scASGC: An adaptive simplified graph convolution model for clustering single-cell RNA-seq data.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables the identification of cellular heterogeneity and developmental trajectories.
• Accurate cell subpopulation identification is critical for scRNA-seq data analysis.
• Existing unsupervised clustering methods struggle with data dropouts, high dimensionality, and computational time, often failing to capture cell-cell associations.

Purpose of the Study:

• To introduce scASGC, a novel unsupervised clustering method for scRNA-seq data.
• To address the limitations of existing clustering techniques, including speed and accuracy.
• To improve the identification of cell subpopulations and potential cell-cell interactions.

Main Methods:

• Developed scASGC, an adaptive simplified graph convolution model.
• Constructed cell graphs to represent cell relationships.
• Employed a simplified graph convolution model for neighbor information aggregation.
• Implemented adaptive determination of optimal convolution layers for diverse graphs.

Main Results:

• scASGC demonstrated superior performance compared to classical and state-of-the-art clustering methods across 12 public datasets.
• The method effectively identified distinct marker genes in a mouse intestinal muscle dataset (15,983 cells).
• scASGC offers a more efficient and accurate approach to scRNA-seq data clustering.

Conclusions:

• scASGC provides a robust and efficient solution for unsupervised clustering of scRNA-seq data.
• The method enhances the discovery of cellular heterogeneity and novel cell subpopulations.
• scASGC facilitates more precise biological insights from complex single-cell datasets.