Premium
Assessing whole genome sequencing variation for Alzheimer’s disease in 4707 individuals from the Alzheimer’s Disease Sequencing Project (ADSP)
Author(s) -
Peloso Gina M.,
Wang Yanbing,
Lin Honghuang,
Sarnowski Chloé,
Pitsillides Achilleas N.,
Lim Elise M.,
Beecham Gary W.,
HamiltonNelson Kara L.,
Ramos Jairo,
Martin Eden R.,
Naj Adam C.,
Thornton Timothy A.,
Wang LiSan,
Boerwinkle Eric,
Farrer Lindsay A.,
Haines Jonathan L.,
Mayeux Richard,
PericakVance Margaret A.,
Seshadri Sudha,
Schellenberg Gerard D.,
Wijsman Ellen,
Fornage Myriam,
Dupuis Josée,
Destefano Anita L.,
Kunkle Brian W.
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.045548
Subject(s) - genetics , dna sequencing , genome wide association study , 1000 genomes project , biology , locus (genetics) , genetic association , allele , genotype , whole genome sequencing , gene , genome , single nucleotide polymorphism
Background The Alzheimer’s Disease Sequencing Project (ADSP) seeks to identify genomic variants contributing to increased risk of and/or protection from Alzheimer’s disease (AD) in multi‐ethnic populations. Here we report on statistical modeling considerations and results from 4707 diverse individuals including 2076 non‐Hispanic White, 1030 Black, 1304 Hispanic, and 297 of other ancestry with whole genome sequencing (WGS) in the ADSP. Methods A total of 4789 individuals were sequenced across 4 sequencing centers and 2 sequencing platforms. Joint calling was carried out by the ADSP data coordinating center GCAD. We evaluated a series of models to explore technical covariate adjustment related to study, sequencing center, and platform and then performed single variant association analysis for AD status adjusting for principal components associated with AD, technical covariates, and a genetic relatedness matrix, limiting to variants with a minor allele count > 30 and a missingness rate ≤ 10%. We also performed gene‐based association of rare (both MAF < 1% and MAF < 5%) coding variants. Results After quality control, we included over 95 million variants across 4707 individuals (2209 cases, 2498 controls). We replicated association at the APOE locus and identified a locus on chromosome 15 that has a suggestive association with AD status (freq = 0.8%, p‐value = 1 × 10 −7 ). Despite joint genotype calling, adjustment for study, sequencing center and platform are necessary to control type‐I error. In gene‐based tests, we found evidence (p < 0.05) of association for 4 of 8 AD candidate genes with low‐frequency or rare variant associations implicated for AD ( TREM2, ABI3, SORL1, and MAPT ) as well as suggestive associations (P < 5 × 10 −5 ) for seven genes ( ZGRF1 on chr4; C1RL on chr12; NAA30 on chr14; NIPA2 on chr15; and ABCC3, BIRC5 and HIC1 on chr17). Conclusion We explored covariate models for the WGS data from ADSP and established that despite joint calling, it is necessary to account for technical effects in the model. We found suggestive novel associations with AD status that will require confirmation.