Graph-regularized dual Lasso for robust eQTL mapping.

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Expression quantitative trait loci (eQTL) mapping is crucial for understanding complex traits.
• Integrating network interactions among genetic markers and genes is a key challenge.
• Existing methods often require preprocessing, ignore network incompleteness/noise, and fail to integrate diverse data.

Purpose of the Study:

• To develop a robust method for eQTL mapping that effectively integrates network information.
• To address limitations of existing Lasso-based eQTL mapping approaches.
• To incorporate genetic marker location and gene-pathway information.

Main Methods:

• Proposed Graph-regularized Dual Lasso (GDL) approach.
• Simultaneously integrates correlation structures among genetic markers and traits.
• Utilizes graph-based regularizers, robust to network incompleteness and noise, avoiding explicit clustering.
• Generalizes GDL to incorporate marker location and pathway data.

Main Results:

• GDL effectively integrates prior knowledge from networks, marker location, and pathways.
• The method demonstrates robustness to incomplete and noisy network data.
• Extensive evaluations show GDL significantly outperforms state-of-the-art eQTL mapping methods on simulated and real datasets.

Conclusions:

• GDL offers a superior approach for eQTL mapping by leveraging network and other biological information.
• The method enhances the dissection of the genetic basis of complex traits.
• GDL provides a flexible and powerful framework for integrating diverse data in genetic analyses.