Limix

Limix implements multi-trait linear mixed models to analyze genetic data by modeling fixed and random effects, observed and hidden covariates, and enhancing power in GWAS and phenotype prediction.

Key Features:

• Multi-trait mixed models: Handles multiple correlated traits simultaneously to enable joint analyses across phenotypes.
• Accounting for covariates: Models both observed and hidden covariates to control confounding in genetic studies.
• Fixed and random effects: Uses linear mixed models to explicitly model fixed effects and random effects structure in the data.
• Stepwise multi-locus regression: Integrates stepwise multi-locus regression within the multi-trait framework to detect multiple associated loci.
• Efficiency and scalability: Optimizes computations for handling large-scale genetic datasets and complex models.
• Improved GWAS power and phenotype prediction: Enhances power in genome-wide association studies and increases accuracy of phenotype prediction through joint and multi-locus modeling.
• Applicability to diverse genetic contexts: Demonstrated for joint GWAS of correlated blood lipid phenotypes, analysis of expression across transcript isoforms, and pathway-based modeling of molecular traits across different environments.

Scientific Applications:

• Joint GWAS of correlated phenotypes: Enables joint genome-wide association analyses for correlated traits such as blood lipid phenotypes to increase detection power.
• Transcript isoform expression analysis: Models expression levels across multiple transcript isoforms of genes to analyze isoform-specific genetic effects.
• Pathway-based molecular trait modeling: Supports pathway-level modeling of molecular traits across different environments to study context-dependent effects.

Methodology:

Employs linear mixed models that model fixed and random effects, incorporates multi-trait mixed-model frameworks, and applies stepwise multi-locus regression while accounting for observed and hidden covariates.