SCCLRR: A Robust Computational Method for Accurate Clustering Single Cell RNA-Seq Data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) enables the study of cellular heterogeneity.
• Identifying cell subpopulations is crucial but challenging due to data complexity (high dimensionality, sparsity, noise).
• Existing computational methods for cell type detection often yield unsatisfactory performance.

Purpose of the Study:

• To develop a novel computational method for robust and accurate cell type detection from scRNA-seq data.
• To improve the identification of cellular subpopulations by learning an intrinsic similarity matrix.
• To address the limitations of current methods in handling scRNA-seq data characteristics.

Main Methods:

• Proposed a new method named SCCLRR (Single-Cell Clustering by Low-Rank Representation).
• Utilized a low-rank representation (LRR) framework to capture global and local data properties.
• Integrated normalized Euclidean distance and cosine similarity for local regularization.
• Employed the alternating direction method of multipliers (ADMM) algorithm to solve the non-convex optimization model.

Main Results:

• Evaluated SCCLRR on nine real-world scRNA-seq datasets.
• Compared SCCLRR performance against seven state-of-the-art methods.
• Demonstrated that SCCLRR significantly outperforms existing methods in clustering scRNA-seq data.
• Showcased the robustness and effectiveness of SCCLRR for cell type detection.

Conclusions:

• SCCLRR is a robust and effective method for clustering scRNA-seq data.
• The proposed approach successfully identifies cell subpopulations by learning an accurate similarity matrix.
• SCCLRR offers an improved solution for analyzing cellular heterogeneity in transcriptomic studies.