A topology-preserving dimensionality reduction method for single-cell RNA-seq data using graph autoencoder.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• High-dimensional single-cell RNA sequencing (scRNA-seq) data requires dimensionality reduction for interpretation.
• Preserving the topological structure of cells in reduced dimensions is a significant challenge.

Purpose of the Study:

• To introduce the single-cell graph autoencoder (scGAE), a novel dimensionality reduction technique for scRNA-seq data.
• To enhance the preservation of topological structure and feature information in scRNA-seq data.

Main Methods:

• scGAE constructs a cell graph and employs a multitask graph autoencoder.
• The method simultaneously preserves topological structure and feature information.
• Extended scGAE for visualization, clustering, and trajectory inference.

Main Results:

• scGAE accurately reconstructs developmental trajectories and separates cell clusters in simulated data.
• Outperforms existing deep learning methods in various scenarios.
• Provides novel insights into cell lineages and preserves inter-cluster distances in empirical data.

Conclusions:

• scGAE is an effective method for dimensionality reduction in scRNA-seq data.
• The approach enhances the biological interpretability of complex single-cell datasets.
• Offers improved performance over current deep learning techniques for scRNA-seq analysis.