Computational Molecular Design Using Tabu Search

Abstract

The focus of this project is the use of computational molecular design (CMD) in the design of novel crosslinked polymers. A design example was completed for a dimethacrylate as part of a comonomer used in dental restoration, with the goal to create a dental adhesive with a longer clinical lifetime than those already on the market. The CMD methodology begins with the calculation of molecular descriptors that describe the crosslinked polymer structure. Connectivity index are used as the primary set of descriptors, and have been used successfully in other CMD projects. Quantitative structure property relationships (QSPRs) were developed relating the structural descriptors to the experimentally collected property data. Models were chosen using Mallows' Cp with correlation coefficient significance. Desirable target property values were chosen which lead to an improved clinical lifetime. Structural constraints were defined to increase stability and ease of synthesis. The Tabu Search optimization algorithm was used to design polymers with desirable properties. Finally, a prediction interval was calculated for each candidate to represent the possible error in the predicted properties. The described methodology provides a list of candidate monomers with predicted properties near the desired target values, which are selected such that the adhesives will show improved propertoes relative to the standard HEMA/BisGMA formulation. The methodology can be easily altered to allow for additional property calculations and structural constraints. This methodology can also be used for molecular design projects beyond crosslinked polymers.