Researchers at MIT and the Whitehead Institute have created IRIS, an artificial intelligence model that reads a cell’s gene activity and estimates which signalling pathways were active at each stage of development, even in cell types it has never seen. The work was published in Nature Methods.
The model was trained on data from thousands of human embryonic stem cells exposed to dozens of combinations of six major signalling pathways. The researchers found that each pathway leaves a gene-activity “fingerprint” that is consistent across cell types.
Tested on cells from mouse embryos, IRIS predicted when and where specific pathways switched on in cells destined to become heart, gut, muscle and spinal cord. It also predicted that activating a particular pathway would promote lung development, which experiments in mouse embryos confirmed.
The team, led by Pulin Li, wants to use the tool to guide stem cells in regenerative medicine, improve organoids for drug testing and study lung diseases such as asthma, lung cancer and pulmonary fibrosis.
