DriverGenePathway: Identifying driver genes and driver pathways in cancer based on MutSigCV and statistical methods

Area of Science:

• Computational biology
• Cancer genomics
• Bioinformatics

Background:

• Existing computational methods for identifying cancer driver genes lack consistency and stability across studies.
• There is a need for user-friendly tools with improved operability and system compatibility for driver gene identification.

Purpose of the Study:

• To develop a user-friendly R package, DriverGenePathway, for identifying cancer driver genes and pathways.
• To integrate established methods like MutSigCV with novel statistical approaches for enhanced accuracy.

Main Methods:

• Developed the DriverGenePathway R package, incorporating MutSigCV's theoretical basis and information entropy for mutation category discovery.
• Employed five hypothesis testing methods (beta-binomial, Fisher, likelihood ratio, convolution, projection) to identify core driver genes.
• Introduced de novo methods to address mutational heterogeneity for driver pathway identification.

Main Results:

• Demonstrated DriverGenePathway's performance on eight TCGA cancer types.
• Successfully confirmed known driver genes, showing high overlap with the Cancer Gene Census list.
• Identified significant driver pathways crucial for cancer development.

Conclusions:

• DriverGenePathway provides a robust and user-friendly platform for cancer driver gene and pathway identification.
• The package enhances consistency and stability in driver gene discovery, aiding cancer research.
• DriverGenePathway is freely available on GitHub, promoting accessibility for the research community.