Application Framework for Computational Chemistry (AFCC) Applied to New Drug Discovery

Application Framework for Computational Chemistry (AFCC) Applied to New Drug Discovery

J. Tindle (University of Sunderland, UK), M. Gray (University of Sunderland, UK), R. L. Warrender (University of Sunderland, UK), K. Ginty (University of Sunderland, UK) and P. Dawson (University of Sunderland, UK)
Copyright: © 2012 |Pages: 17
DOI: 10.4018/jghpc.2012040104
OnDemand PDF Download:
No Current Special Offers


This paper describes the performance of a compute cluster applied to solve three dimensional (3D) molecular modelling problems. The primary goal of this work is to identify new potential drugs. The paper examines the following issues: computational chemistry, computational efficiency, task scheduling, and the analysis of system performance. The philosophy of design for an application framework for computational chemistry (AFCC) is described. Various experiments have been carried out to optimise the performance of a cluster computer, the results analysed and the statistics produced are discussed in the paper.
Article Preview

2. Molecular Modelling

Molecular modelling software allows the user to select atoms from the periodic table and to place them in a three dimensional workspace. In most modelling systems it is possible to build a three dimensional molecular structure by using a colour graphics user interface (GUI), refer to Figure 1. The initial position of the atoms is normally determined by the user calling upon common sense and experience. In all cases the actual position that the atoms assume in the real world is determined by the Laws of Physics. Computational chemistry is a general name for computer based algorithms that may be used to solve this type of problem. There are numerous algorithms that may be deployed and normally this involves computing the minimum value of an energy function to find the optimum solution.

Figure 1.

Molecular modelling workspace


For complex structures in many cases the rate of convergence is relatively slow and it is therefore often necessary to employ high performance computing methods to produce solutions in a reasonable period of time.

2.1. Model Convergence Time

There are three principle factors that influence the time required to produce an acceptable solution.

  • The initial position of the atoms selected by the user. This initial set of atomic positions is the seed for the numerical solver algorithm embedded in Gaussian_09.

  • The number of heavy atoms in the model of the molecular structure.

  • The basis set and the method deployed, for example, the molecular mechanics protocol which is intrinsically fast or the ab initio method MP4 that is relatively slow (Vinter, Davis, & Saunders, 1987; Sastry1, Johnson, Thompson, Goldberg, Martinez, Leiding, & Owens, 2007).

A system that accurately models the Schrodinger Equation will normally require a long time to produce a good solution whereas more approximate methods produce solutions more rapidly.

T is proportional to K * An


T is the time to produce a solution

K is a constant that is associated with method in use

A is the number of heavy (non-hydrogen) atoms

n is a scalar value of 4

When large numbers of atoms are involved in a model the time taken to generate a solution can be very long.

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 14: 6 Issues (2022): Forthcoming, Available for Pre-Order
Volume 13: 4 Issues (2021)
Volume 12: 4 Issues (2020)
Volume 11: 4 Issues (2019)
Volume 10: 4 Issues (2018)
Volume 9: 4 Issues (2017)
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing