UCONGA : Universal Conformer Generation and Analysis.

Abstract

This work presents a solution to the problem of adequately sampling large conformational spaces computationally using a deterministic method that is guaranteed to identify all of the sterically-allowed molecular conformations to within a given bond rotation resolution. This involved development of new conformer generation software called UCONGA and its use, along with other known computational methods, to study various systems with interesting conformational properties. This thesis presents details of the new conformer ensemble generation method UCONGA, and the tools developed simultaneously alongside UCONGA to analyze the generated conformer ensembles. These analysis tools aim to find clusters of similar conformers so that representative or unusual ones can be selected for further study, and can be used on conformer ensembles generated through other methods as well as those generated by UCONGA. It then discusses the extension of UCONGA with a divideand- conquer algorithm to improve its performance with increasing molecular size. Initially, the suitability of various computational chemistry methods for studying conformer ensembles is tested. A series of case studies are then presented sampling different challenges for conformer generation methods. This first case study examines extremely sterically crowded molecules with few stable conformers, which often adopt unusual dihedral angles to avoid steric clashing. The second case study is on molecules bound to surfaces, which require different metrics for the conformational properties, and are in a different chemical environment than the other case studies. The final case study is on highly flexible molecules with large conformer ensembles.