Robust Graph Regularized NMF with Dissimilarity and Similarity Constraints for ScRNA-seq Data Clustering

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (ScRNA-seq) enables detailed cell heterogeneity discovery.
• Clustering ScRNA-seq data is crucial but challenging due to high dimensionality and noise.

Purpose of the Study:

• To develop a robust model for ScRNA-seq data clustering.
• To improve clustering performance by addressing data noise and dimensionality.

Main Methods:

• Proposed a novel Robust Graph regularized Non-Negative Matrix Factorization with Dissimilarity and Similarity constraints (RGNMF-DS) model.
• Utilized complementary similarity and dissimilarity regularizers for matrix decomposition.
• Incorporated a graph regularizer to capture local geometric structure.
• Employed the l2,1-norm to enhance robustness against noise.

Main Results:

• The RGNMF-DS model demonstrated superior performance in clustering ScRNA-seq datasets.
• Experimental results showed outperformance compared to state-of-the-art clustering methods.
• The model effectively handles noise and captures underlying data structures.

Conclusions:

• RGNMF-DS offers a robust and effective approach for ScRNA-seq data clustering.
• The proposed method improves upon existing techniques for analyzing single-cell data.
• This advancement aids in more accurate discovery of cell diversity.