Benchmark on sci-CAR dataset

theislab / multimil

Multimodal weakly supervised learning to identify disease-specific changes in single-cell atlases

https://multimil.rtfd.io/

BSD 3-Clause "New" or "Revised" License

18 stars 3 forks source link

Benchmark on sci-CAR dataset #27

Closed awakeenderman closed 3 years ago

awakeenderman commented 3 years ago

[x] Fine-Tune Multigrate in paired setting
[x] Run MOFA2 on the dataset in paired setting
[ ] Run Seurat v4 on the dataset in paired setting
[ ] Prepare comparison plots

awakeenderman commented 3 years ago

We get poor results on sci-CAR, even in paired setting. lowering KL and integration coefficients, increasing n_iters, and lowering the learning rate may help.

alitinet commented 3 years ago

moved to reproducibility repo