Computational Virology

Treatment of HIV infection is based on combinations of antiviral drugs. To identify the optimal drug combination for each patient, we develop computational methods based on the genetic composition of the virus population of the patient. We predict the probability that the virus will develop escape mutations that would result in drug resistance and eventually in treatment failure. This personalized medicine approach allows for defining drug combinations that are optimized for each individual patient.

Enlarged view:  Network of functional interactions among genes
Network of functional interactions among genes [from doi:10.1016/j.celrep.2018.03.102]

In order to estimate the genetic composition of an intra-​host virus population, we have developed several computational approaches for the analysis of ultra-​deep sequencing data. This challenging task, known as the viral quasispecies assembly problem, involves the assembly of all sequencing reads into an unknown number of unknown viral haplotype sequences.

Enlarged view: HIV-1 strains can be inferred by determining maximal cliques from next-generation sequencing data
Viral quasispecies assembly through clustering of paired-end reads [from doi:10.1371/journal.pcbi.1003515.g001]

In order to make ultra-deep sequencing of virus populations accessible to a broad range of biomedical users, we develop and actively maintain external page V-pipe, a bioinformatics pipeline for the analysis of viral high-throughput sequencing data. V-pipe supports the reproducible analysis of genomic diversity in intra-host virus populations, which is involved in viral pathogenesis and virulence, and enables diagnostic applications.

Selected references:

JavaScript has been disabled in your browser