Genome-wide association studies (GWAS) predominantly rely on additive genetic models, which assume that each additional risk allele contributes linearly to phenotypic variation. While research has been conducted on non-additive effects, their reported contributions are generally minor compared to additive effects, leading most downstream analyses to remain focused on additive modeling. As a result, conventional additive-only approaches may provide an incomplete description of the genetic architecture underlying complex human traits.
This project aims to introduce and apply a factor-based representation of genetic variants that decomposes genotype information into interpretable axes capturing heterozygous and homozygous effects. This alternative parameterisation enables more direct biological interpretation of gene action and provides a richer understanding of how allele configurations contribute to phenotypic diversity.
Using UK Biobank genomic and phenotypic data, we will:
(1) perform GWAS analyses using factor-derived genotype encodings;
(2) compare statistical performance and interpretability with conventional additive models;
(3) quantify the relative contributions of heterozygous and homozygous components to heritability; and
(4) investigate biological pathways associated with these distinct components.
By reframing GWAS from an additive-versus-non-additive paradigm to a heterozygous-versus-homozygous perspective, this research aims to improve interpretability of genetic findings, enhance heritability estimation, and uncover biological mechanisms influencing complex disease risk. Ultimately, these outcomes may support precision medicine by revealing genetic effects that are overlooked in current additive-based analyses.