High-Dimensional Robotics at the Nanoscale — Kino-Geometric Modeling of Proteins and Molecular Mechanisms


Publication Details

Author(s): Budday D
Editor(s): Prof. Dr.-Ing. habil. Sigrid Leyendecker
Title edited volumes: Schriftenreihe Technische Dynamik
Publication year: 2019
Language: English


Proteins are dynamic biomolecules that perform an enormous variety of cellular

functions on a broad range of spatio-temporal scales. Their conformational ensemble

is a crucial determinant of functionality in health and disease, and thus,

its structural and dynamic characterization has been a major research focus

for the last 50 years. While experimental and computational advances have increasingly

enabled atomically detailed insights into molecular mechanisms, the

need for efficient, yet elaborate integrative computational methods to resolve

functional motions across scales remains considerable. Biophysically guided

by protein structure, this thesis lays out a robotics-inspired, kino-geometric

model that efficiently captures small- and large-scale collective motions, with

dihedral angles as torsional degrees of freedom and non-covalent interactions

as constraints. Using geometric tools, a universal theory for geometric rigidity

analysis is developed. The thesis demonstrates the equivalence to existing

topological tools for protein rigidity analysis, decomposing macromolecules into

larger rigid substructures and coordinated motions between them. Furthermore,

the geometrically derived analysis marks a major advance on existing

methods by providing an explicit basis for these coordinated motions. Thriving

on this advantage, the work develops an efficient, high-dimensional motion

planning algorithm to study molecular transitions between different stabilized

substates. The newly formulated algorithm dCC-RRT integrates the principle

of minimal frustration into a rapidly-exploring random tree (RRT), exploiting

non-native, steric contacts that emerge during the transition to redirect conformational

exploration via dynamic, Clash-avoiding Constraints (dCC). The

algorithm outperforms state-of-the-art peer methods and closely approximates

conformational transitions of several example systems known from Molecular

Dynamics simulations, intermediate crystal structures, and other experimental

data, thereby providing a structural basis for allosteric networks that

drive conformational change. Finally, a large-scale, multi-dataset benchmark

analysis demonstrates how our kino-geometric model captures highly conserved,

protein fold specific, dynamic information that often goes undetected

in comparable methods. Extending the methodology from rigidity-theory to

constraint-relaxation based collective motions, the approach bridges insights

from rigidity and normal mode based methods that agree well with a variety of

experimental data and more detailed simulations. Overall, our kino-geometric

modeling approach is a robust and efficient alternative to obtain high-level

insights into molecular mechanisms across scales, with broad applications in

protein engineering, drug design, and human health.

FAU Authors / FAU Editors

Budday, Dominik
Chair of Applied Dynamics

How to cite

Budday, D. (2019). High-Dimensional Robotics at the Nanoscale — Kino-Geometric Modeling of Proteins and Molecular Mechanisms (Dissertation).

Budday, Dominik. High-Dimensional Robotics at the Nanoscale — Kino-Geometric Modeling of Proteins and Molecular Mechanisms. Dissertation, 2019.


Last updated on 2019-29-04 at 17:26