Use this URL to cite or link to this record in EThOS:
Title: Statistical approaches to the study of protein folding and energetics
Author: Burkoff, Nikolas S.
ISNI:       0000 0004 5349 3953
Awarding Body: University of Warwick
Current Institution: University of Warwick
Date of Award: 2014
Availability of Full Text:
Access from EThOS:
Access from Institution:
The determination of protein structure and the exploration of protein folding landscapes are two of the key problems in computational biology. In order to address these challenges, both a protein model that accurately captures the physics of interest and an efficient sampling algorithm are required. The first part of this thesis documents the continued development of CRANKITE, a coarse-grained protein model, and its energy landscape exploration using nested sampling, a Bayesian sampling algorithm. We extend CRANKITE and optimize its parameters using a maximum likelihood approach. The efficiency of our procedure, using the contrastive divergence approximation, allows a large training set to be used, producing a model which is transferable to proteins not included in the training set. We develop an empirical Bayes model for the prediction of protein β-contacts, which are required inputs for CRANKITE. Our approach couples the constraints and prior knowledge associated with β-contacts to a maximum entropy-based statistic which predicts evolutionarily-related contacts. Nested sampling (NS) is a Bayesian algorithm shown to be efficient at sampling systems which exhibit a first-order phase transition. In this work we parallelize the algorithm and, for the first time, apply it to a biophysical system: small globular proteins modelled using CRANKITE. We generate energy landscape charts, which give a large-scale visualization of the protein folding landscape, and we compare the efficiency of NS to an alternative sampling technique, parallel tempering, when calculating the heat capacity of a short peptide. In the final part of the thesis we adapt the NS algorithm for use within a molecular dynamics framework and demonstrate the application of the algorithm by calculating the thermodynamics of allatom models of a small peptide, comparing results to the standard replica exchange approach. This adaptation will allow NS to be used with more realistic force fields in the future.
Supervisor: Not available Sponsor: Leverhulme Trust
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: QA Mathematics ; QD Chemistry