Combining Classifiers for HIV-1 Drug Resistance Prediction

Abstract

This paper applies and studies the behavior of three learning algorithms, i.e. the Support Vector machine (SVM), the Radial Basis Function Network (the RBF network), and k-Nearest Neighbor (k-NN) for predicting HIV-1 drug resistance from genotype data. In addition, a new algorithm for classifier combination is proposed. The results of comparing the predictive performance of three learning algorithms show that, SVM yields the highest average accuracy, the RBF network gives the highest sensitivity, and k-NN yields the best in specificity. Finally, the comparison of the predictive performance of the composite classifier with three learning algorithms demonstrates that the proposed composite classifier provides the highest average accuracy.

A multifaceted analysis of HIV-1 protease multidrug resistance phenotypes

“…Finally, our reported classification accuracies are lower than those reported for genotype-based predictions, but this is partly because we use five categories as opposed to the binary or 3-way classifications commonly used. If we adopt a naive binary classification scheme (scaled resistance < 1.0 is not resistant; scaled resistance >= 1.0 is resistant), our cluster-based classification accuracies using the n-fold cross validation procedure for the entire data set range from 85%-95% excluding TPV(79%), compared with 85%-95% for binary classification schemes reported in the literature [65,72,74] (TPV and DRV were not part of these studies). It is interesting to note that while not the major goal of our paper, we have shown that with the exception of TPV, it may be possible to approach comparable drug resistance prediction accuracy without any genotypic information; this level of accuracy demonstrates the restricted phenotypic space occupied by the virus.…”

“…Many approaches have been used to create prediction models, including regression-based methods[26,64-69], decision trees[70], and other machine learning methods, including artificial neural networks, support vector machines, and others[67,71-74]. Several studies have also comparatively evaluated or combined methods to improve accuracy[67,72,73,75]. Models have also been created for predicting drug resistance phenotype[76] and virological success or failure[77-80] resulting from combination therapies.…”

Computational Mutation Scanning and Drug Resistance Mechanisms of HIV-1 Protease Inhibitors

“…Technically, using the CMS method for a mutation, one only needs to deal with the topology and coordinate files of the Amber program. We note that other scanning methods, such as the CAS and Fluorine Scanning, resulted in correct conformational sampling of protein mutants 78,79…”

“…We note that other scanning methods, such as the CAS and fluorine scanning, resulted in correct conformational sampling of protein mutants. 78,79 Generally speaking, the resistance mechanism of the target mutation can be divided into six groups in the view of thermodynamic rules as shown in Figure 3: decrease in the enthalpy contribution to the binding affinity (A-type), decrease in the entropic contribution to the binding affinity (B-type), decrease in both the enthalpy and entropic contributions (Ctype), no significant change in the enthalpy and entropic contribution (D-type), decrease in the enthalpy contribution compensated with increase in the entropic contribution (E-type), and decrease in the entropic contribution compensated with increase in the enthalpy contribution (F-type). The first three groups (A-, B-, and C-types) always lead to a high level of resistance, whereas the last three groups (D-, E-, and F-types) always lead to no resistance or low resistance.…”

Machine learning approaches to study HIV/AIDS infection: A Review

“…Some of the literatures are regarding of drug development, anti-viral agents development (Kirchmair et al 2011), antiretroviral response prediction (Zazzi et al 2012Prosperi 2011and Prosperi et al 2009), antiretroviral resistance prediction (Zazzi 2016Riemenschneider 2016aRiemenschneider 2016bHeider et al 2013and Kijsirikul 2008, antiretroviral adverse effects prediction (Adrover et al 2015) has been analyzed which were used machine learning approaches, SVM, Expert's Rules and Linear and Non-linear statistical learning algorithms, Radial basis function networks (RBF networks), k-nearest neighbor (kNN) and Virtual screening method to improve their result in Austria, Italy, USA, Germany and Thailand. Prediction of antibody of HIV epitope networks using neutralization titers and a novel computational methods or a simple machine learning methods has been done in USA (Evans et al 2014Hepler et al 2014and Choi et al 2015 Prediction of HIV-1 RT associated RNase H inhibition (Poongavanam 2013) shown good enrichment (80-90%) by receptor-based fl exible docking experiments compared to ligand-based approaches such as FLAP (74%), shape similarity (75%) and random forest (72%) in Denmark.…”