There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
| Principal Investigator | |
| Principal Investigator's Name: | Carolina Dallett |
| Institution: | Roche |
| Department: | Computational Biology |
| Country: | |
| Proposed Analysis: | Feasibility of Alzheimer's Disease (AD) multi-ethnicity gene risk score Deepa Rajami, Carolina Dallett, Sowmi Utiramerur, Clara Su, Darvesh Gorhe, Juhi Pandey Background: Late-onset Alzheimer’s Disease is non-trivial to predict because the disease is polygenic and multi-mechanistic with high heritability (~60-80%) and few universally predictive biomarkers. Associations obtained in one molecular feature modality (eg. Gene expression (2)) do not always correlate with another (eg. SNPs(4)). In the example to the right, manually curated multimodal evidence (red) from Bellenguez et al.*(4), shows minimal overlap with AD related genes from other modalities: methylation (6), siRNA (7), expression (2). Performance is affected when using traditional GWAS-based PRS scores across cohorts or ethnicities due to the method’s inability to differentiate causal and bystander associations and other limitations (eg. Disease being multi-mechanistic). A mechanistic approach to gene identification for PRS scores can learn hidden links from multimodal data to differentiate the causal changes that lead to the disease pathophysiology. Goal: Refine PRS scores to extend across ethnicities. Goal1: Gene prioritization and score generation based on mechanistic understanding of AD pathophysiology and compare to standard PRS calculations, Goal2: Development of classifier based on Gene prioritization score that is applicable to uni-modal patient data Collate and develop an additive PRS from multiple strong prioritized PRS signals from different cohorts to fill in gaps in our mechanistic understanding of the disease. This helps fill in the gaps in our knowledge of unidentified paths and triggers. Use this refined PRS to investigate predictability across ethnicities. Pending Data request: The research team has accumulated significant open source data of orthogonal experiments such as single-cell RNS-seq, bulk-RNA-seq (case-control), siRNA, methyl-seq, etc. For model evaluation, the team needs individual level data and weights’ estimation ( request for study data of the Alzheimer’s Disease Sequencing project from ethnically diverse populations available through NIAGADS - https://dss.niagads.org/datasets/ng00067/) Outcome: a)Validation study replicating the previously reported ability to run conventional PRS methods b)evaluate cross-ethnicity robustness using predicted weights from patients’ variants and phenotypes for PRS calculation using IMC and GNN |
| Additional Investigators |
