Publications

Global Biobank analyses provide lessons for developing polygenic risk scores across diverse cohorts

Auteurs : Y. Wang, S. Namba, E. Lopera, S. Kerminen, K. Tsuo, K. Läll, M. Kanai, W. Zhou, K. H. Wu, M. J. Favé, L. Bhatta, P. Awadalla, B. Brumpton, P. Deelen, K. Hveem, V. Lo Faro, R. Mägi, Y. Murakami, S. Sanna, J. W. Smoller, J. Uzunovic, B. N. Wolford, C. Willer, E. R. Gamazon, N. J. Cox, I. Surakka, Y. Okada, A. R. Martin, J. Hirbo

The researchers assessed the performance of polygenic risk scores (PRSs) in predicting disease risk across diverse global populations using data from the Global Biobank Meta-analysis Initiative (GBMI). They constructed PRSs using two methods: pruning and thresholding (P + T) and PRS-continuous shrinkage (CS). Data from nine biobanks, including the Ontario Health Study, were analyzed for 14 different disease endpoints. Results showed that PRS-CS generally outperformed the P + T method, particularly for diseases with higher SNP-based heritability.