Fig. 2

Illustration of the qualitative validations and predictions performed on the dynamical model. A precursor state is composed of the least number of entities to be “present” (represented in blue) so that all processes of the map can be “switched on” in the future. a For a given phase N, we check for the existence of trajectories from a precursor state leading to a state where at least one marker of that phase phase is “present” and all markers of all other phases are “absent”. b Same as in a, but using a model where all markers of the phase directly preceding phase N (i.e., of phase N − 1) are disabled by a Loss-of-Function (LoF). c For a given phase N and a given marker of that phase, we compute the mutations [Loss-of-Function (LoF) or Gain-of-Function (GoF)] that prevent the existence of trajectories leading to a state where this marker is “present”