0.23.1

CHANGELOG

Improvements

• Added keep_bounds parameter in Metadata.sift and DataStore.filter_cells

0.23.0

CHANGELOG

Breaking changes introduced ⚠️

Improvements

• save_normed_for_query now uses parallel writing.
• modified iter_normed_feature_wise to export multiple formats (dataframe or ndarray).
• multiple fixes in docstrings.
• improved chunk sizing in smoothen_dists in knn_utils.
• changed the location of mu and sigma in Zarr hierarchy to avoid recomputation.
• optimized ann index saving for non local zarr

New functionality

• Added bed_to_sparse_array to writers module.
• Major rewrite of run_marker_search to include multiple summary stats about marker genes

0.22.2

CHANGELOG

Since 0.21.0

Bug fixes

• Fixes in tqdm usage
• Bug fixes and performance improvement in SparseToZarr

Improvements

• Reading obsm slot from H5ad file now supported.
• removed dependency on sparse library. Switched to scipy.sparse.coo_matrix

0.21.0

CHANGELOG

Bug fixes

• The index in anndata object obtained from to_anndata is now set to ids column to avoid name duplication.

Improvements

• The 'var', 'obs' and 'obsm' column in H5ad file are populated as expected when using to_h5ad. Known issue: PCA is still not exported (need to find a solution to generate PCA coords for this purpose).

0.20.1

CHANGELOG

Improvements

• Added assertion checks and changed overwrite behaviour in ZarrMerge

0.20.0

CHANGELOG

Bug fixes

• Log message formatting in RNAassay.mark_hvgs

New features

• Added CSVReader
• Added CSVtoZarr

0.19.8

CHANGELOG

redeploy to due Pypi sync error

0.19.7

CHANGELOG

Bug fixes
Bug fix in OSFDownloader

0.19.6

CHANGELOG

Doc fixes
Optimized requirements for readthedocs

0.19.5

CHANGELOG

Bug fixes
Explicit Unicode decoding in create_zarr_obj_array. The bug was discovered when using H5adReader to read 'obs' entries.