The interest in intrinsically disordered proteins has greatly increased, as it has become clear that they are very widespread, especially in eukaryotic organisms. Functionally, they appear to play a significant role in the control of many cellular processes and signalling pathways and have been, also, associated with a number of diseases ranging from cancer to Alzheimer's. Thus, there is enormous interest in attempts to predict disordered regions in proteins solely from knowledge of their amino acid sequences. In this study, we assess the quality of predictions for 25 groups on predicting disordered regions in 122 target proteins. In addition, we suggest the need of a "knowledge-independent" measure that would enable one to normalize the results of the different CASP experiments and to determine whether the disorder prediction field had improved across the years.
Copyright 2009 Wiley-Liss, Inc.