| We currently have a great deal of experimental data about how particular proteins fold, but no description of protein folding that could be used as a recipe to fold proteins. Here, we explore a model of protein folding called Zipping and Assembly (Z&A). Zipping refers to the growth of topologically local substructures within the chain, and assembly refers to the coming together of already-formed pieces.; We first trace the development of our theoretical understanding of protein folding, and discuss what the current state of experiments and theory tell us about the prospect of developing a recipe for protein folding. In particular, we explain the need for a folding principle, a microscopic description of folding that could be used to predict the folding pathways and native state of an arbitrary sequence of amino acids.; To determine whether Z&A is a general method that can efficiently fold most of sequence space to global minima, we use the HP model, with which it is possible to enumerate full conformation and sequence spaces. We find that Z&A reaches the global energy minimum native states, even though it searches only a very small fraction of conformational space, for most sequences in the full sequence space. Furthermore, a Z&A-based search recapitulates key features of protein folding such as topology-dependent folding rates and characteristic early-vs.-late assemblies.; With the aim of putting the principles of Z&A to practice in an automated algorithm that works with atomically-detailed molecular mechanics models of proteins, we analyze a set of contact metrics extracted from a large database of short REMD simulations of peptide fragments. By building optimal Bayesian classification models, we show that it is possible to achieve up to 27% classification success in predicting native and non-native contacts from short fragment simulations. The most informative metric we tested was contact probability, with much smaller effects coming from multi-body metrics. Predictions for a test protein show how the classification model helps to prioritize the search process with the Zipping and Assembly Method (ZAM). |
