The Prediction of a Protein and Nucleic Acid Structure: Problems and Prospects

Abstract

Recent advances in DNA and protein-sequencing technologies have made an increasing number of primary structures available for theoretical investigations. The prediction of a higher-order protein, and nucleic acid structure in particular, is an area where computational approaches will be able to complement the lack of experimental observations. We review some of the problems related to structure predictions: sequence homology searches, secondary structure prediction in RNAs, and regular structure prediction in proteins. The first two are mathematically well-defined problems, for it is not usually necessary to consider long-range interactions. The solution to a smaller segment is a part of the solution to the entire sequence. Thus, the problem can be solved by dynamic programming algorithms. The prediction of protein structures poses a more complex combinatorial problem, as illustrated in our statistical mechanical treatment. A promising approximation is to calculate locally optimal structures stabilized by relatively short-range interactions, and then to include longer-range effects as interactions between the locally optimal structures.