Statistical properties of liquid protein-water molecular system dynamics

  • Jutharath Voraprateep

Student thesis: Doctoral ThesisDoctor of Philosophy

Abstract

It is considered an established fact that water plays the major role in protein
motion, there is a close connection between the water dynamics and the
protein conformational dynamics.

We report on statistical analysis of such conformational dynamics obtained
using classical molecular dynamics simulations with explicit water. We
investigate specific moments in time when one of the dihedral angles of
a simulated protein (a peptide dialanine) makes a large amplitude change
causing a conformational transition in the peptide. We are interested in
finding statistical correlations between the values of the angle at the moment
of transition and several moments in advance of the transition (between
0.0 and 50.1ps). We also investigate how these correlations change when
conditioned on the presence of water at different locations in space around
the peptide. The challenge is in a large number of parameters that influence
the conformational dynamics, which leads to multivariate probabilities. As
statistical tools, we use pair-copulas and the Kendall's tau correlation.

Copulas are a special way of representing multivariate probabilities. Paircopulas
construction (PCC) decomposes a multivariate probability density
into bivariate copulas, so-called pair-copulas. D-vine is one of graphical
models that give a specific way of decomposing the probability density.
The dependency structure is determined by the bivariate copulas and a
nested set of trees using pair-copula. For this research, we apply the D-vine
to study the statistical correlations between variables describing molecular
conformation of a peptide and the properties of water molecules surrounding
the peptide.

We have found that the dynamics of peptides conformation possesses temporal
correlations well in advance of the moments of conformational transitions.
Moreover, when conditioned on the presence of water molecules
at a few very specific locations in the first hydration shell of the peptide,
these correlations become stronger and longer in time. This quantifies the
influence of water on the conformational transitions and specifies water
molecules that appear critical for the peptide to make successful conformational
transition.
Date of Award20 Jun 2017
LanguageEnglish
SupervisorDmitry Nerukh (Supervisor)

Cite this

Statistical properties of liquid protein-water molecular system dynamics
Voraprateep, J. (Author). 20 Jun 2017

Student thesis: Doctoral ThesisDoctor of Philosophy