scGNN is a novel graph neural network framework for single-cell RNA-Seq analyses

Abstract

Single-cell RNA-sequencing (scRNA-Seq) is widely used to reveal the heterogeneity and dynamics of tissues, organisms, and complex diseases, but its analyses still suffer from multiple grand challenges, including the sequencing sparsity and complex differential patterns in gene expression. We introduce the scGNN (single-cell graph neural network) to provide a hypothesis-free deep learning framework for scRNA-Seq analyses. This framework formulates and aggregates cell–cell relationships with graph neural networks and models heterogeneous gene expression patterns using a left-truncated mixture Gaussian model. scGNN integrates three iterative multi-modal autoencoders and outperforms existing tools for gene imputation and cell clustering on four benchmark scRNA-Seq datasets. In an Alzheimer’s disease study with 13,214 single nuclei from postmortem brain tissues, scGNN successfully illustrated disease-related neural development and the differential mechanism. scGNN provides an effective representation of gene expression and cell–cell relationships. It is also a powerful framework that can be applied to general scRNA-Seq analyses.

Define and visualize pathological architectures of human tissues from spatially resolved transcriptomics using deep learning

“…Parameter setting. Parameters in scGNN to generate embedding are referred to the previous study 10 . In the case study of the AD sample, in analysis on cortical layers 2 & 3, the expressions of 8 well-defined marker genes were log-transformed and embedded by spaGCN with 0.65 resolution.…”

“…The measured gene expression values of the spot are treated as the node attributes, and the neighboring spots adjacent in the Euclidean space on the tissue slice are linked with an undirected edge. This lattice-like spot graph is modeled by our graph neural network (GNN) based tool scGNN 10 , which learns a three-dimensional embedding to preserve the topological relationship between all spots in the spatial space of transcriptomics. The three-dimensional embedding on gene expression is mapped to three color channels as Red, Green, and Blue in an RGB image, which is naturally visualized as an image of the spatial gene expression.…”

Self-supervised contrastive learning for integrative single cell RNA-seq data analysis

“…The choice of k for the widely used k-nearest-neighbors graph affects the size and number of final clusters 7 . Because of the model capacity and scalability of deep learning methods, almost all the recently developed methods are based on antoencoder 5,9,[20][21][22][23][24][25] (AE) or variational autoencoder 8,26,27 (VAE), which can also incorporate the biostatistical models 28,29 seamlessly. However, as AE and VAE methods are unsupervised learning methods, it is very difficult to control and decide what the deep learning models will learn, although some very recent studies try to impose constraints and our prior knowledge about the problem onto the low-dimensional space 5,27 .…”

A Robust and Scalable Graph Neural Network for Accurate Single Cell Classification

“…In fact, the high-order representation and topological relations could been naturally learned by the graph neural network (GNN), and GNN have been proven with improved performance in scRNA-seq data analyses such as imputation and clustering [27–29]. ScGCN[30] is currently the only graph neural network method for annotating cells.…”