An integrated phenomic approach to multivariate allelic association

Area of Science:

• Genetics
• Statistical Genetics
• Biostatistics

Background:

• Genome-wide association studies (GWAS) primarily use association to identify genetic variants linked to phenotypic variation.
• Focus has been on multiple testing issues with single nucleotide polymorphisms (SNPs), overlooking inflated error rates from testing numerous phenotypes.
• Multivariate analyses can detect pleiotropic and monotropic effects while accounting for phenotype non-independence.

Purpose of the Study:

• To present a novel maximum likelihood approach for genetic association analysis that integrates latent and variable-specific tests.
• To provide a method applicable to both individual and family-based genetic data.
• To evaluate the performance of this combined multivariate (CMV) approach against traditional univariate methods.

Main Methods:

• Developed a maximum likelihood statistical framework combining latent factor and variable-specific tests.
• Applied the combined multivariate (CMV) analysis to simulated data and real-world genetic data (Add Health Study).
• Compared the power and performance of CMV analysis with univariate analyses of factor scores and sum scores.

Main Results:

• The CMV analysis demonstrated robust performance with minimal power loss compared to univariate analyses when factor-level association was present.
• CMV analysis maintained statistical power even as the allelic effects deviated from factor loadings, unlike univariate methods.
• The approach successfully examined the association between dopamine receptor D2 TaqIA and substance initiation in the Add Health dataset.

Conclusions:

• The combined multivariate (CMV) approach offers a powerful and flexible method for genetic association studies involving multiple phenotypes.
• This method effectively controls for non-independence between phenotypes and maintains power in complex genetic scenarios.
• The CMV approach provides a valuable tool for dissecting genetic influences on complex traits, with practical implementation resources available.