mengxu98 / scapgnn

Graph Neural Network-Based Framework for Single Cell Active Pathways and Gene Modules Analysis

scapGNN：Graph Neural Network-based Framework for Active Pathway and Gene Module Inference from Single-cell Multi-omics Data

Version: 0.1.4

Installation

1. CRAN
install.packages("scapGNN")

2. GitHub
library(devtools);
install_github("XuejiangGuo/scapGNN")

Notice:
The ConNetGNN() function of scapGNN is implemented based on pytorch, so an appropriate python environment is required:

python >=3.9.7
pytorch >=1.10.0 (CPU)
sklearn >=0.0
scipy >=1.7.3
numpy >=1.19.5

We also provide environment files for conda: /inst/extdata/scapGNN_env.yaml. Users can install it with the command: conda env create -f scapGNN_env.yaml.

Descriptions

scapGNN is a uniform framework based on the graph neural network (GNN) for single-cell active pathway and cell phenotype-associated gene modules analysis. We model a GNN to generate the latent features of the single-cell data, which learns and infers gene-cell associations. It convert the sparse unstable single-cell profiling data into a stable gene-cell association network by aggregating adjacent node information. The genes with dropouts or low expressions were considered by taking into account the association effect with other nodes (Fig 1). For single-cell multi-omics data, we use a network fusion method to merge gene-cell association network from different omics. The random walk with restart (RWR) algorithm further measures pathway activity scores and identify cell phenotype-associated gene modules. We use real and simulated single-cell datasets to benchmark the performance of scapGNN, and found that it performed better than the state-of-art methods in cell clustering, identification of active pathway and cell phenotype-associated gene modules (Fig 2).

Fig 1. An overview of the scapGNN framework. A The input is the gene-cell matrix of scRNA-Seq or gene activity matrix generated from scATAC-seq. A graph-based autoencoder, which contains a deep neural network autoencoder and a graph convolutional autoencoder, learns the latent associations between genes and cells. The RWR algorithm quantifies pathway activity and identifies cell phenotype-associated gene modules. B The main capabilities of scapGNN include inferring single-cell pathway activity profiles, constructing cell cluster association networks, identifying cell phenotype-associated gene modules under multiple cell phenotypes, and quantifying the importance of genes in the pathway.

Fig 2. The workflow of integrating single-cell multi-omics data by scapGNN. First, GNN model of scapGNN constructs gene-cell association network for gene expression profiles of scRNA-seq data and gene activity matrix of scATAC-seq data, respectively. Second, the Brown's method integrates two gene-cell association networks to combined gene-cell association network. Finally, the RWR algorithm is used to calculate pathway activity scores and identify cell phenotype-associated gene modules with multi-omics information.

Codes and used

For detailed functions and usage instructions, please refer to https://cran.r-project.org/web/packages/scapGNN/index.html or use help() to see the documentation for each function after installing scapGNN package.

Reference

The paper is being submitted.