Network sizes and model parameters.

and are the numbers of proteins and links, respectively. ( is used to stop the simulation. is not used as a constraint.) and have units of per gene per million years (Myr). and are probabilities (unitless). and are constraints from the data, while , , and are adjustable parameters. We used Monte Carlo simulations to optimize the parameter values, by minimizing the total symmetric mean absolute percentage error values of the simulated versus the experimental data (see SI). Our values of are substantially lower than because is the rate of mutations leading to the creation of a new PPI (rather than being a simple mutation rate, which would be much higher).