MIT's IRIS model decodes cell signaling fingerprints

MIT researchers led by Pulin Li have published a study in Nature Methods detailing IRIS, a machine learning model that decodes cell-to-cell communication during embryonic development. The model identifies unique gene activity patterns, or 'fingerprints,' left by signaling pathways, allowing it to predict cell fate across different tissue types without prior mapping for each specific cell. Validated on mouse embryos, IRIS successfully guided stem cells toward lung-specific development, confirming its predictive power. This AI-driven method offers a scalable alternative to traditional experimental mapping, potentially accelerating advances in regenerative medicine and organoid research.