Computational discovery of viruses and their hosts

The virus discovery field has been revolutionised by high-throughput sequencing and metagenomic techniques. These approaches enable identification of previously unknown virus genomes with no requirement for virus isolation in cell culture. A key disadvantage of this however, is that in complex metagenomic samples containing nucleic acids from many cellular organisms, it is difficult to determine the specific hosts of novel viruses. The focuses of this thesis were thus on computational approaches to virus discovery in metagenomic data, and on developing techniques to subsequently identify hosts of viruses. Much of the work focused on single-stranded DNA viruses of the phylum Cressdnaviricota. A total of four novel virus families were proposed, namely Naryaviridae, Nenyaviridae, Vilyaviridae, and Draupnirviridae. Hosts of these four families were identified, along with those of four additional lineages from the literature (Redondoviridae, Kirkoviridae, pecoviruses, and hudisaviruses). Among the findings were several virus families that infect important protists of humans (e.g., Entamoeba histolytica, Giardia duodenalis, and Entamoeba gingivalis). Additionally, some members of the Draupnirviridae were found to infect vertebrates, and shown to have donated genes to pathogenic poxviruses of the genus Avipoxvirus. Furthermore, in a search for viruses with potential usage in virocontrol, a number of novel RNA viruses were detected in fungal isolates derived from cases of human infection.