Structural variation detection and association analysis of whole-genome-sequence data from 16,543 Alzheimer's disease sequencing project subjects

Abstract

Introduction: The role of structural variations (SVs) in Alzheimer's disease (AD) remains understudied.

Methods: We analyzed whole-genome sequencing data from the Alzheimer's Disease Sequencing Project (N = 16,543) and identified 400,234 (168,223 high-quality) SVs. Laboratory validation yielded a sensitivity of 82% (85% for high-quality).

Results: We found a burden of singletons (odds ratio [OR] = 1.07, p = 0.0017) and homozygous deletions (OR = 1.14, p < 0.0001) in cases. On AD genes, we observed the ultra-rare SVs associated with the disease, including protein-altering SVs in ABCA7, APP, PLCG2, and SORL1. Twenty-one SVs are in linkage disequilibrium (LD) with known AD-risk variants, exemplified by a 5k deletion in LD (R² = 0.99) with rs143080277 in NCK2. We identified a rare deletion near RNA5SP293 associated with AD (OR = 1.99, p = 1.3 × 10^-5), which was replicated using an independent dataset.

Discussion: This study highlights the pivotal role of SVs in AD genetics.

Highlights: Observed a significant burden of singletons and homozygous deletions in Alzheimer's disease (AD) patients. Identified rare protein-altering structural variations (SVs) in ABCA7, APP, PLCG2, and SORL1. Established linkages between SVs and AD risk-associated single nucleotide variants (SNVs). Discovered a novel deletion near RNA5SP293 linked to AD, replicated independently. Uncovered over-representation of SVs in neuronal function pathways.

Keywords: Alzheimer's disease; NCK2; RNA5SP293; copy number variation; structural variation.

FIGURE 1

Characteristics of high‐quality SVs. (A) Number of high‐quality SVs per individual by ancestry. (B) Principal component analysis of high‐quality SV with MAF > 0.01 and HWE > 1e‐5. (C) The cumulative fractions of variants by AF. (D) The size distribution of high‐quality SVs. (E) The size distribution of high‐quality SVs by SV type. AF, allele frequency; HWE, Hardy–Weinberg equilibrium; MAF, minor allele frequency; SVs, structural variants.

FIGURE 2

Functional annotation of SVs. (A) AnnotSV ranking scores of common (AF ≥ 0.01) and rare (AF < 0.01) high‐quality SVs. The rare SVs are more likely to be deleterious with higher AnnotSV ranking scores. (B) VEP annotation of common (AF ≥ 0.01) and rare (AF < 0.01) high‐quality SVs. The protein‐altering SVs tend to be rare. (C) Percent of singletons in a specified functional category by VEP. The coding sequence variants include stop gain, start/stop lost, frameshift, inframe deletion/insertion, and missense mutations. SVs that were annotated (by VEP) to be able to cause transcript ablation/amplification, stop gain, start/stop lost, frameshift, inframe deletion/insertion, missense mutation, and affecting splice acceptor/donor were classified as protein‐altering variants. AF, allele frequency; SVs, structural variants; VEP, Variant Effect Predictor.

FIGURE 3

Ultra‐rare deletion and duplication on SORL1. (A) Deletion and duplication of SORL1. (B) The 192 Kb duplication covers part of SORL1 and SC5D. (C) The 8.45 Kb deletion covers exon 6 of SORL1.

FIGURE 4

Association of SVs with AD and enrichment analysis. (A) Association of SVs with AD. The red line represents an FDR of 0.05. The gray line represents an FDR of 0.2. One SV near RNA5SP293 and one SV on RABGAP1 with a FDR < 0.2 and replication p < 0.1 were marked red. (B) Controls carrying the duplication on RABGAP1 (chr9:122989084‐122989182) showed significantly lower cerebrospinal fluid amyloid beta levels. AD patients carrying the duplication on RABGAP1 (chr9:122989084‐122989182) showed significantly lower memory scores. p‐values were obtained by the Wilcoxon rank‐sum test. (C) Enrichment analysis for high‐quality SVs (nominal p < 0.05) that are not in problematic regions. AD, Alzheimer's disease; BP, biological process; CC, cellular component; FDR, false discovery rate; MF, molecular function; KEGG, Kyoto Encyclopedia of Genes and Genomes; SVs, structural variants.