Show Notes
️ Episode 39: Whole-Genome Polygenic Score Inference — Fast and Memory-Efficient Algorithms for Scalable PRS
In this episode of Base by Base, we dive into the methodological breakthroughs of Zabad et al. (2025) in The American Journal of Human Genetics, where the authors introduce an updated variational inference framework (VIPRS v0.1) and novel data structures that together enable polygenic risk score computation across tens of millions of variants with dramatically reduced runtime and memory requirements .
Key highlights: Zabad and colleagues design a compressed sparse row (CSR) format for linkage-disequilibrium matrices that cuts storage by over 50-fold; they implement quantization and triangular-only representations to shrink memory footprint while preserving accuracy; they rewrite core inference routines in C/C++, leverage single-precision arithmetic, multithreading, and BLAS/SIMD optimizations to achieve more than an order-of-magnitude speedup per iteration; and they showcase whole-genome scoring on UK Biobank data of up to 18 million variants, completing analysis in under 20 minutes using less than 15 GB of RAM .
Conclusion: By uniting efficient LD compression, numeric optimizations, and parallel inference strategies, VIPRS v0.1 makes whole-genome polygenic scoring both accessible and practical for large-scale genomic studies, paving the way for more comprehensive and accurate genetic risk prediction .
Reference: Zabad S., Haryan C. A., Gravel S., Misra S., Li Y. (2025). Toward whole-genome inference of polygenic scores with fast and memory-efficient algorithms. American Journal of Human Genetics, 112, 1–19. https://doi.org/10.1016/j.ajhg.2025.05.002
License: This episode is based on an open access article published under the Creative Commons Attribution 4.0 International License (CC BY 4.0) – https://creativecommons.org/licenses/by/4.0/
