Deep learning‐based identification of transcriptomic signatures associated with Alzheimer's disease progression. (December 2021)

• Record Type: Journal Article
• Authors: Wang, Qi; Readhead, Benjamin; Chen, Kewei; Su, Yi; Reiman, Eric M.; Dudley, Joel
• Is Part Of: Alzheimer's & dementia. Volume 17(2021)Supplement 3
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-12
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.051202
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Post‐mortem brain tissues and peripheral blood monocytes have been used to characterize some of the molecular networks and drivers associated with Alzheimer's disease (AD). In this study, we capitalize on the use of a deep learning approach to characterize the molecular changes associated with the severity of AD‐related neuropathology and cognitive impairment. Method: Starting from the RNA‐seq data obtained from postmortem brain tissues in the ROSMAP cohort (n = 634), we adopted a deep learning framework, which consists of joint supervised classification of the expression profiles from neuropathologically confirmed AD and control cases, followed by a mapping of the entire cohort with heterogenous diagnoses to the same transcriptomic space and applying unsupervised dimensionality reduction, and obtaining a progressive trajectory that mirrors AD pathological severity and cognitive impairment. We then applied the model to external transcriptomic datasets from different brain regions (MAYO, n = 266; MSBB, n = 214) and blood monocyte samples (ROSMAP, n = 551) to evaluate the consistency of these findings across multiple cohorts and tissues. Network analysis was also carried out to identify key gene modules present in the model underlying the progression. Result: The AD severity indexes (SI) calculated from the trajectory in ROSMAP brain samples were very closely correlated with all the AD neuropathology biomarkers (R ∼ 0.5, p < 1e‐11) and global cognitive function (R = ‐0.68, p < 2.2e‐16). In the external datasets from multiple brain regions (MAYO/MSBB), it also significantly (p < 1e‐3) predicted neuropathology and clinical severity, supporting the application of the model to multiple tissues and cohorts broadly. Network analysis of the "index genes" significantly contributed to the model resolved two discrete gene modules that are implicated in vascular and metabolic dysfunction respectively. In the ROSMAP monocyte samples, the trajectory derived for one of the gene modules also shows strong correlation (p < 1e‐3) with cognitive function. Conclusion: This study illustrates the promise of deep learning methods to multi‐omic data to characterize the molecular networks associated with increasingly severe clinical and neuropathological stages of neurodegenerative diseases like AD. It also offers novel approaches to the discovery of potential drug targets and/or biomarkers.