Molecular Design in Chemical and Biological Systems
Open Access
- Author:
- Lehmann, Andreas
- Graduate Program:
- Chemical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- October 28, 2004
- Committee Members:
- Costas D Maranas, Committee Chair/Co-Chair
Margaret K Duda, Committee Member
Kristen Ann Fichthorn, Committee Member
Barbara Jane Garrison, Committee Member - Keywords:
- quantum chemistry
molecular design
protein design
protein engineering - Abstract:
- The focus of this thesis is the development of methods for computational molecular design with the goal to aid the experimentalist and reduce laboratory cost. Two topical complexes are treated: small molecules that are of interest in the chemical industry and much larger protein molecules in biological systems. These two types of molecules require the use of different structure-property relationships and methodical approaches of how the search is conducted. In the first part, the combination of quantum chemical methods with optimization techniques for molecular design of small molecules is examined. A hydrofluorocarbon refrigerant design example and a solvent design example illustrate the proposed framework. The hydrofluorocarbon compounds are optimized for their heats of formation and the potential solvents are searched for capacity, selectivity and environmental safety. In both examples, a genetic algorithm is applied to generate and screen candidate molecules. The molecular properties are evaluated using a combination of quantum chemical calculations and group contribution methods. The feasibility of the proposed approach for small molecules is assessed and it is found that establishing a proper trade-off between accuracy of the quantum chemical method and computational expense is vital. In the second part, optimization techniques are combined with force field methods for protein design. The focus here lies on making a clear distinction in how to account for different protein properties such as protein stability and function. A framework is proposed for predicting promising mutants of a protein with improved binding functionality for a known ligand of this protein. The method is applied to two design examples. The first example validates the method by comparing computationally designed mutations with experimental data. In the second example, predictions are given for promising multiple mutations of the plant protein concanavalin A, which will lead to enhanced binding of glucose.