Quasispecies reconstruction based on vertex coloring algorithms

The viral quasispecies represent a set of related variants in a virus population (e.g. from an infected patient) that contain similar mutations due to the rapid and mutation-prone replications in viruses. The characterization of viral quasispecies in a highly divergent virus population is of great interest in biomedical research, in particular, to identify virulent and drug-resistant mutations in viral genomes for diagnosis of infectious diseases and targeted drug design. In recent years, next-generation sequencing (NGS) techniques have been widely used for deep sequencing of virus populations, in an attempt to characterize low abundant viral quasispecies containing specific mutations associated with virulence or drug-resistance. However, because of the short length of NGS reads, it remains a challenge to reconstruct viral quasispecies from NGS sequencing data. In this paper, we formulate the viral quasispecies reconstruction as the vertex coloring problem on a read conflict graph, and then apply heuristic algorithms to solve it. We compared our new algorithms with one existing software tool on three simulated datasets for HIV quasispecies reconstruction. The results showed our methods can improve the accuracy on the inference of the identities and quantities of viral quansispecies in a virus population.