QUBIC2: a novel and robust biclustering algorithm for analyses and interpretation of large-scale RNA-Seq data

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Biclustering large-scale gene expression data can identify functional gene modules.
• Existing methods struggle with RNA-Sequencing (RNA-Seq) and single-cell RNA-Sequencing (scRNA-Seq) data due to high zero counts.

Purpose of the Study:

• To develop a novel biclustering algorithm, QUBIC2, for comprehensive and accurate detection of significant biclusters.
• To address the challenges posed by zero-enriched expression data in RNA-Seq and scRNA-Seq.

Main Methods:

• QUBIC2 utilizes a novel left-truncated mixture of Gaussian model for multimodality assessment.
• A dropout-saving expansion strategy optimizes gene modules using information divergency.
• A rigorous statistical test validates the significance of all identified biclusters.

Main Results:

• QUBIC2 demonstrated superior bicluster detection performance compared to five existing algorithms on benchmark datasets.
• The algorithm showed robust and improved performance across microarray, bulk RNA-Seq, and scRNA-Seq data.
• QUBIC2 effectively handles zero-enriched expression data, a common issue in scRNA-Seq.

Conclusions:

• QUBIC2 is a powerful and versatile biclustering algorithm suitable for various gene expression data types.
• The algorithm offers improved accuracy and comprehensiveness in identifying functional gene modules.
• QUBIC2 provides a valuable tool for genomic and transcriptomic data analysis.