A genotypic method for the identification of HIV-2 coreceptor usage - PowerPoint PPT Presentation

A genotypic method for the identification of HIV-2 coreceptor usage Matthias Döring Max Planck Institute for Informatics AREVIR meeting May 26, 2017

HIV-2 is prevalent in West Africa and Europe Ibe S, Sugiura W. Recombinant Forms of HIV-2. Encyclopedia of AIDS . 2014. May 26, 2017 2

HIV-2 has a milder course of infection than HIV-1 commons.wikimedia.org/wiki/File:Hiv-timecourse.png HIV-2 CD4 count Course of infection for HIV-1 Course of infection for HIV-2 HIV-2 viral load May 26, 2017 3

Coreceptors are necessary for HIV cell entry Maraviroc Engelman and Cherepanov. The structural biology of HIV-1: mechanistic and therapeutic insights. Nature Review Microbiology . 2012; 10, 279-290. May 26, 2017 4

Why treat HIV-2 with coreceptor antagonists? PIs INI NNRTIs CCR5 Antagonist NRTI + NRTIs NNRTI FI May 26, 2017 5

Determining coreceptor usage before presccribing X4-capable R5 Courtesy of Nico Pfeifer May 26, 2017 6

The advantages of genotypic approaches • Interpretable results • Cost efficiency • Reduced technological bias • No standardized assay for HIV-2 May 26, 2017 7

Examples of technological bias Identifier of No. of X4- No. of R5 V3 loop of the X4-capable sequence Decision X4-capable capable isolates Isolate isolates DQ870430 1 21 CKRPGNKTVVPITLMSGLVFHSQPINKRPRQAWC R5 NARI-12 1 5 CKRPGNKTVLPITLMSGLVFHSQPINTRPRQAWC R5 3 Exclude GU204945 1 CKRPGNKTVRPITLLSGRRFHSQVYTVNPKQAWC 310248 1 1 CRRPGNKTVVPITLMSGLVFHSQPINKRPRQAWC X4-capable HIV-2 samples with discordant annotations of coreceptor usage but identical V3 loops May 26, 2017 8

Approach of geno2pheno[coreceptor-hiv2] May 26, 2017 9

Reproduction of known markers and novel markers X4-probabilities of X4-capable variants as predicted by Top-scoring (75% of total weight) features of the predictive model geno2pheno[coreceptor-hiv2] Visseaux et al. Molecular Determinants of HIV-2 R5-X4 Tropism in the V3 Loop: Development of a New Genotypic Tool. J Infect Dis . 2012; 205:111–120. May 26, 2017 10

Performance comparison to Visseaux et al. SVM Visseaux et al. Sensitivity 73.5% 85.3% Specificity 96% 94.0% 10-fold nested CV performance on the test data set (N = 84 ) No significant difference at 𝛽 = 0.05 P-value (McNemar’s test): 0.37 Why should you use our tool? Döring M, Borrego P, Büch J, Martins A, Friedrich G, Camacho RJ, et al. A genotypic method for determining HIV-2 coreceptor usage enables epidemiological studies and clinical decision support. Retrovirology. 2016;13:85. May 26, 2017 11

Validation on an independent test set Performance on nine novel HIV-2 samples • geno2pheno[coreceptor-hiv2]: 9/9 correct • Rules-based approach: 7/9 correct May 26, 2017 12

Interpretation for ROD10 (H18L + K29T) May 26, 2017 13

Web service Overview of results coreceptor-hiv2.geno2pheno.org CSV output Visualization May 26, 2017 14

Conclusions www.linkedin.com First web service for HIV-2 coreceptor prediction Tool enables epidemiological studies Features of HIV-2 coreceptor usage in the V3 loop • Discriminatory features occur at the end • Individual amino acids are highly predictive • Net charge is highly predictive May 26, 2017 15

Thank you for your attention and thanks to … Thomas Lengauer Nico Pfeifer Achim Büch Georg Friedrich Max Planck Institute Max Planck Institute Max Planck Institute Max Planck Institute for Informatics, for Informatics, for Informatics, for Informatics, Saarbrücken Saarbrücken Saarbrücken Saarbrücken Pedro Borrego Nuno Taveira Rolf Kaiser Ricardo Camacho Rega institute, University of Lisbon University of Lisbon Institute for Virology, KU Leuven University of Cologne May 26, 2017 16

Backup Slides May 26, 2017 17

Properties of HIV-2 Relation to HIV-1 Local prevalence Cpz: Chimpanzee albanydailystar.com High prevalence Adapted from Ibe S, Sugiura W Recombinant Forms of HIV-2. Low prevalence Encyclopedia of AIDS . 2014. Milder course of infection kpbs.org Reeves JD, Doms RW. MM: Sooty Human immunodeficiency virus type 2. mangabey J Gen Virol . 2002; 83:1253–1265. Marlink et al. Reduced rate of disease development after HIV-2 infection as compared to HIV-1. Science . 1994; 265:1587–90. May 26, 2017 18

