MT-HESS

MT-HESS performs Bayesian hierarchical multivariate analysis to identify associations between single nucleotide polymorphisms (SNPs) and gene expression levels across multiple tissues, cell types, or conditions.

Key Features:

• Bayesian Hierarchical Modeling: MT-HESS employs a Bayesian hierarchical model to jointly analyze SNPs and gene expression across multiple conditions.
• Multivariate Model Search: It performs a fully multivariate model search across all possible linear combinations of SNPs to detect joint genetic effects.
• Correlation Modeling: The method explicitly models correlations between condition- or tissue-specific responses to account for shared and distinct regulatory patterns.
• Hierarchical Structure: Hierarchical priors enable sharing of information across genes to improve detection of variants that regulate multiple genes.
• Enhanced Detection Power: Simulation studies demonstrate increased power, identification of new genetic hotspots, and improved prediction when analyzing multiple tissues jointly.

Scientific Applications:

• eQTL mapping: Identify expression quantitative trait loci by testing joint associations between SNPs and gene expression across tissues, cell types, or conditions.
• Hotspot discovery: Detect genetic hotspots and loci with cross-gene regulatory effects that may be missed by univariate approaches.
• Integrative genomics of complex traits: Analyze large-scale predictor-response associations across diverse biological conditions to study genetic regulation underlying complex traits and diseases.

Methodology:

Implements a Bayesian sparse regression algorithm that considers all linear combinations of SNPs in a multivariate context, incorporates a model for correlation between condition-specific responses, uses hierarchical priors to share information across genes, and is validated via simulation studies.