“…These efforts intensified after the disorder prediction was introduced into the biannual CASP experiments in 2002. [14][15][16] The disorder predictors are categorized into four types: 17 1. propensity-based methods based on relative propensity of amino acids to form disorder/ordered regions: GlobPlot, 18 FoldIndex, 19 IUPred, 20 and Ucon; 21 2. machine learning-based predictors: DISOPRED2, 22 DISpro, 23 RONN, 24 ProfBval, 25,26 PONDR predictors, 9,14,[27][28][29][30] PreDisorder, 23,32 NORSnet, 21 DisEMBL, 18 and Spritz; 31 3. consensus-based methods that combine predictions from multiple base predictors: metaPrDOS, 33 GS-MetaServer, 34 MD, 35 PONDR-FIT, 36 and MFDp; 17 4. structural models-based approaches that make use of predicted tertiary structure models: PrDOS 37 and DISOCLUST. 38 The results from a recent comparative review 39 and the CASP8 competition, 16 demonstrate that the consensus-based methods, such as the GS-MetaServer, 34 MD, 35 and MFDp, 17 generally outperform other methods.…”