LTSOFT

LTSOFT implements a conditioning approach using liability threshold modeling to integrate clinical covariates (age, BMI, smoking status, gender) and genetic covariates (known associated variants) into case-control association studies to improve statistical power while accounting for disease prevalence and non-random ascertainment.

Key Features:

• Integration of Clinical and Genetic Covariates: Incorporates clinical covariates (age, BMI, smoking status, gender) alongside genetic covariates (known associated variants) in case-control analyses.
• Novel Conditioning Approach: Implements a conditioning strategy based on liability threshold modeling that estimates model parameters for each known variant.
• Accounting for Disease Prevalence and Non-Random Ascertainment: Incorporates published disease prevalence data and models non-random ascertainment to mitigate correlation structures between candidate and known variants, including across chromosomes.
• Improved Statistical Performance: Demonstrates controlled false-positive rate and enhanced test statistics versus no conditioning and standard conditioning methods in simulation studies and empirical datasets, especially for low-prevalence diseases.
• Enhanced Disease Gene Discovery: Increases power to detect novel disease-associated variants in studies with multiple known risk variants.

Scientific Applications:

• Case-control association studies: Analysis of genetic associations in case-control datasets that include clinical covariates and known associated variants.
• Low-prevalence disease studies: Application in diseases with low prevalence where standard logistic regression approaches lose power.
• Disease gene discovery in complex disorders: Facilitates discovery of novel genetic associations in disorders characterized by numerous known risk variants.

Methodology:

LTSOFT applies a conditioning approach based on liability threshold modeling, estimates parameters for known variants, incorporates published disease prevalence data, and models correlation structures induced by non-random ascertainment as an alternative to standard logistic regression.