Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.690494
Title: Advances in Monte Carlo techniques with application to lattice protein aggregation
Author: Xu, Yuanwei
ISNI:       0000 0004 5923 8689
Awarding Body: University of Warwick
Current Institution: University of Warwick
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Motivated by an intricate mechanism to transport folded proteins across biological membranes, known as the Twin-arginine translocation (Tat) pathway, we construct lattice protein models in an attempt to study the aggregation of the membrane protein TatA, which plays an integral role during active Tat translocation. We develop force field that characterizes intra- and inter-residue interactions, as well as how each residue interacts with its environment. Although written with the Tat process in mind, this thesis is mainly devoted to developing efficient Monte Carlo schemes for biomolecular simulations, which are often challenged and impeded by complex energy landscapes. To tackle the local trap problem that is typical in Metropolis sampling, the idea of dynamic weighting is incorporated into the parallel tempering (PT) algorithm. Our results show that, when applied to the lattice-protein model, the modified PT algorithm is capable of locating the low energy state much more quickly, but does not produce reliable estimates for equilibrium expectations. A modern method for free energy calculation, called the multistate Bennett acceptance ratio (MBAR) estimator, is reviewed from a statistical perspective, reminiscent of the underlying statistical theory which the method is based upon. Instead of adopting the common practice of using MBAR as a post-simulation analysis tool, we propose a new approach that integrates MBAR into simulation, allowing the simulation to benefit from the statistical optimality of the MBAR estimator. We show that the MBAR-enhanced Monte Carlo improves simulation efficiency of the lattice-protein aggregation model and, since it can also be applied to continuous models, provides a promising alternative to the study of more realistic systems. The new method is then applied to our model of TatA, where the protein features both a transmembrane and an amphipathic helix. The effect of individual helices on dimerization was studied and problem with the move set was identified. In this thesis, we used pull move and translation move as our Monte Carlo trial moves. Implementation details of pull moves, which are often omitted by many researchers who use them for sampling configuration space, are given in Chapter 1. We show that, for our double-helix TatA model, pull moves are no longer efficient moves and therefore, for future study of more realistic systems, we point to several methods which all attempt to design efficient trial moves. Aggregation of more than two polymer chains was also considered in this thesis.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.690494  DOI: Not available
Keywords: QA Mathematics ; QH301 Biology
Share: