SV-integrated imputation to enable cross-cohort discovery of Alzheimer’s disease-associated structural variants

This project will test whether structural variants (SVs) contribute to Alzheimer’s disease risk across large genetic datasets by enabling SV imputation from genotyping array data. Our objective is to use an SV-integrated imputation framework and a whole-genome-sequencing-based reference panel to infer SV genotypes in cohorts that do not have direct SV calls, thereby increasing power for cross-cohort SV discovery and replication.
We will support local SV imputation in eligible Alzheimer’s disease cohorts using existing genotype array data and an SV-integrated reference panel. Participating cohorts will perform analyses under their own governance and data-access approvals, after which association summary statistics will be shared, harmonized and meta-analyzed across studies.
We will evaluate the association of imputed SVs with Alzheimer’s disease phenotypes, primarily case-control status, using cohort-appropriate regression models and standard covariate adjustment. We will perform quality control of imputation and association outputs, harmonize variant representation across studies, and meta-analyze SV association signals across participating datasets. We will assess replication, heterogeneity across cohorts and ancestries, and overlap with known Alzheimer’s disease GWAS loci to determine whether SVs may explain additional risk beyond single nucleotide variant(SNV)-focused analyses.
This project will involve collaboration with international Alzheimer’s disease genetics research groups and consortia that will conduct approved cohort-level analyses locally and contribute harmonized summary results for cross-study meta-analysis, enabling larger sample sizes, broader population diversity, and increased power to detect structural-variant associations.

Most genetic studies of Alzheimer’s disease focus on small DNA changes because those can be compared across many cohorts. Larger DNA changes, called structural variants, are harder to measure and are therefore missing from many studies. This project will use a new method to estimate structural variants from existing genotyping array data, using a reference panel built from whole-genome sequencing data.
By making structural variants measurable in many more participants, we aim to test these variants across multiple Alzheimer’s disease cohorts and combine results through meta-analysis. This could improve discovery of genetic risk factors, help determine whether known Alzheimer’s disease risk regions also contain important structural variants, and broaden genetic studies beyond standard single-variant analyses.