Cardiff University | Prifysgol Caerdydd ORCA
Online Research @ Cardiff 
WelshClear Cookie - decide language by browser settings

Interactive drug-design: using advanced computing to evaluate the induced fit effect

Anthopoulos, Athanasios 2013. Interactive drug-design: using advanced computing to evaluate the induced fit effect. PhD Thesis, Cardiff University.
Item availability restricted.

PDF - Accepted Post-Print Version
Download (3MB) | Preview
[img] PDF - Supplemental Material
Restricted to Repository staff only

Download (288kB)


This thesis describes the efforts made to provide protein flexibility in a molecular modelling software application, which prior to this work, was operating using rigid proteins and semi flexible ligands. Protein flexibility during molecular modelling simulations is a non-­‐trivial task requiring a great number of floating point operations and it could not be accomplished without the help of supercomputing such as GPGPUs (or possibly Xeon Phi). The thesis is structured as follows. It provides a background section, where the reader can find the necessary context and references in order to be able to understand this report. Next is a state of the art section, which describes what had been done in the fields of molecular dynamics and flexible haptic protein ligand docking prior to this work. An implementation section follows, which lists failed efforts that provided the necessary feedback in order to design efficient algorithms to accomplish this task. Chapter 6 describes in detail an irregular – grid decomposition approach in order to provide fast non-­‐bonded interaction computations for GPGPUs. This technique is also associated with algorithms that provide fast bonded interaction computations and exclusions handling for 1-­‐4 bonded atoms during the non-­‐bonded forces computation part. Performance benchmarks as well as accuracy tables for energy and force computations are provided to demonstrate the efficiency of the methodologies explained in this chapter. Chapter 7 provides an overview of an evolutionary strategy used to overcome the problems associated with the limited capabilities of local search strategies such as steepest descents, which get trapped in the first local minima they find. Our proposed method is able to explore the potential energy landscape in such a way that it can pick competitive uphill solutions to escape local minima in the hope of finding deeper valleys. This methodology is also serving the purpose of providing a good number of conformational updates such that it is able to restore the areas of interaction between the protein and the ligand while searching for optimum global solutions.

Item Type: Thesis (PhD)
Status: Unpublished
Schools: Pharmacy
Subjects: R Medicine > RS Pharmacy and materia medica
Uncontrolled Keywords: interactive drug-design; computing; molecular modelling
Date of First Compliant Deposit: 30 March 2016
Last Modified: 10 Oct 2017 15:46

Actions (repository staff only)

Edit Item Edit Item


Downloads per month over past year

View more statistics