Validate csr matrix in X and raw.X

pyproject.toml

@@ -20,7 +20,7 @@ classifiers = [
 license = "BSD-3-Clause"
 license-files = ["LICEN[CS]E*"]
 dependencies = [
-    "cap-anndata>=0.5.0",
+    "cap-anndata>=0.5.2",
     "numpy>=2.0.2",
     "pandas>=2.2.3",
     "scipy>=1.13.1",

src/cap_upload_validator/errors.py

@@ -113,3 +113,18 @@ class AnnDataNonStandardVarError(CapException):
             If there are other species you wish to upload to CAP, please contact
             support@celltype.info and we will work to accommodate your request.
         """ 
+
+class CSCMatrixInX(CapException):
+    name = "CSCMatrixInX"
+
+    def __init__(self, locations: list[str]):
+        """
+        locations: list of matrix locations, e.g. ['X'], ['raw.X'], or ['X', 'raw.X']
+        """
+        super().__init__()
+        self.locations = locations
+        loc_str = " and ".join(locations)
+        self.message = (
+            f"The CSC matrix is found in {loc_str}. "
+            "Gene expression matrix must be stored in CSR or dense format!"
+        )

src/cap_upload_validator/upload_validator.py

@@ -3,7 +3,7 @@
 from scipy.sparse import issparse
 from cap_anndata import CapAnnData, read_h5ad
 import logging
-from h5py import Dataset
+from h5py import Dataset, Group
 
 from .gene_mapping import (
     GeneMap,
@@ -22,6 +22,7 @@
     AnnDataNonStandardVarError,
     BadAnnDataFile,
     AnnDataNoneInGeneralMetadata,
+    CSCMatrixInX,
 )
 from typing import Optional
 
@@ -68,6 +69,7 @@ def validate(self, report_success: bool = True) -> None:
             if cap_adata.raw is not None:
                 cap_adata.raw.read_var(columns=[])
 
+            self._validate_x_and_raw_x_formats(cap_adata)
             self._check_X(cap_adata)
             self._check_obsm(cap_adata)
             self._check_obs(cap_adata)
@@ -287,3 +289,48 @@ def _remove_gene_version(ensemble_ids: pd.Index) -> pd.Index:
         clean_index = ensemble_ids.to_series().apply(lambda x: x.split(".")[0])
         clean_index = pd.Index(clean_index)
         return clean_index
+
+    def _is_csc(self, group_or_dataset) -> bool:
+        """
+        Returns True if HDF5 object represents a CSC sparse matrix.
+        """
+        if not isinstance(group_or_dataset, Group):
+            return False
+
+        encoding = group_or_dataset.attrs.get("encoding-type", None)
+        return encoding == "csc_matrix"
+
+    def _is_csr(self, group_or_dataset) -> bool:
+        if not isinstance(group_or_dataset, Group):
+            return False
+        return group_or_dataset.attrs.get("encoding-type", None) == "csr_matrix"
+
+    def _is_dense(self, group_or_dataset) -> bool:
+        return isinstance(group_or_dataset, Dataset)
+
+    def _validate_x_and_raw_x_formats(self, cap_adata: CapAnnData) -> None:
+        """
+        Validate that X and raw.X (if exists) are dense or CSR.
+        Raise CSCMatrixInX otherwise.
+        """
+        locations = []
+
+        f = cap_adata.file
+
+        # X
+        x = f["X"]
+        if self._is_csc(x):
+            locations.append("X")
+        elif not (self._is_dense(x) or self._is_csr(x)):
+            locations.append("X")
+
+        # raw.X
+        if "raw" in f and "X" in f["raw"]:
+            raw_x = f["raw/X"]
+            if self._is_csc(raw_x):
+                locations.append("raw.X")
+            elif not (self._is_dense(raw_x) or self._is_csr(raw_x)):
+                locations.append("raw.X")
+
+        if locations:
+            raise CSCMatrixInX(locations=locations)

test/test_upload_validator.py

@@ -1,7 +1,12 @@
 import pytest
 import numpy as np
 import pandas as pd
+
 import anndata as ad
+from packaging import version
+if version.parse(ad.__version__) >= version.parse("0.11.0"):
+    ad.settings.allow_write_nullable_strings = True
+
 import scipy.sparse as sp
 from pathlib import Path
 import tempfile
@@ -27,6 +32,7 @@
     AnnDataNonStandardVarError,
     CapMultiException,
     AnnDataNoneInGeneralMetadata,
+    CSCMatrixInX,
 )
 
 TMP_DIR = Path(tempfile.mkdtemp())
@@ -262,3 +268,75 @@ def test_ontology_id_instead_general_metadata(names_provided, with_none):
         with context:
             adata.read_obs(GENERAL_METADATA)
             v._check_obs(adata)
+
+
+def write_adata_with_matrix(path, X, raw_X=None):
+    adata = ad.AnnData(X=X)
+
+    if raw_X is not None:
+        adata.raw = ad.AnnData(X=raw_X)
+
+    adata.write_h5ad(path)
+
+
+def test_csc_in_x_raises(tmp_path):
+    p = tmp_path / "test.h5ad"
+
+    X = sp.csc_matrix(np.eye(5)) # CSC
+    write_adata_with_matrix(p, X=X)
+
+    v = UploadValidator(p)
+
+    with pytest.raises(CSCMatrixInX) as e:
+        with read_h5ad(p, edit=False) as cap_adata:
+            v._validate_x_and_raw_x_formats(cap_adata)
+
+    assert "X" in e.value.message
+
+
+def test_csc_in_raw_x_raises(tmp_path):
+    p = tmp_path / "test.h5ad"
+
+    X = sp.csr_matrix(np.eye(5)) # valid CSR
+    raw_X = sp.csc_matrix(np.eye(5)) # invalid CSC
+
+    write_adata_with_matrix(p, X=X, raw_X=raw_X)
+
+    v = UploadValidator(p)
+
+    with pytest.raises(CSCMatrixInX) as e:
+        with read_h5ad(p, edit=False) as cap_adata:
+            v._validate_x_and_raw_x_formats(cap_adata)
+
+    assert "raw.X" in e.value.message
+
+
+def test_csc_in_both_raises(tmp_path):
+    p = tmp_path / "test.h5ad"
+
+    X = sp.csc_matrix(np.eye(5))
+    raw_X = sp.csc_matrix(np.eye(5))
+
+    write_adata_with_matrix(p, X=X, raw_X=raw_X)
+
+    v = UploadValidator(p)
+
+    with pytest.raises(CSCMatrixInX) as e:
+        with read_h5ad(p, edit=False) as cap_adata:
+            v._validate_x_and_raw_x_formats(cap_adata)
+
+    assert "X and raw.X" in e.value.message
+
+
+def test_dense_and_csr_pass(tmp_path):
+    p = tmp_path / "test.h5ad"
+
+    X = np.random.rand(5, 5) # valid dense
+    raw_X = sp.csr_matrix(np.eye(5)) # valid CSR
+
+    write_adata_with_matrix(p, X=X, raw_X=raw_X)
+
+    v = UploadValidator(p)
+
+    with read_h5ad(p, edit=False) as cap_adata:
+        v._validate_x_and_raw_x_formats(cap_adata) # should not raise