MOST

MOST performs multivariate association testing of multiple correlated traits to improve detection of genetic variants influencing complex human diseases.

Key Features:

• Multivariate Analysis: Employs multivariate test statistics to analyze multiple correlated traits simultaneously as a complement to univariate approaches.
• Trait Flexibility: Handles continuous and binary traits, and mixtures of both.
• Family Data Accommodation: Supports studies that include family data.
• Covariate Adjustment: Adjusts for covariates such as ancestry variables, age, and gender.
• Generalized Linear Models (GLMs): Relates marginal distributions of multivariate traits to genetic variants and covariates via GLMs without modeling dependencies among traits or family members.
• Score-Type Statistics: Constructs score-type statistics that are computationally efficient and numerically stable in the presence of covariates and can be combined across studies with different designs and arbitrary patterns of missing data.
• Linkage Disequilibrium (LD) Consideration: Implements a strategy to determine genome-wide significance while accounting for linkage disequilibrium among genetic variants.

Scientific Applications:

• Meta-analysis of cardiovascular cohorts: Applied to five major cardiovascular cohort studies, identifying a new pleiotropic locus (HSCB) associated with four analyzed traits.
• Cross-study integration and power enhancement: Integrates data from multiple studies and accommodates various study designs to increase power for detecting genetic associations.

Methodology:

MOST uses multivariate test statistics and GLMs relating marginal trait distributions to genetic variants and covariates without modeling trait or family dependencies, constructs score-type statistics that are combinable across studies with arbitrary missing data, and applies a strategy for genome-wide significance that accounts for linkage disequilibrium among variants.