ABSTRACT HIV-1 under antiretroviral treatment selects for genetically-linked mutations that are correlated due to constraints on protein structural stability and function, which contribute to fitness. Project 5 studies are concerned with analyzing pairs (or higher-order) patterns of antiretroviral resistance mutations and their combined biophysical, biochemical, and structural effects on drug-resistance and viral fitness. During the past funding period, new statistical methods were developed to identify correlative mutational patterns present in genetically unlinked Gag and protease deep sequencing data. Potts Hamiltonian probabilistic models were constructed from protease sequence alignments to identify mutational patterns that lead to drug-resistance. To extend the past findings, it is proposed to identify genetically-linked patterns of antiretroviral resistance mutations from full-length, individual viruses from clade B or non-clade B HIV-infected patients during antiretroviral treatment. To investigate structural constraints in HIV proteins that influence selection of resistance mutations, Potts models of protein sequence covariation will be developed utilizing sequence and structural data. The combination of a novel full-length sequencing approach and virology expertise by Torbett will be complemented by bioinformatics and modeling expertise of Levy to serve the following specific aims: 1) Identify genetically-linked drug-resistance mutations (pairs or higher order) from HIV in longitudinal patient samples utilizing Multi-read Barcode-Assisted Single Molecule Sequencing (MrBASMS). Covariant mutations will be functionally and structurally characterized using previously described biochemical, biophysical and virological assays to validate their role in the rise of drug resistance. 2) Both full-length, from 1), and HIV sequence data from databases and structural information will be utilized to construct Potts models of drug naïve and drug-experienced protease, reverse transcriptase, integrase and Gag. Potts models will be used to investigate the effects of epistatic mutational combinations on fitness, as well as predict HIV protein residues at risk for drug-resistance mutation development. These studies will provide critical insight into HIV genetic barriers that must be overcome to develop resistance to multiple inhibitor combinations. The MrBASMS sequencing of HIV quasispecies from longitudinal patient samples will be led by Torbett and Sarafianos, along with outside collaborator Routh (UTMB). The biochemical, structural and virological validation of mutational covariants will be led by Torbett, Sarafianos and Levy, along with assistance from Core 2. Levy will develop Potts models from HIV sequence data and protein structural information obtained from Projects 1, 2, and Core 1.