The extended Central Dogma of molecular biology is simply represented as DNA --> RNA --> amino acid sequence --> 3D structure --> function, where the last three steps comprise the current protein structure/function paradigm.
     We noticed, however, that many proteins contain regions, sometimes quite large, that apparently don't fold into specific structures, but rather remain as flexible ensembles. In many cases, the disorderliness of these ensembles is required for function: these regions are "intrinsically disordered." With the advent of NMR methods of structure determination, several proteins have recently been shown to be intrinsically disordered from their amino- to their carboxy-termini. The existence of such proteins calls for a re-assessment of the current protein structure/function paradigm (see Wright and Dyson, J. Mol. Biol. 293: 321-331 [1999]).
     Since amino acid sequence is known to determine protein folding, we reasoned that sequence should determine disorder as well. To test this hypothesis, we used simple data analysis, neural networks and other machine learning tools. Our results suggest that, overall, proteins with mixtures of order and disorder are very likely the most common type in nature.
     These results encouraged us to consider possible roles of unfolded protein states in the realm of molecular biology, thereby leading us to a new classification scheme for structure/sequence relationships, a new classification scheme for molecular recognition, and a proposed critical role for disordered regions in the evolution of molecular biological networks. More recently we have initiated analysis of predicted disorder on a genomic basis. The results are startling in the large variation in the amount of putative disorder across the currently known genomes.
     If verified, our predictions will ultimately require a significant restructuring of the extended Central Dogma of molecular biology in order to include the important functions carried out by intrinsically disordered proteins. These observations are of course crucial to any attempts to deduce function from amino acid sequence and to the "structural genomics" project.