Polarbear

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