Support vector machine prediction of HIV-1 drug resistance using the viral nucleotide patterns

Abstract

The drug resistance of the human immunodeficiency virus (HIV), which is due to its fast and error-prone replication, is a key factor in the failure to combat the HIV epidemic. Performing pre-therapy drug resistance testing and administering appropriate drugs, or a combination of drugs, is therefore very useful. Genotyping tests HIV drug resistance by detecting specific mutations known to confer drug resistance. It is also cheaper than phenotypic testing and can be computerised, but it requires being able to know which mutations confer drug resistance. Previous research using pattern recognition techniques has been promising, but needs to be improved. It is also important that the techniques be highly adaptive when faced with new mutations or drugs. A relatively recent addition to pattern recognition techniques is the Support Vector Machine (SVM). SVMs have proved very successful in many benchmark applications such as face recognition and text recognition, and have also performed well in many computational biology problems. 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 the viral DNA. We use as a case study that part of the HIV DNA that codes for the two enzymes, reverse transcriptase and protease, which are critical for the replication of the virus. To evaluate the performance of SVMs we used cross-validation to measure the unbiased estimate on 2045 data sets. The accuracy of classification and the area under the receiver operation characteristics (ROC) curve was used as a performance measure. We also developed other prediction models based on popular classification algorithms, namely neural networks, decision trees and logistic regression and these are explored in the paper. The results show that SVMs are a highly successful classifier and perform better than the other techniques with performance ranging between 94.1% and 96.3% accuracy and 81.3% and 97.5% area under the ROC curve.

Key words:

• machine learning
• HIV
• drug resistance
• support vector machines