FEATURE SELECTION FOR GENERALIZED VARYING COEFFICIENT MIXED-EFFECT MODELS WITH APPLICATION TO OBESITY GWAS.

Area of Science:

• Genetics
• Biostatistics
• Genomics

Background:

• Obesity is a complex trait influenced by genetic and environmental factors.
• Genome-wide association studies (GWAS) have identified numerous genetic loci associated with body mass index (BMI).
• Existing statistical models may not fully capture the intricate genetic influences on BMI over time and across individuals.

Purpose of the Study:

• To develop a robust two-step gene-detection procedure for identifying significant single nucleotide polymorphisms (SNPs) associated with BMI.
• To analyze genetic impacts on both the mean BMI trend and its age-dependent variability.
• To account for individual genetic variations in a generalized varying coefficient mixed-effects model framework.

Main Methods:

• Utilized data from a genome-wide association study (GWAS) on obesity.
• Proposed a two-step procedure for generalized varying coefficient mixed-effects models with ultrahigh dimensional covariates.
• Employed Monte Carlo simulations to validate the method's performance.
• Conducted causal inference for selected SNPs.

Main Results:

• Successfully identified significant SNPs impacting the mean BMI trend, including known "fat genes."
• Discovered SNPs that significantly influence the age-dependent variability of BMI.
• Demonstrated the procedure's applicability to longitudinal data with various response types (continuous, binary, count).

Conclusions:

• The proposed method effectively detects SNPs influencing BMI, offering insights into genetic contributions to obesity.
• The procedure accounts for individual genetic variations and is adaptable to diverse longitudinal data.
• This work advances genetic analysis for complex traits like obesity and supports potential causal inference.