Browsing by Subject "Polygenic Risk Scores (PRS)"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Polytranscriptomic risk score for Alzheimer Disease in a large diverse multi‐center brain bank study
    (Wiley, 2025-01-03) Cieza, Basilio; Yang, Zikun; Reyes-Dumeyer, Dolly; Lee, Annie J.; Dugger, Brittany N.; Jin, Lee-Way; Murray, Melissa E.; Dickson, Dennis W.; Pericak-Vance, Margaret A.; Vance, Jeffery M.; Foroud, Tatiana M.; Mayeux, Richard; Tosto, Giuseppe; Neurology, School of Medicine
    Background: Alzheimer’s disease (AD) missing heritability remains extensive despite numerous genetic risk loci identified by genome‐wide association or sequencing studies. This has been attributed, at least partially, to mechanisms not currently investigated by traditional single‐marker/gene approaches. Polygenic Risk Scores (PRS) aggregate sparse genetic information across the genome to identify individual genetic risk profiles for disease prediction and patient risk stratification. Recent advancements have pivoted on innovative approaches utilizing OMICS data to construct such risk scores. Method: We employed a random forest algorithm to identify a list of gene candidates from bulk RNA sequencing data in prefrontal cortex from 565 AD brain samples (non‐Hispanic Whites, n = 399; Hispanics, n = 113; African American, n = 12) across six U.S. brain banks. Subsequently, we calculated their effect size on Braak staging using regression models to construct a polytranscriptomic risk score (PTRS). We employed two distinct models: “Ethnicity‐Agnostic” Model (randomly assigning samples to training and testing samples) and “Ethnicity‐Aware” Model (assigning NHW samples to training and Hispanics to testing sample). Analysis of variance and the receiver operating characteristics area under the curve (ROC AUC) was used to evaluate PTRS’s classification performances. We validated findings using the Religious Orders Study/Memory and Aging Project study (ROS/MAP, n = 1,095). Result: We found a significant difference in PTRS between samples with low vs. high Braak stages (≤4 vs. ≥5, p = 1*E‐04; Figure 1 upper panel). AUC was found to be 79‐81%, consistently in both Ethnicity‐Agnostic and Ethnicity‐Aware models (Figure 1 lower panel). Finally, the PTRS in ROS/MAP yielded a similar classification performance (p = 2*E‐04, AUC = 77%). Conclusion: Contrary to prior studies, we developed a PTRS with optimal transferability across ethnicities. This underscores the importance of developing novel tools to stratify and harmonize large brain repositories for AD.
  • Thumbnail Image
    Item
    Variational Autoencoder-based Model Improves Polygenic Prediction in Blood Cell Traits
    (bioRxiv, 2025-01-18) Li, Xiaoqi; Pang, Minxing; Wen, Jia; Zhou, Laura Y.; Raffield, Laura M.; Zhou, Haibo; Yao, Huaxiu; Chen, Can; Sun, Quan; Li, Yun; Biostatistics and Health Data Science, Richard M. Fairbanks School of Public Health
    Genetic prediction of complex traits, enabled by large-scale genomic studies, has created new measures to understand individual genetic predisposition. Polygenic Risk Scores (PRS) offer a way to aggregate information across the genome, enabling personalized risk prediction for complex traits and diseases. However, conventional PRS calculation methods that rely on linear models are limited in their ability to capture complex patterns and interaction effects in high-dimensional genomic data. In this study, we seek to improve the predictive power of PRS through applying advanced deep learning techniques. We show that the Variational AutoEncoder-based model for PRS construction (VAE-PRS) outperforms currently state-of-the-art methods for biobank-level data in 14 out of 16 blood cell traits, while being computationally efficient. Through comprehensive experiments, we found that the VAE-PRS model offers the ability to capture interaction effects in high-dimensional data and shows robust performance across different pre-screened variant sets. Furthermore, VAE-PRS is easily interpretable via assessing the contribution of each individual marker to the final prediction score through the SHapley Additive exPlanations (SHAP) method, providing potential new insights in identifying trait-associated genetic variants. In summary, VAE-PRS presents a novel measure to genetic risk prediction by harnessing the power of deep learning methods, which could further facilitate the development of personalized medicine and genetic research.