Journal of Microbiology

Research Support, Non-U.S. Gov't

Improved Prediction of Coreceptor Usage and Phenotype of HIV-1 Based on Combined Features of V3 Loop Sequence Using Random Forest: Shungao Xu , Xinxiang Huang , Huaxi Xu , Chiyu Zhang; J. Microbiol. 2007;45(5):441-446.; DOI: https://doi.org/2592 [pii]

Abstract: HIV-1 coreceptor usage and phenotype mainly determined by V3 loop are associated with the disease progression of AIDS. Predicting HIV-1 coreceptor usage and phenotype facilitates the monitoring of R5-to-X4 switch and treatment decision-making. In this study, we employed random forest to predict HIV-1 biological phenotype, based on 37 random features of V3 loop. In comparison with PSSM method, our RF predictor obtained higher prediction accuracy (95.1% for coreceptor usage and 92.1% for phenotype), especially for non-B non-C HIV-1 subtypes (96.6% for coreceptor usage and 95.3% for phenotype). The net charge, polarity of V3 loop and five V3 sites are seven most important features for predicting HIV-1 coreceptor usage or phenotype. Among these features, V3 polarity and four V3 sites (22, 12, 18 and 13) are first reported to have high contribution to HIV-1 biological phenotype prediction.

Search