Reconstructing signaling histories of single cells via perturbation screens and transfer learning.

Transferable signaling response signatures learned from in vitro perturbation data enable accurate reconstruction of signaling histories in diverse in vivo cell types.

• Why it matters: Understanding cellular signaling histories in vivo is crucial for manipulating cell states and guiding interventions, but high-throughput methods for this are currently lacking.
• What they did: The authors generated a signaling perturbation atlas on human pluripotent stem cells and trained IRIS, a neural-network model, then applied it to mouse embryo single-cell data to infer signaling activities and histories.
• The result: This approach uncovers conserved signaling signatures across cell types, reveals heterogeneity, and reconstructs developmental signaling histories, facilitating targeted cell fate engineering.