Refining neural network predictions for helical transmembrane proteins by dynamic programming.
For transmembrane proteins experimental determination of three-dimensional structure is problematic. However, membrane proteins have important impact for molecular biology in general, and for drug design in particular. Thus, prediction method are needed. Here we introduce a method that started from the output of the profile-based neural network system PHDhtm (Rost, et al. 1995). Instead of choosing the neural network output unit with maximal value as prediction, we implemented a dynamic programming-like refinement procedure that aimed at producing the best model for all transmembrane helices compatible with the neural network output. The refined prediction was used successfully to predict transmembrane topology based on an empirical rule for the charge difference between extra- and intra-cytoplasmic regions (positive-inside rule). Preliminary results suggest that the refinement was clearly superior to the initial neural network system; and that the method predicted all transmembrane helices correctly for more proteins than a previously applied empirical filter. The resulting accuracy in predicting topology was better than 80%. Although a more thorough evaluation of the method on a larger data set will be required, the results compared favourably with alternative methods. The results reflected the strength of the refinement procedure which was the successful incorporation of global information: whereas the residue preferences output by the neural network were derived from stretches of 17 adjacent residues, the refinement procedure involved constraints on the level of the entire protein.