Applications of machine learning to problems in biomolecular function and dynamics

Veevers, Ruth (2020) Applications of machine learning to problems in biomolecular function and dynamics. Doctoral thesis, University of East Anglia.

[img]
Preview
PDF
Download (9MB) | Preview

Abstract

Biomolecules such as proteins and nucleic acids are involved in all biological processes. As they take part in these processes, biomolecules often undergo motions and changes in their conformation that are related to their function. This thesis presents research into and development of methods to support the study of the dynamics of these changes and their relationship to the biomolecular function.

Due to the scale of the structures and speed of the changes, common methods of determining (or “solving”) the structures of biomolecules cannot capture the change in conformation. Detail of the changes must be extrapolated from changes observed between multiple solved states of the same structure. We present a novel method of visualising potential motions of atoms comprising biomolecules, estimated from solved structures at the start and end of the trajectory. Comparisons show that our method produces atomic coordinates that pass closer to known intermediates than those produced by similar existing methods.

Our visualisations treat each atom as an individual body, but the conformational changes of proteins can be broken down into the motions of “dynamic domains”, which are sections of proteins that move semi-rigidly, controlled by flexible hinge bending regions. Tools such as the DynDom program identify and analyse the motions of these dynamic domains displayed between pairs of solved structures. We designed and developed DynDom6D, a new version of the DynDom program for very large macromolecules that assigns atoms to domains or hinge bending regions using 6-dimensional k-means clustering.

Item Type: Thesis (Doctoral)
Faculty \ School: Faculty of Science > School of Computing Sciences
Depositing User: Chris White
Date Deposited: 08 Apr 2021 09:31
Last Modified: 08 Apr 2021 09:31
URI: https://ueaeprints.uea.ac.uk/id/eprint/79640
DOI:

Actions (login required)

View Item View Item