Network-guided regression for detecting associations between DNA methylation and gene expression.

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• High-throughput biological data offer insights into cellular states but require advanced statistical methods.
• Aberrant DNA methylation is a cancer hallmark, yet its link to gene expression is unclear.

Purpose of the Study:

• To introduce Network-sparse Reduced-Rank Regression (NsRRR), a novel statistical framework.
• To leverage gene interaction networks for identifying associations between DNA methylation and gene expression.

Main Methods:

• Developed a multivariate regression framework, NsRRR.
• Utilized prior biological knowledge in the form of gene interaction networks.
• Validated the model using simulations and The Cancer Genome Atlas (TCGA) ovarian cancer data.

Main Results:

• NsRRR demonstrates superior variable selection accuracy compared to models without network information.
• The method effectively identifies associations between DNA methylation and gene expression signatures.
• Application to TCGA ovarian cancer data showcases practical utility.

Conclusions:

• NsRRR provides a powerful approach for analyzing complex biological data.
• Integrating network information enhances the discovery of gene expression-DNA methylation relationships.
• This framework can advance understanding and treatment of diseases like cancer.