SinNLRR: a robust subspace clustering method for cell type detection by non-negative and low-rank representation

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) offers unprecedented resolution for biological studies.
• Clustering cells in scRNA-seq data is crucial for understanding cellular heterogeneity and diversity.
• Existing clustering methods often fall short due to the unique challenges of scRNA-seq data, including high noise and the need for specialized algorithms.

Purpose of the Study:

• To develop a novel and robust computational method for cell type detection in scRNA-seq data.
• To address the limitations of existing clustering approaches by leveraging similarity learning.
• To improve the accuracy and reliability of cell clustering in the presence of high-throughput sequencing noise.

Main Methods:

• Proposed SinNLRR, a novel scRNA-seq cell type detection method based on similarity learning.
• Incorporated non-negative and low-rank structures into the similarity matrix.
• Utilized the alternating direction method of multipliers for optimization and developed an adaptive penalty selection method.
• Integrated the learned similarity matrix with spectral clustering, t-distributed stochastic neighbor embedding, and Laplace score.

Main Results:

• SinNLRR demonstrated robust and accurate cell clustering performance across diverse scRNA-seq datasets.
• The method effectively resolves cellular heterogeneity and diversity.
• Achieved superior results compared to existing scRNA-seq clustering methods.

Conclusions:

• SinNLRR provides a powerful new tool for scRNA-seq data analysis, enhancing cell type detection.
• The similarity learning framework offers a promising direction for future computational biology research.
• The method's robustness and accuracy make it valuable for biological discovery using scRNA-seq data.