v0.4.2

Bugfixes:

• Fixed a bug where only the last layer of an AnnData objects was appended.
• Fixed a small error with the creation of the conda env

v0.4.1

fixed links

v0.4.0

changed name from rapids_singlecell to rapids-singlecell
Added ability to install from pipy
Added updated yaml recipes for Conda environments

Minor Changes:

• minor fix for rapids-23.02 in rank_genes_groups_logreg
• accelerated _get_mean_var
• improvements for tl.tsne:
  * added method argument
  * added metric argument

v0.3.3

cunndata:

• switchted to anndata's getitem function

cunndata_funcs:

• added .layers support for:
  • regress_out
  • scale
  • normalize_total
• changed mean and var calcuations to make them more memory efficent

scanpy_gpu

• changed TSNE defaults

v0.3.2

cunndata

*added .varm
*added .__repr__

cunndata_funcs

*added pca
*added TSVD and Incremental PCA support
*added support for .varm

scanpy_gpu

*added TSVD and Incremental PCA support
*added .uns support for leiden and louvain

v0.3.1

• fixed Louvain and Leiden Clustering for rapids version > 22.08.
• added support for use_weights for leiden and louvain clustering
• added support for neighbors_key for leiden and louvain clustering
• added possibility to use different neighbourhood graphs
• added support multi_target_regresion in cudata_funcs.regress_out
• added support for __getitem__ to be used on .var
• added support for slices to be used on __getitem__

v0.3.0

v.0.3.0

cunnData

• added support for spatial transcriptomics
• added .obsm to cunnData
• moved methods from class to cunnData_funcs

cunnData_funcs

• most methods of cunnData are now functions the can be used on the cunnData object.

scanpy_gpu

• split functions into separate files

decoupler_gpu

• added first two accelerated decoupler functions:
  • run_mlm
  • run_wsum

v.0.2.2

Minor Bugfixes

• Fixed dtype bug after using normalize_pearson_residuals.

v0.2.1

added support for:

• highly_variable_genes based on poisson_gene_selection
• pyMDE embedding
• harmony_integrate batch correction
• some improvements for tsne, leiden, louvain

v.0.2.0

v0.2.0

new functionalities and improvements:

• support for layers in cunndata
• support for selecting HVGs based on seurat_v3 and pearson_residuals
• layers support for highly_varible_genes
• new function for normalizing based on pearson_residuals
• calculate_qc got sped up
• layers support for pca and rank_genes_groups_logreg
• new function for calculating embedding_density