Changelog

Changelog#

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]#

[v0.2.5]#

Added#

  • Added batch_size parameter to Kernel.compute_neighbors() for backend-agnostic batching #60

  • Added subset parameter to plot_confusion_matrix() for filtering cells #60

  • Added Python 3.14 support #62

Changed#

  • Default postfixes now include underscore: "_pred" and "_conf" instead of "pred" and "conf" #60

  • Switched from MathJax to KaTeX for documentation math rendering #61

  • Updated GitHub Actions to latest versions (checkout v5, setup-uv v7) #61

  • Updated test matrix to Python 3.11/3.14 #62

Fixed#

  • Fixed PackageNotFoundError when checking rapids availability in conda environments #60

  • Fixed stale k-NN state when compute_neighbors() fails #60

  • Fixed plot_confusion_matrix() handling of NaN values in both y_true and y_pred #60

  • Fixed plot_confusion_matrix() handling of mismatched category sets #60

  • Fixed plot_confusion_matrix() handling of float categories #60

  • Fixed importlib.resources.files() compatibility with Python 3.14 #62

[v0.2.4]#

Changed#

  • Deprecated Python 3.10 support #55

  • Updated to cookiecutter-scverse template v0.6.0 #54

[v0.2.3]#

Changed#

  • Improved returning probabilities (after mapping categorical obs fields) to always return a DataFrame #49

Added#

  • Add the possibility to adjust the library size after .map_layers #50

  • Added the option to turn off post-processing of the presence scores, so that they can be first smoothed and then processed, like in HNOCA-tools #51

[v0.2.2]#

Added#

  • Enabled subsetting categories before mapping .obs values #46

Changed#

  • Updated the README a bit #44

  • Updated tutorials to work with new parameter names #43

Fixed#

  • Fixed a small bug where hvg masks would not be propagated correctly to joint pca computation #48

[v0.2.1]#

Changed#

  • Move some duplicated docstrings into a central _docs.py file #41

Added#

  • Added some tests for edge cases in the MappingOperator class #41

  • Treat faiss-cpu and faiss-gpu separately #41

[v0.2.0]#

Added#

  • Added a tutorial on same-modality query to reference mapping #38

  • Added a tutorial on data smoothing #37

  • Added an option to return the mapping probabilities for categorical .obs mapping #39

  • Added a MappingOperator class which allows for iterative mapping matrix applicatino in self-mapping mode #35

  • Add the umap method to compute symmetric k-NN connectivities in self-mapping mode #34

Changed#

  • Refectored the neighbors classes into a Neighobrs and a Kernel class and moved symmetrization into the Kernel class #36

[v0.1.4]#

Changed#

  • Rename mapping methods to map_obs, map_obsm, and map_layers, and improve support for numerical .obs annotations #30.

[v0.1.3]#

Added#

  • Added a tutorial on spatial contextualization and niche identification #23.

  • Implemented a self-mapping mode with only a query dataset #21.

  • Allow importing a pre-computed dataset of transfered expression values #21.

  • Allow importing pre-computed neighborhood matrices #21.

  • Add a tutorial on spatial contextualization and niche identification #21.

  • Add an equal-weight kernel #22.

[v0.1.2]#

Added#

  • Included tests for the check module, and more tests for the main classes #15.

  • Implemented the computation of presence scores, following HNOCA-tools #16.

  • Added a groupby parameter to expression transfer evaluation #16.

  • Added a test_var_key parameter to expression transfer evaluation #19.

  • Added a tutorial on spatial mapping #19.

[v0.1.1]#

Changed#

  • Switched to vcs-based versioning #5.

Added#

  • Added PyPI badge.

[v0.1.0]#

Initial package release.