MIT AI Decodes How Cells Communicate, Study Reveals

A human embryo starts as a clump of identical stem cells and ends up with a liver, a brain, and a few hundred other specialized tissues, each built from cells that somehow knew what to become. That decision doesn't happen in isolation. Cells trade chemical messages with their neighbors, and those messages travel through signaling pathways, multi-step relay systems that convert an external signal into a change in gene activity inside the cell. The result tells a cell where it sits in the body, what point in development it has reached, and what job it needs to do next.

Scientists have wanted a way to read that relay backward, to take a cell and reconstruct the sequence of signals that shaped it. That kind of reconstruction could explain how tissues assemble correctly, and how that assembly breaks down in diseases like cancer or birth defects. The obstacle has been scale: researchers have assumed each signaling pathway behaves differently depending on the cell type it's acting in, which would mean mapping every pathway separately across every cell type, a project too large for standard lab methods to tackle.

A team at MIT and the Whitehead Institute, led by assistant professor Pulin Li and graduate student Nicholas Hutchins, set out to test that assumption.