Student Number : 0213068F - MSc Dissertation - School of Computer Science - Faculty of Science / Drug resistance of the HI virus due to its fast replication and error-prone mutation is a key factor in the failure to combat the HIV epidemic. For this reason, performing pre-therapy drug resistance testing and administering appropriate drugs or combination of drugs accordingly is very useful. There are two approaches to HIV drug resistance testing: phenotypic (clinical) and genotypic (based on the particular virus’s DNA). Genotyping tests HIV drug resistance by detecting specific mutations known to confer drug resistance. It is cheaper and can be computerised. However, it requires being able to know or learn what mutations confer drug resistance. Previous research using pattern recognition techniques has been promising, but the performance needs to be improved. It is also important for techniques that can quickly learn new rules when faced with new mutations or drugs. A relatively recent addition to these techniques is the Support Vector Machines (SVMs). SVMs have proved very successful in many benchmark applications such as face recognition, text recognition, and have also performed well in many computational biology problems where the number of features targeted is large compared to the number of available samples. This paper explores the use of SVMs in predicting the drug resistance of an HIV strain extracted from a patient based on the genetic sequence of those parts of the viral DNA encoding for the two enzymes, Reverse Transcriptase or Protease, which are critical for the replication of the HIV virus. In particular, it is the aim of this reseach to design the model without incorporating the biological knowledge at hand to enable the resulting classifier accommodate new drugs and mutations. To evaluate the performance of SVMs we used cross validation technique to measure the unbiased estimate on 2045 data points. The accuracy of classification and the area under the receiver operating characteristics curve (AUC) was used as a performance measure. Furthermore, to compare the performance of our SVMs model we also developed other prediction models based on popular classification algorithms, namely neural networks, decision trees and logistic regressions. The results show that SVMs are a highly successful classifier and out-perform other techniques with performance ranging between (94.13%–96.33%) accuracy and (81.26% - 97.49%) AUC. Decision trees were rated second and logistic regression performed the worst.

The development of essential medicines is being slowed by a lack of efficiency in drug development as ninety per cent of drugs fail at some stage during clinical evaluation. This attrition in drug development is seen not because of a reduction in pharmaceutical research expenditure nor is it caused by a declining understanding of biology, if anything, these are both increasing. Instead, drugs are failing because we are unable to effectively predict how they will work before they are given to patients. This is due to limitations of the current methods used to evaluate a drug's toxicity and efficacy prior to its development. Quite simply, these methods do not account for the full complexity of biology in humans. Systems pharmacology models are a likely candidate for increasing the efficiency of drug discovery as they seek to comprehensively model the fundamental biology of disease mechanisms in a quantit- ative manner. They are computational models, designed and hailed as a strategy for making well-informed and cost effective decisions on drug viability and target druggability and therefore attempt to reduce this time-consuming and costly attrition. Using text mining and text classification I present a growing landscape of systems pharmacology models in literature growing from humble roots because of step-wise increases in our understanding of biology. Furthermore, I develop a case for the capability of systems pharmacology models in making predictions by constructing a model of interleukin-6 signalling for rheumatoid arthritis. This model shows that druggable target selection is not necessarily an intuitive task as it results in an emergent but unanswered hypothesis for safety concerns in a monoclonal antibody. Finally, I show that predictive classification models can also be used to explore gene expression data in a novel work flow by attempting to predict patient response classes to an influenza vaccine.