Figure 1From: Bayesian approaches to reverse engineer cellular systems: a simulation study on nonlinear Gaussian networksGoodness of fit and parsimony of the inferred models. Plots of the RMSE (a) and average number of parents per variable (b) for the nonlinear models, as a function of the parameter α of the hyperbolic tangent function (continuous blue curves); the dashed red curves refer to the RMSE and average number of parents for the linear regression model. The dash-dotted green curve in (b) represents the average number of parents in the differential equation model (i.e. the average number of true parents). Further analyses showed that, for α → + ∞, the RMSE saturates at 0.247, and the average number of parents saturates at 3.4.Back to article page