pertpy
pertpy copied to clipboard
Add linear mixed effects model to DE interface
Description of feature
Description of feature
They can be used to either perform tests on all cells, including the sample as random effect, or on pseudobulk when there are designs that require random effects.
Methods
- MAST (https://github.com/scverse/multi-condition-comparisions/issues/19)
- DREAM
- Statsmodels LME
Model specification
- Neither formulaic nor patsy currently allow for random effects. There's a PR open to add support https://github.com/matthewwardrop/formulaic/pull/34
- formulae allows specifying random effects: https://github.com/bambinos/formulae
- Alternatively, it could be possible to specify two formulas, one for fixed and one for random effects.
Migrated from https://github.com/scverse/multi-condition-comparisions/issues/20