thyra 3.2.0__tar.gz → 3.2.1__tar.gz

{thyra-3.2.0 → thyra-3.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: thyra
-Version: 3.2.0
+Version: 3.2.1
 Summary: A modern Python library for converting Mass Spectrometry Imaging (MSI) data into SpatialData/Zarr format - your portal to spatial omics
 License: MIT
 License-File: LICENSE

{thyra-3.2.0 → thyra-3.2.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "poetry.core.masonry.api"
 
 [tool.poetry]
 name = "thyra"
-version = "3.2.0"
+version = "3.2.1"
 description = "A modern Python library for converting Mass Spectrometry Imaging (MSI) data into SpatialData/Zarr format - your portal to spatial omics"
 authors = ["Theodoros Visvikis <t.visvikis@maastrichtuniversity.nl>"]
 maintainers = ["Theodoros Visvikis <t.visvikis@maastrichtuniversity.nl>"]

{thyra-3.2.0 → thyra-3.2.1}/thyra/__init__.py

@@ -36,7 +36,7 @@ warnings.filterwarnings(
     category=FutureWarning,
 )
 
-__version__ = "3.2.0"
+__version__ = "3.2.1"
 
 # Import key components - avoid wildcard imports
 try:

{thyra-3.2.0 → thyra-3.2.1}/thyra/converters/spatialdata/streaming_converter.py

@@ -1278,18 +1278,28 @@ class StreamingSpatialDataConverter(BaseSpatialDataConverter):
 
                 nnz = len(mz_indices)
                 if nnz > 0:
-                    # Count entries per column (vectorized)
-                    np.add.at(col_counts, mz_indices, 1)
-                    total_nnz += nnz
-
-                    # Accumulate for average spectrum (vectorized)
+                    # Accumulate for average spectrum (vectorized). Left
+                    # outside the bounds check to match the COO pre-scan,
+                    # which also averages over every spectrum read rather
+                    # than every spectrum stored.
                     np.add.at(total_intensity, mz_indices, resampled_ints)
 
-                    # TIC and occupancy. Both are indexed by grid position,
-                    # so both need the bounds check: a reader yielding a
+                    # Column counts, TIC and occupancy all describe a
+                    # spectrum that is going to get a row, so all three sit
+                    # behind the bounds check: a reader yielding a
                     # coordinate outside the declared dimensions would
                     # otherwise wrap round and land on an unrelated pixel.
+                    #
+                    # col_counts and total_nnz size the CSC arrays, and the
+                    # scatter pass skips exactly the spectra this rejects
+                    # (row_of_grid gives them no row). Counting them here
+                    # reserved slots nothing ever wrote: the memmap is
+                    # zero-filled, so they surfaced as explicit zeros at
+                    # row 0, out of order within their column, which is a
+                    # matrix scipy reports as non-canonical.
                     if 0 <= y < n_y and 0 <= x < n_x:
+                        np.add.at(col_counts, mz_indices, 1)
+                        total_nnz += nnz
                         tic_values[y, x] = resampled_ints.sum()
                         occupancy[y * n_x + x] = True
                     else: