Performance assessment of single features. (A) Performance of single sequence features on Seq-set and Seq-control. CM, correlated mutation; DT, sequence distance; DO, co-localization within the same disordered region; CE, residue co-evolution; DM, co-localization within the same domain; RC, residue conservation; PP, PolyPhen-2 score; and SF, SIFT score. (B) Performance of single structural features on Str-set and Str-control. DT, structural distance; SP, shortest path distance; SS, pairwise secondary structure state; PK, co-localization within the same pocket; BF, B-factor; LN, Laplacian norm; TP, topological features; DP, depth and protrusion indices; and HB, number of hydrogen bonds. The disorder feature is divided into the sequence group, because this feature is derived from protein sequence, and all our structural features are calculated based on atomic coordinates.