Risk score modeling of multiple gene to gene interactions using aggregated-multifactor dimensionality reduction

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Multifactor Dimensionality Reduction (MDR) is used to detect gene-gene (GxG) interactions in complex diseases.
• Current MDR methods use a single GxG interaction for a binary risk model, neglecting cumulative effects.

Purpose of the Study:

• To introduce Aggregated-Multifactor Dimensionality Reduction (A-MDR) for detecting multiple GxG interactions.
• To develop a higher-resolution disease susceptibility risk score by aggregating effects from numerous GxG interactions.

Main Methods:

• Exhaustively search for and detect significant GxG interactions to build an epistasis-enriched gene network.
• Develop an aggregated epistasis-enriched risk score considering multiple GxG interactions simultaneously.
• Introduce new statistical measures (pOR, pRR, pChi) for detecting multiple GxG interactions.

Main Results:

• A-MDR was evaluated through extensive simulations.
• Application to Juvenile Idiopathic Arthritis (JIA) data identified GxG interactions in the folate pathway affecting methotrexate (MTX) response.
• The aggregated epistasis-enriched risk score, using 82 GxG interactions, achieved 82% accuracy in distinguishing MTX responders from non-responders.

Conclusions:

• A-MDR offers an innovative approach within the MDR framework to analyze aggregated GxG interaction effects.
• The method enhances the understanding of complex disease susceptibility and treatment response by incorporating multiple genetic interactions.