ADRD Epistasis

We will perform a series of genetic association analyses of AD & ADRD. We will evaluate epistasis associations of all pairs of genes at the transcription, proteomic, and rare variant burden score levels. We will follow-up any gene-level analyses with SNP-level analyses for targeted regions of interest. We will identify gene-gene interactions using imputed expression that influence ADRD risk. Analyses will be performed by first imputing expression in brain regions of key importance to ADRD risk – the frontal cortex, hippocampus, and amygdala, and possibly other brain regions as expression reference panels grow. We will follow-up with cell type-specific imputed expression and imputed proteomic expression and high-impact missense and loss-of-function rare variant burden scores. We will also perform all APOE e4-by-variant, -gene, and -transcript interactions. All analyses will use a mixed model association framework, accounting for relatedness and stratification, using appropriate covariates. We will perform analyses within genetic ancestry groups, then perform trans-ancestry meta-analysis using a random-effects meta-analysis. We will also perform sex-stratified analyses to explore sex-specific genetic effects.
We will identify network hub genes as those with the most pairwise interactions. Simulations have shown that while statistical power is limited for polygenic traits like ADRD, we are well-powered (>90% power) to correctly identify network hub genes as those with >=10 interactions at nominal significance (1e-5).
We will perform enrichment tests of associations using a priori defined gene sets, including >8,000 gene sets (from MSigDB), protein-protein interaction network modules (StringDB), and gene sets representing brain region- and cell type-specific expression. This will identifying key cell types, pathways and networks that underlie ADRD risk via interactions.

Gene-gene interactions in ADRD risk have never been systematically and exhaustively explored across the entire genome, but we have developed a new method to exhaustively and efficiently test all pairwise gene-gene interaction associations with complex traits. This has identified novel pathways and core genes integrating multiple pathways in other complex traits. We propose an exhaustive, genome-wide gene-gene interaction study of ADRD, in which we will uncover new aspects of ADRD biology. We will perform analyses at the variant, gene, and gene expression levels to assess multiple layers of ‘omics data, and determine which molecular pathways, brain regions, and individual cell types are enriched in gene-gene interactions contributing to ADRD risk.