Figure 5.

The probability of mass shift induced by different intervention types

100 PBNs were built on the same cell state pair (Departure: 1010101110, Destination: 1001010011) using the sampled network approach. The boxplots show the probability of mass shift induced by different intervention types: The intervention types are denoted by UX_DY_IZ where U and D stand for departure and destination cell states, X and Y are the values of the TF in the departure and destination cell states and I stands for Intervention and Z is the value that the TF is set to after intervention, where X, Y, and Z are binary values. Example representation: U1_D1_I0: TF’s value in departure cell state = 1, TF’s value in destination cell state = 1 and Intervention is setting the TF to 0. Group 1 (U0_D0_I0 and U1_D1_I1) has almost no effect on the steady-state distribution while in Group 2 (U0_D0_I1 and U1_D1_I0) the steady-state probability mass of cell states is distributed over other states and the initial cell states are lost. In Group 3 (U0_D1_I0 and U1_D0_I1), the steady-state probability mass is moved toward the departure cell state while in Group 4 (U1_D0_I0 and U0_D1_I1), most of the steady-state probability distribution is moved toward the destination cell state.