Protein Structure Computation

Abstract

The sequencing of the human genome and subsequent increase in understanding of the processes of the human body have brought great expansion in the scope of therapeutic drug applications. In the last year, well-publicised reports of possible treatments for diseases such as Alzheimer's, Huntington's, and breast cancer have been seen; the common theme to these afflictions (and many others) is that they are caused or sustained by certain proteins. These molecules are of central importance to biochemistry. This paper seeks to demonstrate how computer science can be and has been of benefit to research in this field. Firstly, an introduction to the fundamental principles of protein structure and behaviour is given. With this foundation prepared, the basic problems in computational biochemistry (folding, alignment, annotation, and docking) are summarized. Applications of solutions to the above problems are varied, but we focus on one of much interest worldwide: that of rational drug development. This topic is described, with illustrations of recent projects at both desktop and globally-distributed scales. In particular, the use of the FFT for convolution-based correlation scoring is outlined. An implementation of a docking program employing an established FFT algorithm is presented. The program defines simple Java classes for representing proteins and other molecules, and also an interface for docking algorithms. This permits several algorithms to be applied in turn to a pair of molecules, each progressively improving the quality of docking. An additional refinement method is introduced to this framework. Besides the limitations of the FFT method, this work has highlighted problems with using Java for substantial molecular processing.