GeneCluster 0.1.0

A gene clustering model using deep learning

GeneCluster

GeneCluster, a gene clustering method for Single Cell RNA-seq data.

Requirements

python --- 3.8.10

scanpy --- 1.8.2

umap-learn --- 0.5.3

torch --- 1.8.1

torchvision --- 0.9.1

faiss-gpu --- 1.7.1

Usage

The raw single-cell data in h5 format first goes through DataProcess.py to get the gene-cell in csv format, and the csv format is fed into the model for training.

# Run main.py directly with specified parameters
python main.py --nmb_cluster 20 --batch 128 --epochs 100 \
               --lr 0.01 --data_path normal_data.csv \
               --ckpt_path train_res &

# Alternatively, run the GeneCluster package with parameters
GeneCluster --data_path path/to/dataset.csv --epochs 100 --nmb_cluster 20 --batch 128 --ckpt_path train_res   

or

sh train.sh

Please note that the model outputs gene embeddings extracted by training (features.npy), which can be used to predict gene co-expression relationships and identify gene modules.

Here is an example on pbmc3k.h5ad scRNA-seq data.

Example

from genecluster import main

args = {
    "data_path": "path/to/dataset.csv",
    "epochs": 100,
    "nmb_cluster": 10,
    # other arguments
}

main(args)

Contact

Yuting Bai ([e-mail])