Low Rank Subspace Clustering via Discrete Constraint and Hypergraph Regularization for Tumor Molecular Pattern...

Area of Science:

• Computational biology
• Bioinformatics
• Machine learning for cancer research

Background:

• Tumor clustering is vital for cancer classification and treatment.
• Traditional methods like NMF-based models have limitations in accuracy.
• Subspace clustering, particularly low-rank representation, offers improved performance.

Purpose of the Study:

• To propose a novel Low Rank Subspace Clustering model (DHLRS).
• To enhance cancer class discovery using discrete constraints and hypergraph regularization.
• To identify molecular patterns in tumor gene expression data.

Main Methods:

• Developed a novel Low Rank Subspace Clustering model (DHLRS).
• Employed discrete constraints for direct learning of cluster indicators.
• Utilized Schatten-norm for low-rank approximation and hypergraph regularization for complex gene relationships.

Main Results:

• The DHLRS model demonstrated effectiveness in clustering tasks.
• Successfully identified molecular patterns in tumor gene expression datasets.
• Experiments on synthetic and real data validated the proposed method's performance.

Conclusions:

• DHLRS provides an effective approach for tumor clustering and cancer class discovery.
• The integration of discrete constraints and hypergraph regularization improves subspace clustering.
• The model aids in understanding the intrinsic geometrical structure of gene expression data.