Drug resistance mutations in HIV: new bioinformatics approaches and challenges

[Display omitted] •Machine learning is increasingly used to predict and understand drug resistance in HIV.•Phylogenetics helps to track the emergence and spread of HIV drug resistance mutations.•Deep sequencing accurately measures within-host HIV genetic diversity and low frequency resistant variants.•Confidentiality, securely and accessibly sharing of sequence data and associated metadata remain a challenge. Drug resistance mutations appear in HIV under treatment pressure. Resistant variants can be transmitted to treatment-naive individuals, which can lead to rapid virological failure and can limit treatment options. Consequently, quantifying the prevalence, emergence and transmission of drug resistance is critical to effectively treating patients and to shape health policies. We review recent bioinformatics developments and in particular describe: (1) the machine learning approaches intended to predict and explain the level of resistance of HIV variants from their sequence data; (2) the phylogenetic methods used to survey the emergence and dynamics of resistant HIV transmission clusters; (3) the impact of deep sequencing in studying within-host and between-host genetic diversity of HIV variants, notably regarding minority resistant variants.