Font Size: a A A

Development and application of linear-scaling quantum methods for the study of biological macromolecules

Posted on:2004-07-24Degree:Ph.DType:Thesis
University:University of MinnesotaCandidate:Khandogin, Yang (Jana)Full Text:PDF
GTID:2450390011453882Subject:Chemistry
Abstract/Summary:
Recent advances in linear-scaling algorithms for electronic structure methods and continuum solvation models have opened a door for applications to solution-phase biological macromolecules. In this thesis, a new set of macromolecular quantum descriptors derived from the linear-scaling quantum/solvation methods is presented for the electrostatic and chemical characterization of biological macromolecules in solution. The descriptors include solvent-polarized electrostatic surface potential maps, equilibrated atomic charges, Fukui reactivity indices, approximate local hardness maps, and relative proton potentials. These descriptors can be obtained with modest computational costs and offer promise in applications to modern structure-based drug design.; They are applied to study regioselectivity of HIV-1 nucleocapsid protein toward small electrophilic agents, the stepwise protonation mechanism of the C-terminal zinc knuckle domain, and the distinctive electrostatic binding mode to SL2- and SL3-genomic RNA. The new insights gained from this work may contribute to the structure-based design of anti-HIV therapies. The electrostatic descriptors are further applied to study the inter- and intramolecular charge variations of A and B form DNA's and RNA's in solution, the solvent polarization on the electrostatic potential around DNA's, RNA's, and hammerhead ribozyme, the conformational dependence of the solvated electrostatic surface potential of the phosphate-binding protein mutant, and the order of pK a values of the acidic residues in ovomucoid turkey third domain.; This thesis also presents an implementation of the linear-scaling smooth COSMO (conductor-like screening model) solvation method and its analytic gradients. The integration of the smooth COSMO method into the d-orbital semiempirical framework MNDO/d allows stable gradient-based geometry optimizations and transition-state searches for biological molecules involving third-row elements. The combined MNDO/d and smooth COSMO method is tested on the geometry optimizations of tri-, tetra- and pentavalent phosphorus compounds, and applied to study the mechanism of the transphosphorylation reaction of a 3-ribose, 5-methyl phosphodiester model system and the results are compared with those from a recent QM/MM simulation. The combined MNDO/d and smooth COSMO method offers a great potential for the study of solution-phase chemical reactions of large systems and can aid in the parameterization of QM/MM potentials.
Keywords/Search Tags:Linear-scaling, Method, Biological, Potential
Related items