Table 3.

Probability that the distributions of Table 2 features for WT and mutated structures are the same. The Mann-Whitney-Wilcoxon [28] test examines whether two samples are likely to derive from the same population taking as the null hypothesis that the distributions are equal. Therefore, a low p-value, here p = 0.05, after corrections is interpreted as that the feature may be used to distinguish between WT and mutated protein structures. The p values below 0.05 are presented in bold.

Predicted Structural Feature	P-value After Correction (BH)
Root mean square deviation of atomic position	5.815e-01
Confidence score	2.032e-02
Template modeling Score	5.6e-01
Solvent accessible surface area	1.6e-05
Total Gibbs free energy	2.03e-02
Pairwise distance-dependent energy	2.56e-02
Sidechain orientation-dependent energy	2.56e-02
Distance-related energy	2.56e-02
Angle-related energy	4.78e-02