Protein Structure Analysis and Prediction

The protein structure prediction problem consists of three separate subproblems: modeling the protein structure, modeling the protein energetics, and the problem of searching for low-energy structures in high-dimensional space in the presence of many local minima. This study presents a new approach to protein structure prediction based on innovative methods in all three subproblems. The protein structure is modeled using natural coordinates, with less than one degree of freedom per residue. The protein energetics are modeled using two-dimensional statistical potentials and a linearized Hamiltonian. The two-dimensional statistical potentials are modeled as a function of the inter atomic distance and the local chain compression. The linearized Hamiltonian accounts for the fact that there is no one-to-one mapping between protein energy mechanisms and statistical potentials. The search for low-energy structures is done using a phenomenological structure model consisting of local structure fragments and tertiary contacts. The search space is reduced by searching along the finite set of possible tertiary contacts. The initial implementation of this approach is producing promising results. The results of this study also suggest a new method for structural alignment using natural coordinates, and new evidence for long-range sequence patterns.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Biophysics and Computational Biology

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2015

Author dc:creator

• Hunter, Cornelius George

Contributors dc:contributor

• Subramaniam, Shankar

• Biology, Molecular

Language dc:language

eng

Identifier

(MiAaPQ)AAI3017108

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/85416