Despite reported examples of proteins with natively disordered regions, their commonness and function in nature is unknown. Solutions to these problems cannot be found by searching existing protein structure databases since they are likely to be strongly biased against natively disordered proteins. The approach proposed in this project consists of disorder estimation from primary sequence information through an intelligent data analysis. Preliminary results indicate that disordered regions can have crucial function and that such regions are evidently very common in nature. If confirmed by an in depth study, this represents a major shift in the world view of protein structure and function. The aims of the current study are to further investigate this problem by: a) enlarging and improving a preliminary data set of proteins with disordered regions; b) improving the preliminary protein disorder prediction system accuracy, interpretability, and accessibility; c) testing the validity of the obtained knowledge. The techniques for analyzing massive volumes of data are likely to serve as stepping-stones towards the development of sophisticated prediction systems for some important problems in biochemistry. Possible biological applications include studies aimed at understanding protein structure and function, as well as improving protein crystallization process efficiency.
