Polarbear translates between different single-cell data modalities
install polarbear conda environment (in linux)
conda env create -f environment_polarbear.yml
conda activate polarbear
download data from this link to ./data/
train/evaluate the semi-supervised Polarbear model:
bash run_polarbear.sh babel semi
bash run_polarbear.sh random semi
train/evaluate the Polarbear-coassay model (that only uses co-assay data into training):
bash run_polarbear.sh babel coassay
bash run_polarbear.sh random coassay
generate performance comparison plots between Polarbear and Polarbear co-assay
mkdir -p result/
Rscript bin/evaluate_polarbear.R
Reference: Ran Zhang, Laetitia Meng-Papaxanthos, Jean-Philippe Vert, William Stafford Noble. Semi-supervised single-cell cross-modality translation using Polarbear