportions_sparse_csr(data, indptr, ns):
partitioned[i, :] = -(np.partition(-data[start:end], maxidx))[:maxidx]
partitioned[i, :] = np.partition(partitioned[i, :], maxidx - ns)
partitioned = partitioned[:, ::-1][:, : ns[-1]]
acc = np.zeros((indptr.size - 1), dtype=data.dtype)
acc = np.zeros((indptr.size - 1), dtype=np.float64)
prev = 0
# can’t use enumerate due to https://github.com/numba/numba/issues/2625
for j in range(ns.size):
Expand Down
19 changes: 19 additions & 0 deletions tests/test_qc_metrics.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -115,6 +115,25 @@ def test_top_proportions(request: pytest.FixtureRequest, array_type):
assert (segfull == propfull).all()


@pytest.mark.parametrize(
"dtype",
[np.int8, np.int16, np.uint8, np.uint16, np.uint32],
ids=lambda d: d.__name__,
)
def test_top_segments_sparse_small_int(dtype):
"""Row sums must not overflow or fail for small integer dtypes."""
dense = (
np.random
.default_rng(0)
.integers(1, min(np.iinfo(dtype).max, 1000), (3, 60))
.astype(dtype)
)
ns = np.array([5, 10])
expected = top_segment_proportions(dense, ns)
result = top_segment_proportions(sparse.csr_matrix(dense), ns) # noqa: TID251
np.testing.assert_allclose(result, expected)


# While many of these are trivial,
# they’re also just making sure the metrics are there
def test_qc_metrics(adata_prepared: AnnData):
Expand Down
Loading