1.4.0

What's Changed

• 1.4.0 by @dbdimitrov in #138

Full Changelog: 1.3.0...1.4.0

1.3.0

What's Changed

• v1.3.0 by @dbdimitrov in #125

Full Changelog: 1.2.1...1.3.0

1.2.1

What's Changed

• v1.2.1 by @dbdimitrov in #119
• set required Python to >=3.8: the <3.12 causes a broad incompatibility issue in package management by @deeenes in #114

New Contributors

• @deeenes made their first contribution in #114

Full Changelog: 1.2.0...1.2.1

1.2.0

What's Changed

• 1.2.0 by @dbdimitrov in #110
• Update _tileplot.py for plotnine >=0.13 by @redst4r in #108

New Contributors

• @redst4r made their first contribution in #108

Full Changelog: 1.1.0...1.2.0

1.1.0

1.1.0 (12.04.2024)

---

• Added a check for the subset of cell types in li.multi.dea_to_lr. Related to #92.

• Split Local and Global Bivariate metrics. Specifically, I reworked completely the underlying code, though the API should remain relatively unchanged. With the exceptions of: 1) lr_bivar is now removed and bivar has been renamed to bivariate. This allowed me to remove a lot of redundancies between the two functions. 2) nz_threshold has been renamed to nz_prop for consistency with expr_prop in the remainder of the package. Related to #44.

• li.mt.bivariate parameter mod_added has been renamed to key_added due to this now refer to both .obsm and .mod - depedening whether an AnnData or MuData object is passed.

• Added Global Lee's statistic, along with a note on the weighted product that upon z-scaling it is equivalent to Lee's local statistic.

• The Global L statistic and Global Moran's R are themselves basically identical. See Eq.22 from Lee and Eq.1 in Supps of SpatialDM.

• Changed the li.mt.bivar parameter function_name to local_name for consistency and to avoid ambiguity with the newly-added global_name parameter.

• Added bumpversion to manage versioning. Related to #73.

• Added max_runs and stable_runs parameters to enable the inference of robust causal networks with CORNETO. Related to #82.

• Optimized MISTy such that the matrix multiplication by weights is done only once, rather than for each target. Users can now obtain the weighted matrix via the misty.get_weighted_matrix function.

• MISTy models are now passed externally, rather than being hardcoded. This allows for more flexibility in the models used. As an example, I also added a RobustLinearModel from statsmodels. Related to #74.

• Removed forced conversion to sparse csr_matrix matrices in MISTy. Related to #57.

1.0.5

What's Changed

• v1.0.5 by @dbdimitrov in #91

• Added ScSeqComm Method, implemented by @BaldanMatt (#68)

• Added functions to query a metabolite-receptor interactions database (MetalinksDB), including:
  => li.rs.get_metalinks to get the database
  => li.rs.get_metalinks_values to get the distinct annotation values of the database
  => describe_metalinks to get a description of the database

• Added a metabolite-mediated CCC tutorial in spatially-resolved multi-omics data (#45).

• Changed hardcoded constants to be defined in _constants.py

• Excluded CellChat from the default rank_aggregate method

• Fixed return logic of SpatialBivariate

• li.mt.process_scores is now exported to li.mt

• Changed the default max_neighbours in li.ut.spatial_neighbors to 1/10 of the number of spots.

Full Changelog: 1.0.4...1.0.5

1.0.4

See Release notes

v1.0.3

See release notes.

v1.0.2

See Release Notes

1.0.1 Stable Release

See Release notes