A Hybrid Clustering Algorithm for Identifying Cell Types from Single-Cell RNA-Seq Data.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) provides insights into cell differentiation and subtype variation.
• Unsupervised machine learning, particularly clustering, is vital for scRNA-seq analysis.
• Existing methods face challenges like high dimensionality, unstable results, and complex parameter tuning.

Purpose of the Study:

• To develop an improved unsupervised method for identifying cell subpopulations in scRNA-seq data.
• To address limitations of current clustering techniques in scRNA-seq analysis.
• To offer a robust and user-friendly approach for cell subpopulation identification.

Main Methods:

• A novel method combining structure entropy and k-nearest neighbors (KNN) is proposed.
• The method identifies natural communities by minimizing structure entropy, eliminating the need to pre-specify cluster numbers.
• Applied and validated on eight diverse scRNA-seq datasets.

Main Results:

• The proposed method demonstrated superior performance compared to three established benchmark methods (nonnegative matrix factorization, single-cell interpretation via multikernel learning, and structural entropy minimization principle).
• Consistent improvements were observed across multiple scRNA-seq datasets.
• The approach successfully identified distinct cell subpopulations without prior knowledge of cluster numbers.

Conclusions:

• The structure entropy and KNN-based method offers a robust and effective solution for cell subpopulation identification in scRNA-seq data.
• This approach overcomes key limitations of existing clustering methods, enhancing the analysis of single-cell data.
• The findings suggest a promising new direction for computational analysis in single-cell genomics.