PREDICTION OF WHOLE REACTION PATHS FOR LARGE MOLECULAR-SYSTEMS

Reaction pathways for large systems such as proteins or other macromolecules are difficult to model using standard methods owing to the many degrees of freedom in such systems. Standard Euler-type methods which require knowledge of a transition structure, from which the path to two energy minima may be obtained are very inefficient when a whole reaction path is required since only small steps may be used with such methods. Furthermore, the location of a transition may itself be very difficult for large systems. Given these problems, an alternative approach (as suggested by Elber and Karplus, Chem. Phys. Lett., 1987, 139, 375), based on minimizing a functional of the entire path appears very attractive. This approach has not, previously, been evaluated for quantum-mechanical reaction surfaces, only for molecular mechanical surfaces. The assessment of a scheme, based on the Elber-Karplus approach, within both an ab initio and semi-empirical molecular orbital framework is presented. The method is evaluated by comparing the predicted paths with those obtained by the much used Gonzalez-Schlegel method for three model systems (isomerization of HCN, S(N)2 reaction of F- and CH3F and the addition of HF to ethene). The method is also tested on reactions without a transition state (hydride attack on an ester and a thioester). In the latter case, the conventional methods are more difficult to apply. The extension of the method to describe reactions in solution is discussed.