Interpretable decoding of cell fate from a snapshot of combinatorial signaling

How combinatorial cell signaling controls cellular decisions in the face of crosstalk is a fundamental problem in biology. A key open question is whether a single snapshot of signaling is sufficient to predict cell fate, especially given substantial evidence that signaling dynamics shape fate decisions. Here, we show that a snapshot of combinatorial signaling accurately predicts cell fate at the single-cell level in a model for human embryonic patterning. To this end, we developed Sig2Fate, a quantitative method integrating iterative immunofluorescence, information theory, and machine learning. Cell fate is encoded by combinatorial yet redundant signaling that reduces to a single angular coordinate in the high-dimensional signaling space, providing a simple interpretation of the signal-to-fate map. This map generalizes across variations in BMP concentration and pharmacological perturbations of ERK, Wnt, and YAP signaling, enabling prediction of drug responses from control data alone when signaling crosstalk is accounted for. Our findings provide a framework for predicting and explaining complex phenotypes from signaling perturbations across biological systems.