Figure 2. A. Percentage of the best models in CASP12–15 falling into three interface accuracy bins:

poor: ICS<0.50, red; good: 0.50 ≤ ICS < 0.75, blue; and high: ICS ≥ 0.75, green. B. Box-and-whisker plots for the best ICS scores generated in each round. The distributions are presented under different difficulty categories.