AI's Virtual Cell Challenge Aims to Simulate Human Cells

science

Scientists are racing to build computer models that mimic the human cell well enough to run biological experiments virtually instead of in a lab.

On Nov. 19, the Arc Institute in Palo Alto, Calif., announced the winners of its second annual Virtual Cell Challenge, an AI-backed competition to predict how cells respond when genes are switched off. About 1,000 teams took part in the final round, a jump from last year, and the contest drew more than 1,600 teams across early and final phases.

The winning models, including one from Illumina and one from independent researchers, showed AI could learn from biological data to forecast cell behavior. The broader effort aims at virtual cells for faster drug discovery, though more data is needed before such models can match real biology.