The current costs of drug discovery are extremely high and need to be addressed if diseases such as AIDS and malaria are to be combated. The major reasons for the high costs are the use of expensive in vitro methods and the high failure rate of drugs at clinical testing phases. In silico techniques hold tremendous potential in addressing the high costs. In silico drug design can be done at a fraction of the cost of in vitro techniques, and can be used in synergy with in vitro techniques by doing much of the screening before any experimental studies, thereby reducing the chemical space to be searched experimentally. In silico techniques can also enhance the quality of drug candidates sent to clinical phases, increasing the probability of success. In this study techniques were investigated to build analogous ligand libraries with scaffolds and molecular building blocks through a user guided process, including the development of the LIGLIB program, which is an Open Source package for lead development. The Markush molecular enumeration technique was implemented in C++ with a Python front-end extending it to the Python molecular visualization tool, Chimera. The software makes use of chemical graphs to make permutations according to user inputs, generating an output library in SMILES and Mol2 format, the later of which is generated by Corina. As part of the validation of the software it was used in a lead discovery experiment which targeted Plasmodium falciparum Glutathione S-transferase. The developed software was able to generate a series of suitable molecules, thereby validating the Markush molecular enumeration technique as well as its implementation in LIGLIB.
Dissertation (MSc (Bioinformatics))--University of Pretoria, 2008.