The hypothetical active site lattice. An approach to modelling active sites from data on inhibitor molecules.

Microcomputer-assisted methods are described that allow the mathematical construction of a hypothetical active site lattice (HASL) which can model enzyme-inhibitor interactions and provides predictive structure-activity relationships for a set of competitive inhibitors. The inhibitor set can be structurally dissimilar, including acyclic and cyclic moieties normally refractory to classical parameter-based quantitative structure-activity relationship strategies. With use of three-dimensional Cartesian coordinates representing energy-minimized inhibitor conformations, a four-dimensional space-filling description is generated, wherein the fourth dimension can be a user-selected physiochemical property such as hydrophobicity or electron density. The multidimensional lattices thus generated are used to quantitatively compare molecules to one another. Composite lattices of more than one molecule are merged with binding data to form a HASL capable of predicting inhibitor binding and relative orientation. Details of the algorithms and assumptions utilized are illustrated for competitive inhibitors of yeast glyoxalase-I and E. coli dihydrofolate reductase.