AI model predicts chemical effects on gene expression, speeding drug discovery

A Michigan State‑led AI team built a gene‑expression predictor that reads a molecule’s structure and forecasts how it will turn genes on or off, shrinking the search for drug candidates from millions to a handful. Trained on millions of public experiments, the model already flagged promising compounds for aggressive liver cancer and untreatable lung fibrosis, showing how AI can accelerate real‑world therapeutics.