Loading Curie Briefs...
Getting the latest healthcare news for you
Getting the latest healthcare news for you
A new AI model called IRIS can decode the chemical signals that guide embryonic cells toward their specialized fates — think heart, lung, or muscle tissue. Developed by MIT's Whitehead Institute, IRIS identifies unique gene-activity "fingerprints" left by signaling pathways, and works across different cell types without needing to be retrained. The findings, published in Nature Methods, could supercharge stem cell engineering and organoid research.
Scientists at MIT's Whitehead Institute have developed an AI model called IRIS that can read the chemical "conversations" cells have during embryonic development. As an embryo grows, cells constantly send and receive molecular signals that determine what type of cell they'll become — brain, liver, lung, and so on. Until now, mapping these signaling histories required painstaking, cell-type-by-cell-type experiments. IRIS changes that.
The model works by detecting unique gene-activity "fingerprints" that each signaling pathway leaves behind in a cell. Crucially, these fingerprints are consistent across different cell types — meaning IRIS can decode signaling histories broadly, without being retrained for every new cell type. Think of it like a voice recognition system trained in one language that can still pick up patterns in others.
When tested on mouse embryo cells, IRIS accurately predicted signaling activity in cells destined to become heart, gut, muscle, and spinal cord tissue. It even identified the precise signals needed to generate lung-specific cells — a prediction later confirmed experimentally.
Key Takeaways:
Why it matters: By mapping the exact signals that steer stem cells toward specific fates, IRIS gives researchers a practical roadmap for engineering tissues in the lab — accelerating drug testing, organoid development, and ultimately regenerative therapies for damaged organs.
