CellDISECT v0.1.5

What's Changed

• Gene importance, Cell Generation, and Single covariate handling by @ArianAmani in #34

Full Changelog: v0.1.3...v0.1.5

CellDISECT v0.2.0b1

0.2.0b1 (2024-03-16)

• Beta release with compatibility updates
• Updates:
  • Compatible with Google Colab
  • Support for newer torch versions (>=2.1.0, <2.3.0)
  • Support for newer scvi-tools versions (>=1.0.0, <=1.3.0)
• Note:
  • This beta version is designed to work with modern environments while the main branch remains for reproducibility
  • Although this version introduces breaking changes, it is still backward compatible with the previous version (training and numerical process could be different due to different torch versions, hence slightly different results)

Full Changelog: v0.1.3...v0.2.0b1

CellDISECT v0.1.3

Release version 0.1.3

CellDISECT v0.1.2

CellDISECT v0.1.2

Initial release of Cell DISentangled Experts for Covariate counTerfactuals (CellDISECT).

Features

• Causal generative model for disentangling known covariate variations
• Counterfactual prediction capabilities
• Compatible with Python >=3.9, <3.11
• Integration with scvi-tools, scanpy, and other single-cell analysis tools

Dependencies

• torch >=2.1.0, <2.3.0
• scvi-tools >=0.20.3, <1.0.0
• anndata >=0.10.8, <0.10.9
• And other dependencies as specified in pyproject.toml

Installation

pip install celldisect