Fig. 4

SCOUR performance on synthetic (a-c Smaller Synthetic Model and d-f Bigger Synthetic Model) and biological models (g-i S. cerevisiae and j-l E. coli) using noisy and low sampling frequency training and test data. Solid lines represent accuracy, sensitivity, specificty, and PPV performance of SCOUR on each model for each step of the framework. Dashed lines represent the PPV if interactions were randomly classified. Error bars represent the standard error of the mean. (n = 30 from independent autogenerated training replicates)