Predicting chemical reactivity by computer



Mathematical models for predicting the fate of pollutants in the environment require reactivity parameter values —that is, the physical and chemical constants that govern reactivity. Although empirical structure-activity relationships have been developed that allow estimation of some constants, such relationships generally hold only within limited families of chemicals. Computer programs are under development that predict chemical reactivity strictly from molecular structure for a broad range of molecular structures. A prototype computer system called SPARC (SPARC Performs Automated Reasoning in Chemistry) uses computational algorithms based on fundamental chemical structure theory to estimate a variety of reactivity parameters (e.g., equilibrium/rate constants, UV-visible absorption spectra, etc.). This capability crosses chemical family boundaries to cover a broad range of organic compounds. SPARC does not do “first principles” computation, but seeks to analyze chemical structure relative to a specific reactivity query in much the same manner in which an expert chemist would do so. Molecular structures are broken into functional units with known intrinsic reactivity. This intrinsic behavior is modified for a specific molecule in question with mechanistic perturbation models. To date, computational procedures have been developed for UV-visible light absorption spectra, ionization pKa, hydrolysis rate constants, and numerous physical properties. This paper describes the logic of the approach to chemistry prediction and provides an overview of the computational procedures. Additional papers are in preparation describing in detail the chemical models and specific applications.