Empirical Potential Functions for Prediction of Protein Folding

点击翻译按钮获取中文摘要

One of the grand challenges in biophysics is the prediction of protein folding from the amino acid sequence. The investigator and coworkers have determined an empirical mathematical function that can recognize the native conformation for many proteins compared to many thousands of decoy conformation constructed from pieces of other protein crystal structures. They are now improving this function by testing different functional forms, training against more natives, and challenging it with wider classes of nonnative folds. If the function can be made smooth enough, it could quide a search toward the native structure as part of a structure prediction algorithm. In the current excitement over the human genome project, it is easy to forget that gene sequences are like blueprints for a house, and we are more concerned with the house than the blueprints. At the moment, we have no idea about the biological and medical role that about a third of the sequences play once they are translated into proteins, the three-dimensional molecular machines that make cells work. How proteins function depends critically on how these long chains fold up into particular shapes. This project is concerned with making sense of those thousands of mysterious sequences by trying to predict their three-dimensional structures. Workers in this laboratory have developed computer methods to choose the correct structure for a given sequence out of a lineup of many incorrect ones, as in a huge multiple choice exam. They are now moving toward answering an essay exam, where the sequence is given, and the structure must not merely be recognized, but constructed. Until this can be done, much of the value of the massive sequencing projects cannot be realized.