Diversity and complexity of HIV-1 drug resistance: A bioinformatics approach to predicting phenotype from genotype

Abstract

Drug resistance testing has been shown to be beneficial for clinical management of HIV type 1 infected patients. Whereas phenotypic assays directly measure drug resistance, the commonly used genotypic assays provide only indirect evidence of drug resistance, the major challenge being the interpretation of the sequence information. We analyzed the significance of sequence variations in the protease and reverse transcriptase genes for drug resistance and derived models that predict phenotypic resistance from genotypes. For 14 antiretroviral drugs, both genotypic and phenotypic resistance data from 471 clinical isolates were analyzed with a machine learning approach. Information profiles were obtained that quantify the statistical significance of each sequence position for drug resistance. For the different drugs, patterns of varying complexity were observed, including between one and nine sequence positions with substantial information content. Based on these information profiles, decision tree classifiers were generated to identify genotypic patterns characteristic of resistance or susceptibility to the different drugs. We obtained concise and easily interpretable models to predict drug resistance from sequence information. The prediction quality of the models was assessed in leave-one-out experiments in terms of the prediction error. We found prediction errors of 9.6 -15.5% for all drugs except for zalcitabine, didanosine, and stavudine, with prediction errors between 25.4% and 32.0%. A prediction service is freely available at http:͞͞cartan.gmd.de͞geno2pheno.html.

Predicting Human Immunodeficiency Virus (HIV) Drug Resistance Using Recurrent Neural Networks

“…We compared the results with those published previously (Beerenwinkel, N. et al 2002;James, R. 2004). All results were averaged using 10-fold cross-validation.…”

“…The columns 1, 2 and 3 correspond to the results reported by James (2004) using KNN, the classic decision tree using ID3 and a variant of a decision tree developed respectively. The column 4 represents the results obtained by Beerenwinkel et al (2002) using a classic decision tree. *Note the results of classical decision trees by James are different than those of Beerenwinkel due to a different number of cases; Beerenwinkel et al used more cases to decision tree training.…”

“…In that case a value of 1 is given to the analyzed amino acid position and a 0 to all the others. Mutual information profiles have also been used to represent each sequence of the Protease enzyme (Beerenwinkel, N. et al 2002).…”

“…Wang and Larder (2003) used neural networks to predict resistance to the Protease inhibitor Lopinavir. Beerenwinkel et al (2002) used decision trees while James (2004) used decision trees and the k-nearest neighbor technique (KNN) to predict the resistance of protease inhibitors. Recently, convex optimization techniques have been used together with Least Absolute Shrinkage, and Selection Operator (LASSO) and Support Vector Machine (SVM) models (Rabinowitz, M. et al 2006) for regression or classification of the protease and reverse transcriptase resistance (see review by Cao et al (2005) for details).…”

Discovering human immunodeficiency virus mutational pathways using temporal Bayesian networks

“…A good part of these models is devoted to predicting phenotypic HIV resistance to antiretroviral drugs using different approaches such as decision trees [8] or neural networks [9]. Other models have tried to identify relevant associations between clinical variables and HIV disease [10].…”