scPEDSSC: proximity enhanced deep sparse subspace clustering method for scRNA-seq data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) is vital for analyzing cellular heterogeneity.
• Clustering scRNA-seq data to identify cell types faces challenges like high dimensionality, noise, and sparsity.

Purpose of the Study:

• To introduce scPEDSSC, a novel deep sparse subspace clustering method.
• To improve cell type identification in scRNA-seq data through proximity enhancement.

Main Methods:

• scPEDSSC utilizes a deep auto-encoder with a two-part generalized gamma (TPGG) distribution to learn a self-expression matrix (SEM).
• The SEM and its second power are used to generate a similarity matrix for clustering.

Main Results:

• scPEDSSC demonstrated superior performance compared to eight state-of-the-art methods on twelve real biological datasets.
• Experimental validation confirmed the effectiveness of the proposed scPEDSSC method.

Conclusions:

• scPEDSSC offers a robust and effective approach for cell type identification in scRNA-seq data.
• The method addresses key challenges in scRNA-seq data analysis, advancing single-cell studies.