Life Simulator

Flight simulators, weather forecasts, crash-test modelling, and VR moonwalks are all virtual replications of real-world phenomena that provide tangible benefits despite the artifice. Researchers have applied this logic to microscopy, developing an AI system to simulate protein distribution in cells at a scale microscopy cannot achieve. The system was trained on 1.23 million microscopy images, covering 12,800 proteins, and can predict the pattern of a chosen protein on a basic microscopy image (pictured, simulating many different proteins distributed on a single cell). They generated a dataset of 30 million simulated images, creating thousands of virtual cells and a platform for computational investigation of biology. Researchers can ask questions about complex predicted protein network distribution and activity that real images cannot answer, opening avenues of exploration for human health and disease that can be explored before turning to real-world experiments.

Written by Anthony Lewis

Anthony Lewis

Writer

Ant was the web editor in the early days of BPoD, and has since worked with the UK’s top science organisations as a writer and multimedia producer. He now splits his time between freelance work and managing multimedia at The Lancet.

Website

• Video (sped up 2 x original) from work by Huangqingbo Sun and colleagues
• Department of Bioengineering, Stanford University, Stanford, CA, USA
• Video originally published with a Creative Commons Attribution 4.0 International (CC BY-NC 4.0)
• Published in bioRxiv, May 2026 (not peer reviewed)