0.12.0

CHANGELOG

New features
Multimodal data integration by graph merging. Integrated graphs are supported by run_umap, run_clustering and run_leiden_clustering
Added smart_label method to DataStore

Added tutorials
Expanded the multi-modal data handling tutorial to demonstrate graph integration functionality
Added MNIST dataset notebook

Improvements
MtxDirReader now supports custom index offset
Warning suppression for division by zero in run_umap
plot_cluster_tree supports non-integer labels

Dev notes
.codecov.yml moved to github dir and edited to not show checks along with github tests
removed push_pypi.sh from root as pypi released are now handled by github actions

0.11.0

CHANGELOG

Breaking changes introduced in run_umap and run_unified_umap method of DataStore
fit_n_epochs parameter renamed to n_epochs. tx_n_epochs parameter is no longer valid.

Improvements
~100% coverage in docstrings in assays module
Fixed a bug in file counter in fetch_dataset
Improved iteration counter in smoothen_dists. Potentially more memory optimized now.
Removed unnecessary transform method in umap module. Disabled removal of low weight edges. Removed call to fuzzy_simplical_set because steps are already performed upstream.

New feature
Added DensMap algorithm to umap module.

0.10.1

CHANGELOG

Improvements
Removed holoviews and EoN as dependencies.
Added hierarchy_pos from EoN directly into plots.py
Added docstrings in assays module
Added few FAQs and minor improvements to documentation

0.10.0

CHANGELOG

Improvements
Improved logging. Moved a lot of unnecessary log messages to DEBUG level so that they will not be shown by default.
Progress bars for dask operations are now rendered through tqdm.
tqdm progress bar now can be modified by users. Optimized tqdm usage for jupyter notebooks.
handle_download now uses the requests library rather than OS specific download utilities. Shows progress of the download.
More API tests were added for scATAC-Seq data, utilities, mapping functions and downloading.

New feature
Added set_verbosity global function that can change the verbosity of log messages. This exposes the loguru log levels.
Exposed verbosity in DataStore.run_umap

For developers
Switched to 'black' code style.
Docs theme was switched to jupyter-book-theme. Most of the docs file converted from rst to md format and are now parsed using My-ST parser.

0.9.2

CHANGELOG

Improvements
Added several more tests
Small bug fixes plots, reader and writers modules

0.9.1

CHANGELOG:

Improvements

• Improved testing for DataStore class (thanks to @JohanRodhe for contribution)
• Improved default color palettes (inspired by Scanpy)
• Improve legends display for plot_layout. Cleaned up code in plots.plot_scatter and all affected DataStore wrappers.

Known issues

• Tests fail on MacOS
• Unable to reproduce results of Leiden and UMAP on Github actions. Can reproduce locally

Other changes

• Minimized README for Github repo
• Improved logo layout for Github README
• Improved logo formats for Github and RTD
• Using Github actions for testing now. Not using CircleCI anymore. This was motivated by difficulty of changing testing OS.

0.9.0

CHANGELOG

Package name on PyPi changed to scarf
To install Scarf:
pip install scarf

0.8.5

CHANGELOG:

Fixes in github publish workflow

0.8.4

CHANGELOG:

Several fixes in Manifest.in including renaming from Manifest.ini
This release is to test if github actions based pypi upload works

0.8.3

CHANGELOG:

Additions

Added two new vignettes. zarr_explanation vignettes integrated in toc tree. plotting vignette (thanks to @JohanRodhe) not integrated in toc yet.

Improvements

Bug fix in ZarrMerge. The bug caused omission of sample names in ids
Minor docstring fixes in DataStore.plot_layout by @JohanRodhe
FIxes in plots.shade_scatter by @JohanRodhe