High-throughput and efficient multilocus genome-wide association study on longitudinal outcomes.

Area of Science:

• Genetics
• Bioinformatics
• Machine Learning

Background:

• High-dimensional genomic data, including complex longitudinal phenotypes, are increasingly common in medical research.
• Existing analytical approaches often underutilize longitudinal trait data in genetic analyses like genome-wide association studies (GWAS).
• There is a need for efficient methods to perform multilocus GWAS on longitudinal data.

Purpose of the Study:

• To develop a high-throughput machine learning approach for multilocus GWAS using longitudinal traits.
• To couple Empirical Bayesian Estimates from mixed-effects modeling with a novel ℓ0-norm algorithm.
• To enable scalable and efficient analysis of complex genetic data.

Main Methods:

• Developed a novel machine learning approach combining Empirical Bayesian Estimates from mixed-effects modeling and a ℓ0-norm algorithm.
• Designed for high-throughput multilocus genome-wide association studies (GWAS) with longitudinal traits.
• Validated through extensive simulations and application to real-world data.

Main Results:

• The proposed approach demonstrated accurate selection of single nucleotide polymorphisms (SNPs) with high power and robust false positive control.
• Achieved significant speed improvements (over 1000x faster) compared to existing methods, enabling genome-wide multilocus analysis of longitudinal traits.
• Successfully identified known SNPs associated with Alzheimer's Disease (AD) in the AD Neuroimaging Initiative dataset.

Conclusions:

• The novel approach offers a scalable and computationally efficient solution for analyzing high-dimensional genomic data with longitudinal phenotypes.
• Facilitates true multilocus analysis, potentially identifying complex genetic architectures of diseases.
• Provides a valuable tool for genetic research, particularly in complex diseases with longitudinal data.