Decline of CD4+ cell count in HIV-2 is slower N = 32 N = 31 Disease-free: CD4+ cell count ≥ 400 copies per 𝜈𝑚 Marlink et al. Reduced rate of disease development after HIV-2 infection as compared to HIV-1. Science . 1994; 265:1587–90. May 26, 2017 19

Viral coreceptor usage and maraviroc treatment X4-capable R5 • R5-viruses can use only CCR5 • X4-capable viruses can use CXCR4 www.aidsinfo.nih.gov Treatment with maraviroc www.aidsmap.com • R5-viruses are inhibited • X4-capable variants can still replicate www.aidsinfo.nih.gov May 26, 2017 20

Broad use of HIV-2 coreceptors in vitro van der Ende et al. Journal of General Virology . 2000 Coreceptor use of multiple isolates Coreceptor use in vivo “CCR5 and CXCR4 appear to be the major coreceptors for HIV-2 infection of PBMC.” — Mörner et al. AIDS Resarch and Human Retroviruses . 2002 Bron et al. Journal of Virology . 1997 Promiscuity of the HIV-2 ROD strain May 26, 2017 21

Clinical staging of HIV-infection HIV staging scheme from the CDC AIDS Primary Stage 1 Stage 2 Stage 3 Stage 4 Infection • Acute • Asymptomatic • Weight loss • Tuberculosis • HIV wasting syndrome • Asymptomatic • Lymphadenopa • Respiratory • Chronic thy infections diarrhea • Pneumocystis pneumonia WHO staging May 26, 2017 22

Interactions between HIV and cell receptors HIV surface Host membrane Delhalle et al. Phages and HIV-1: from display to interplay. Int J Mol Sci . 2012;13(4):4727-94 May 26, 2017 23

The clinical data pyramid Few data points Phenotypes Clinical data Genotypes Many data points May 26, 2017 24

Support vector machines: separable case May 26, 2017 26

Learning with support vector machines Maximize the margin: Separate the two classes: May 26, 2017 27

Learning with support vector machines Inseparable case: Cannot be fulfilled! May 26, 2017 28

Learning with support vector machines R5 Optimization problem Classification rule X4-capable May 26, 2017 29

SVM optimization problem Inseparable case with margin M Inseparable case with cost parameter C November 23, 2015 30

Kernel SVM function November 23, 2015 31

Nu-support vector classification Motivation for nu-SVM: interpretability 𝜉 is bounded: 𝜉 ∈ [0,1] • • Upper bound for the ratio of “errors” • Lower bound for the ratio of support vectors Schölkopf B, Smola AJ, Williamson RC, Bartlett PL. New Support Vector Algorithms. Neural Comput. 2000; 12:1207-1245. May 26, 2017 32

Contingency tables and other measures Reference X4-capable R5 X4-capable TP FP Prediction R5 FN TN Structure of confusion tables Specificity Sensitivity 𝐺𝐺 𝐺𝑈 𝑈𝑈 𝑈𝐺 + 𝐺𝑈 = 1 − 𝑈𝐺 + 𝐺𝑈 = 1 − 𝐺𝑈𝑈 𝑈𝑈𝑈 = 𝑈𝑈 + 𝐺𝐺 November 23, 2015 33

Mc Nemar’s Test Test the marginal homogeneity: 𝑞 𝑏 + 𝑞 𝑐 = 𝑞 𝑏 + 𝑞 𝑑 and 𝑞 𝑑 + 𝑞 𝑒 = 𝑞 𝑐 + 𝑞 𝑒 Visseaux et al. X4-capable R5 X4-capable a b SVM R5 c d Structure for McNemar’s test Test statistic Distribution Hypothesis 𝜓 2 = 𝑐 − 𝑑 2 𝐼 0 : 𝑞 𝑐 = 𝑞 𝑑 Chi-squared 𝑐 + 𝑑 𝐼 1 : 𝑞 𝑐 ≠ 𝑞 𝑑 distribution with 1 df Rejection of the null hypothesis The SVM does not predict very different labels from the approach by Visseaux et al. November 23, 2015 34

Construction of the learning data set Distribution of genotype-phenotype pairs Group Tropism Count A R5 61 A X4-capable 46 B R5 12 B X4-capable 5 D X4-capable 1 U R5 1 N=126 (74 R5, 52 X4-capable) May 26, 2017 35

AUCs of different kernel functions are similar Best-performing model Linear SVM AUC = 0.94 Interpretation Marginal role of higher- order interactions Results from 10 runs of 10-fold CV May 26, 2017 36

Distribution of transformed decision values X4-capable R5 Choosing an FPR-based classification cutoff Setting an FPR cutoff at 5% is reasonable! May 26, 2017 37