Figure 6

Normalized root squared error (NRSE) data point by data point before (at left) and after (at right) the optimization for the test set. Black squares are not considered when calculating the NRSE because the same protein used to perturb the model was then measured.