Figure 2.

Nemenyi test for pairwise comparison of the top four models of the top-ranking cluster generated by four different modeling approaches: ab initio docking, AIR-guided docking, superimposition and CA-CA-guided docking. For each case, the four modeling methods are ranked based on their performance, with a rank of 1 for the best performing method and 4 for the worst. The average rank of each method is shown. Two methods are significantly different from each other if their average rank difference is larger than the CD at the significance level of 0.05, which is 0.62. Methods are connected with a bar if they are not significantly different from each other.