A component overlapping attribute clustering (COAC) algorithm for single-cell RNA sequencing data analysis and...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Next-generation sequencing and computational tools facilitate large-scale single-cell RNA sequencing (scRNA-seq) data analysis.
• scRNA-seq data present challenges like low gene expression and missing data, complicating the identification of functional gene sets and regulatory networks.
• Linking cell types to diseases and therapeutics via scRNA-seq profiles remains a complex task.

Purpose of the Study:

• To develop a novel algorithm, Component Overlapping Attribute Clustering (COAC), for localized gene co-expression network analysis in scRNA-seq data.
• To demonstrate COAC's utility in cell type identification, disease correlation, and pharmacogenomics biomarker discovery.

Main Methods:

• Developed the Component Overlapping Attribute Clustering (COAC) algorithm for localized gene co-expression network analysis.
• Inferred gene subnetworks from decomposed scRNA-seq data matrices.
• Applied COAC to scRNA-seq data for cell type identification, survival analysis, and drug response prediction.

Main Results:

• COAC identified single-cell gene subnetworks with 83% accuracy for cell type identification.
• COAC-inferred subnetworks from melanoma patients' scRNA-seq data correlated significantly with survival rates (TCGA).
• COAC identified localized gene subnetworks as pharmacogenomics biomarkers, predicting drug responses in cancer cell lines (AUC 0.728-0.783) (GDSC).

Conclusions:

• COAC provides a robust method for identifying network-based diagnostic and pharmacogenomics biomarkers from large-scale scRNA-seq data.
• The algorithm facilitates a deeper understanding of cell-specific gene regulation in health and disease.
• COAC is a valuable tool for advancing precision medicine through scRNA-seq data analysis.