Summary of the DREAM8 Parameter Estimation Challenge: Toward Parameter Identification for Whole-Cell Models
Parameter and prediction error trajectories of the top five scoring teams (A: Crux, B: New Dream, C: ICM Poland, D: Alucinatori, E: Whole-Sale Modelers). For comparison, insets depict each team’s trajectory using common x- and y-scales defined in A. Submissions are colored by time. Light blue indicates each team’s first submission; dark blue indicates each team’s last submission. Light and dark blue dots indicate the first and last submissions, respectively. Trajectories show that Team Crux employed a deterministic algorithm which steadily improved their performance over the competition, whereas the other teams employed stochastic algorithms which randomly sampled the parameter space, resulting in non-monotonically increasing performance. F: Box plots of the estimated values of each unknown parameter. Blue indicates the 15 modified parameters; green indicates the 15 unknown and unmodified parameters. G: Correlations between the estimated value of each unknown parameter and the prediction error.