computer modeling in molecular biology.pdf

4 Julia M. Goodfellow and Mark A. Williamsbeen assessed [35] and found to alter some properties of the system. Methods to efficientlyapproximate the long range interactions have been applied recently tobiological systems [36] in an effort to minimise these truncation effects.The underlying constraint in an application of simulation techniques to macromolecularproblems is often computer resources which may limit the size of thesystem, the sophistication of the force field and/or the length of time for which thesimulation can be undertaken. The latter is important as recent papers are emphasisingthe need for simulations of a nanosecond or longer [37, 381 in order to obtainrepresentative samples of the equilibrium configurations of large biological macromolecules.Important recent developments which will help to reduce the user timenecessary to carry out macromolecular simulations include the use of multipletimestep algorithms [39] and parallel architecture computers. Specific changes toalgorithms have been made to take advantage of a number of different types of parallelhardware [40-431.1.4 Prediction of Molecular ConformationOne of the major applications of molecular simulation algorithms is the predictionof the conformation of macromolecules. Simulated annealing methods are now usedroutinely to refine both X-ray crystallographic [44] and NMR solution structure [45]of proteins and DNA, often prior to conventional least-squares restrained refinement.In other applications, only limited experimental data may be available for therequired structure and these data may only be used to generate the starting modelwhich is refined using molecular dynamics. For example, one may be interested inpredicting the solution structure of a molecule given its crystal structure (Chapter 6),predicting the conformation of a complex between receptor and ligand when onlythe structures of the uncomplexed molecules are known [46], or predicting the effectof chemical modification on a structure [47, 481. There are numerous examples ofthese types of applications [7-91. An example of the use of simulation to study theimportant problem of the binding of peptides to the MHC class I antigen isdescribed in Chapter 7. Energy minimization and molecular dynamics may also beused to refine rule-based or homology built models of proteins such as thosedescribed in Chapters 2 and 7.