GGD-Lasso

GGD-Lasso implements a graph-regularized dual Lasso approach to map expression quantitative trait loci (eQTL) by integrating genetic markers, trait correlation structures, and prior network information to improve robustness against incomplete or noisy networks.

Key Features:

• Graph-Based Regularization: Employs graph-based regularizers to model prior networks and integrate correlation structures among genetic markers and traits.
• Robustness to Incomplete and Noisy Networks: Refines partial or noisy prior networks to maintain reliable eQTL signal recovery despite network imperfections.
• Integration of Additional Biological Information: Incorporates spatial locations of genetic markers and gene–pathway information alongside marker–trait correlations.
• No Preprocessing Clustering Required: Eliminates the need for explicit preprocessing clustering steps by directly leveraging network regularization.
• Empirical Evaluation: Demonstrated performance improvements on simulated and real datasets compared to existing Lasso-based eQTL mapping methods.

Scientific Applications:

• eQTL Mapping: Identification of expression quantitative trait loci linking genetic markers to gene expression variation.
• Genetic Architecture of Complex Traits: Dissection of marker–trait relationships to study heritability and genetic influences on complex traits.
• Integrative Genomics: Integration of marker locations, pathway annotations, and network priors for multi-layer genomic analyses relevant to personalized medicine and functional genomics.

Methodology:

Integrates correlation structures among genetic markers and traits using graph-based regularization within a dual Lasso framework, refines incomplete networks by leveraging available prior knowledge, and supports incorporation of marker location and pathway information.