Detecting Protective Rare Variants for Alzheimer’s Disease from Whole‐Genome Sequencing in the Amish

Abstract

AbstractBackgroundA shift in focus from risk to resilience for Alzheimer’s disease (AD) encourages efforts to uncover novel AD biological mechanisms. Rare variants identified through whole genome sequencing (WGS) may represent an important and understudied component of complex trait genetics. While population‐based studies are powered to discover associations with common genetic variants, founder populations are better powered for discovery of previously unknown rare alleles that have risen to higher frequency due to genetic drift.MethodWe examined WGS data from the Mid‐Western Amish population in a genome‐wide search for rare coding variants shared only among cognitively‐unimpaired (CU) individuals. The cognitive status of each individual was assigned via consensus review of medical history and neuropsychological testing. Allele frequencies were calculated across all samples and for CU and cognitively‐impaired (CI) groups separately. We defined rare variants as having an allele frequency <0.05 across all samples. Variants with a minor allele count (MAC)>10 in the CU group, and 0 in the CI group were annotated to determine likely loss of function.ResultAfter extensive QC, 1,048 samples were available to estimate overall allele frequency. Allele frequencies of rare variants (n = 11,854,817) within 634 CU (mean age = 81.53±6.10, 60% female) and within 184 CI (mean age = 84.77±6.60, 62% female) individuals were compared and 51,616 variants with MAC >10 were found only in CU. There were 316 unique coding variants (288 missense, 4 inframe_insertion, 6 inframe_deletion, 11 frameshift, 4 stop_gained, 3 splice_donor) and 180 synonymous variants located within 425 unique genes. Among them, 7 missense variants and 7 synonymous variants had MAC >20 and the remaining variants had MAC of 11‐20. The mean allele frequency of the 7 missense variants was 0.0038 in TOPMed compared to 0.0154 in our data. Two synonymous variants were located within proposed AD genes (NYAP1 and ZNF423). Additionally, several untranslated region or nonsense‐mediated decay transcript variants existed within 26 known AD gene regions, among which 8 of them are reported to harbor protective variants for AD.ConclusionNumerous rare variants potentially impacting gene function were found only in CU individuals, providing a rich resource for further investigation of genes that may provide protection from cognitive impairment.