SAILERX

Pytorch implementation and tutorial for SAILERX.

Setup

Clone the repository.

git clone https://github.com/uci-cbcl/SAILER.git

Navigate to the root of this repo and setup the conda environment.

conda env create -f deepatac.yml

Activate conda environment.

conda activate deepatac

Data

Please download data here and setup your data folder as the following structure:

SAILERX
|___data  
    |___...

Usage

For full description, please see

python train.py -h

Standard training

To train with one multimodal sc-deq data (scRNA-seq + scATAC-seq). Using PBMC 10k as an example.

python train.py -d pbmc10k -cuda 0 --pos_w 20

Training with data from two batches

To train with multiple multimodal sc-deq data (scRNA-seq + scATAC-seq). Using PBMC 10k + 3k as an example.

python train.py -d pbmc_batch -cuda 0 --pos_w 20 -batch True

Hybrid training

To train with multiple multimodal sc-deq data (scRNA-seq + scATAC-seq). Using PBMC 10k + 3k as an example.

python train.py -d pbmc_hybrid -cuda 0 --pos_w 20 -batch True -t hybrid

Evaluation

To generate the embedding, please use the following command to load a specific model you want, subsititute PATH_TO_CKPT to the path to the ckpt you want to use (i.e., ./models/main/398.pt).

python eval.py -d pbmc10k --name main -cuda 0 -l PATH_TO_CKPT

For more info, please use

python train.py -h

Or see evaluation examples here.