Omnigenic Model: Neural Network for Disease Risk Prediction Sparks Academic Benchmarking Debate

anshulkundaje · x · 2026-08-06

Researchers introduced a new preprint for the Omnigenic Model (OGM), a neural network architecture designed for human genetics. It leverages biological system structure to aggregate genetic variations for disease risk prediction.

However, academics pointed out that using deep learning to improve Polygenic Risk Scores (PRS) is an emerging trend. They emphasized the need for rigorous benchmarking and referencing of different efforts (such as G2PT), warning that the field could otherwise become messy and difficult to navigate.

Related event: Neural Network Architecture OGM Proposed for Human Disease Risk Prediction(2 posts)→

Original post →

More from Research

Research channel →