Protein Structure Prediction

Abstract

In this chapter, we assume that most readers know a little about proteins and protein structure prediction, but that their knowledge might now be fuzzy and out of date. In order to refresh their knowledge and get novices up to speed, we concentrate on the most basic principles. We outline the categories of protein structure prediction before examining the underlying methods used in 'successful' prediction tools. Each method will be accompanied by examples of biological applications from recent papers and a section highlighting their advantages and disadvantages. HISTORYWhile knowledge about protein structure has a long history going back to Astbury's early experiments on hair and silk (and beyond), its current form came into being with the solution, by X-ray crystallography, of the first globular proteins at near atomic resolution. These consisted of two closely related globin structures (by Kendrew and Perutz and coworkers), followed by the first enzyme structure (lysozyme, by Phillips and co-workers). These structures were somewhat unexpected at the time: given the near crystalline regularity of Astbury's hair and silk structures (along with the structure for DNA proposed then), the globular proteins were described by Kendrew as lacking symmetry and visceral. Even worse, the structure of lysozyme did not immediately provide any clue as to how it, or any enzyme, might work! As more structures were solved, some of these early unexpected aspects were partially rectified: symmetry was rediscovered in the structure of triosephosphate isomerase, and plausible catalytic mechanisms for lysozyme and other enzymes were established. However, despite progress, it remains true that proteins do not yield their secrets easily. When a new structure is solved, it is often still difficult to describe its structure in a systematic way and place it in relation to other structures. Similarly, if a structure is solved for a protein of unknown function, it is not necessarily easy or possible to suggest a function from just the structure.