The application of Artificial Intelligence to the conformational analysis of strained molecules

Abstract

An algorithm for predicting the conformations of strained molecules using Artificial Intelligence techniques is described and illustrated with some typical examples.

The computer program LUDI: A new method for the de novo design of enzyme inhibitors

“…Both limitations could at least in principle be overcome by performing a conformational analysis of the compounds in the active site, for example with distance geometry methods [9] and/or by using larger datasets e.g. by converting 2D structures to 3D structures with the programs CONCORD [10] or WIZARD [11]. It is clear, however, that such a procedure will be very demanding both in CPU time and in disc space.…”

Automated conformational analysis: Algorithms for the efficient construction of low-energy conformations

“…A conformational entity is a group of one or more connected units whose conformations are all linked such that its conformation can be changed without requiring any other entities in the molecule also to change conformation. They were first introduced during the development of the method by which WIZARD relieves strain in criticised conformations [4]. The entities in the molecule are discovered by examining the types of unit connectivity present.…”

“…For example, in the entity connectivity graph in Fig. 9 the order in which the entities are joined would be [1,2,5,9,10,3,6,4,7,11,8]. The entity sequence is then converted into the equivalent unit join-order list, ensuring that each unit always adjoins at least one previous unit in the list.…”

Fundamentals of drug design from a biophysical viewpoint

“…Thus, results from such comparisons will depend on the model used to generate the three-dimensional structure of the test compounds. Structures can be generated using knowledge based approaches such as the programs CONCORD or WIZARD (Rusinko III et al 1988;Leach et al 1990), or minimum energy configurations from quantum mechanical or empirical force field calculations can be used. Similarity indices have also been derived as trajectory averages from molecular dynamics simulations.…”