Abstract
Genetic risk for Alzheimer's Disease (AD) varies across populations. We hypothesized that three-dimensional (3D) genome architecture variations could offer novel epigenetic understanding of ancestry-specific genetic risk. Herein, we performed Hi-C analyses of frontal cortex from APOE ε4/ε4 individuals with African (AF) or European (EU) ancestry who also had single nuclei ATAC-seq and RNA-seq data available. Ancestry-specific 3D genome architecture was found at both compartment and chromatin loop levels. EU genomes have more active compartments than AF genomes, consistent with our previous report of higher chromatin accessibility in EU than AF genome. Of the over one million chromatin loops identified by the DeepLoop pipeline, we called 12,082 putative EU-specific and 2885 putative AF- specific loops. The AF-specific loops are smaller (median size =158 kb) and likely represent promoter-enhancer interactions, while EU-specific loops are larger (median size = 496 kb) and enriched for CTCF loops. We found that differently expressed genes between AF and EU ancestries were significantly enriched at the putative ancestry-specific loop loci (Fisher-test; p<2.2×10 (-16) ; OR=5.13). High confidence HiC-QTLs (N=38) were identified after filtering with CTCF consensus sequence and chromatin accessibility-QTLs. Our study demonstrates variations in 3D genome structure between ancestries, which may contribute to the ancestry-specific genetic risk.