scPEGEnhanced Graph Convolutional Sparse Subspace Clustering Method for scRNA-Seq Data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Accurate cell type identification from single-cell RNA sequencing (scRNA-seq) data is crucial for biological analysis.
• Challenges in scRNA-seq data include high dimensionality, noise, and sparsity, hindering robust clustering.
• Existing clustering methods often struggle with these inherent data characteristics.

Purpose of the Study:

• To propose a novel proximity-enhanced graph convolutional sparse subspace clustering method (scPEGSSC) for scRNA-seq data.
• To improve the accuracy and robustness of cell type identification in scRNA-seq analysis.
• To overcome the limitations of current clustering techniques when applied to complex single-cell data.

Main Methods:

• scPEGSSC utilizes a graph autoencoder to learn a self-expression matrix (SEM).
• A similarity matrix is generated from the SEM and further enhanced by its square.
• The method employs proximity enhancement and graph convolutional sparse subspace clustering principles.

Main Results:

• scPEGSSC was evaluated on thirteen diverse, real-world biological datasets.
• The proposed method demonstrated superior performance compared to eleven state-of-the-art single-cell clustering techniques.
• Consistent improvements in clustering accuracy were observed across the majority of tested datasets.

Conclusions:

• scPEGSSC offers a significant advancement in scRNA-seq data clustering.
• The method effectively handles the inherent challenges of scRNA-seq data, leading to more reliable cell type identification.
• scPEGSSC represents a promising tool for researchers in genomics and computational biology.