GW-SEM: A Statistical Package to Conduct Genome-Wide Structural Equation Modeling.

Area of Science:

• Genetics
• Psychiatry
• Statistical Genetics

Background:

• Advanced phenotyping improves genetic association studies.
• Multivariate methods like SEM are valuable for psychiatric and substance use phenotypes.
• Integrating SEM into genome-wide association studies (GWAS) is computationally challenging.

Purpose of the Study:

• To develop an efficient method for fitting SEMs in GWAS.
• To enable testing of complex hypotheses about genetic associations with multiple phenotypes or latent constructs.
• To overcome computational limitations of traditional SEM fitting in large-scale genetic analyses.

Main Methods:

• Developed a novel method using a diagonally weighted least squares (DWLS) estimator for four common SEMs.
• Applied the method to test SNP associations with multiple phenotypes or latent constructs on a genome-wide basis.
• Compared DWLS performance with full information maximum likelihood (FIML) and existing multivariate GWAS software.

Main Results:

• The DWLS estimator demonstrated strong correspondence with traditional FIML parameters and p-values.
• The new method efficiently fits SEMs, expanding the scope of testable models in GWAS.
• Simulations and power analyses confirmed the method's viability and efficiency.

Conclusions:

• The developed DWLS-based method offers an efficient approach to integrate SEMs into GWAS.
• This facilitates more complex genetic analyses in psychiatric and behavioral genetics.
• The method enhances the power to detect genetic associations and test sophisticated hypotheses.