Scarf

CHANGELOG
Changelog of 0.19.0 is also included here

**Critical bug fix**
Fixed critical bug in `iter_normed_feature_wise` that caused usage of the wrong parameter in the `normed` method call.

**New features**
Added `save_normed_for_query` which will create a new slot in the assay and save all the normalized values for the quick lookup. `get_cells_vals` was improved to utilize these calculated normalized values when available.

**Improvements**
`iter_normed_feature_wise` now features any selection of features.
Added `title` parameter to `plot_layout` and `plot_unified_layout` that allows using custom titles in the plots.
`create_zarr_obj_array` now supports compression and disabling of chunking.

**Documentation changes**
API page now included class __init__ level docstrings.

CHANGELOG

**Critical bug fix**
`ZarrMerge` bug which led to +1 counts for all the features was fixed
Fixed a bug in `DataStore.mark_hvgs` that caused the failure of ANN search when reusing 0 dims.

**Improvements**
Improved handling for groups with no markers in `DataStore.export_markers_to_csv` and `DataStore.get_markers`
`hide_title` and `title_size` parameters included in `plot_scatter` and `shade_scatter` and exposed in `DataStore.plot_layout` and `DataStore.plot_unified_layout`

**Documentation changes**
Improved instructions for installation on Windows
API page now includes classes from `readers`, `writers` and `metadata` modules

**Dev notes**

CHANGELOG

**Improvements**
Bug fixes in `find_markers_by_regression` and `find_markers_by_rank` causing unnecessary dataframe (gene-wise chunk) renaming on the wrong axis
Bugfix in `handle_download` causing wrong tar extraction target
Bugfix in `_scatter_make_colors` causing wrong cmap choice
`plot_marker_heatmap` raises error now when no marker gene is found

**Documentation changes**
Removed Dask warning from homepage
code style made black-like on all the vignettes

**Dev notes**

CHANGELOG


**New features**
Rewrote `meld_assay` module and exposed functionality of `coordinate_melding` through `add_melded_assay` method of `DataStore`. Coordinate intersections are done using `binary_search` function that employs a Numba accelerated implementation of binary search algorithm. This means that we don't need to use Bedtools anymore, which is hard to install on windows. `create_counts_mat` function modified to used normalized data and perform sparse write operations. `meld_assay` function renamed to `coordinate_melding`

**Added tutorials**
Added GeneScores section to scATAC-Seq basic workflow

**Improvements**
Added `lsi_skip_first` parameter to `make_graph` and `AnnStream`. The first LSI dimension usually captures sequencing depth so this parameter helps users control if they want to keep or discard this dimension.

**Documentation changes**
Changed order of vignettes in the TOC tree.
Added pseudotime vignette to colab list
Improved installation instructions for Windows

CHANGELOG

CHANGELOG

**Improvements**
Fixes in `get_markers` in `DataStore` to allow backward compatibility with marker search results.