v1.4.4

• Fix: reorder dependencies to work around uv resolver order-dependence (astral-sh/uv#5161), which caused uv pip install palantir to resolve to incompatible ancient versions of scanpy/numba/llvmlite

v1.4.3

• Avoid warning about coloring in plot_trajectories
• Fix: forward eigvec_key to fallback method in early_cell()
• Performance optimizations (thanks to @svdrecbd for PR #176):
  • Faster max-min waypoint sampling: O(N·K²) → O(N·K)
  • PCA slicing optimization to avoid redundant recomputation
  • Sparse LU solver for absorption probabilities
  • Vectorized perspective matrix and Dijkstra shortest paths
  • Vectorized Markov chain construction and terminal state detection
  • Optional sklearn backend for diffusion kernel (kernel_backend parameter)
• Fix: run_pca now correctly preserves existing X_pca when using custom pca_key

v1.4.2

• Fix "lightgray" error in plot_trend
• Address pygam incompatibility with newer scipy versions (automatic skip in tests, user warnings)
• Address fcsparser incompatibility with NumPy 2.0 (automatic skip in tests, user warnings)
• Use different joblib backend in Python 3.12+ with joblib < 1.5 to avoid ResourceTracker warnings
• More convenient cell-name handling for integers (e.g., in spatial data) with automatic type conversion

v1.4.1

• update LICENSE file to be consistent with MIT - license
• implement plot_trajectories to show multiple poaths on the UMAP
• Drop leiden dependency to allow Python >= 3.13. igrph is used instead.

v1.4.0

• Made pygam an optional dependency that can be installed with pip install palantir[gam] or pip install palantir[full]
• Added proper conditional imports and improved error handling for pygam
• Enhanced run_magic_imputation to return appropriate data types for different inputs
• Updated code to use direct AnnData imports for newer compatibility
• Improved version detection using importlib.metadata with graceful fallbacks
• Fixed Series indexing deprecation warnings in early cell detection functions
• Expanded and standardized documentation with NumPy-style docstrings throughout the codebase
• Added comprehensive type hints to improve code quality and IDE support
• Remove dependency from _ methods in scanpy for plotting.
• add pseudotime_interval argument to control path length in palantir.plot.plot_trajectory

v1.3.6

• run_magic_imputation now has a boolean parameter sparse to control output sparsity
• bugfix: run_local_variability for dense expression arrays now runs much faster and more accurate

v1.3.4

• avoid devision by zero in select_branch_cells for very small datasets
• make branch selection robust against NaNs
• do not plot unclustered trends (NaN cluster) in plot_gene_trend_clusters

v1.3.3

• optional progress bar with progress=True in palantir.utils.run_local_variability
• avoid NaN in local variablility output
• compatibility with scanpy>=1.10.0

v1.3.2

• require python>=3.8
• implement CI for testing
• fixes for edge cases discoverd through extended testing
• implement plot_trajectory function to show trajectory on the umap
• scale pseudotime to unit intervall in anndata

v1.3.1

• implemented palantir.plot.plot_stats to plot arbitray cell-wise statistics as x-/y-positions.
• reduce memory usgae of palantir.presults.compute_gene_trends
• removed seaborn dependency
• refactor run_diffusion_maps to split out compute_kernel and diffusion_maps_from_kernel
• remove unused dependency tables