Systems biology provides a way to derive a holistic view of the role of the microbiome in health and disease. Such work relies on mapping the interaction between complex biological systems such as the bacterial microbiome network with other, less studied, components of the microbiome such as the virome.
These networks can be delineated using high-throughput and reductionist experimentation to define the nodes and edges (Fig. 1).
A deeper understanding of how the microbiome impacts human function requires an understanding of the function of the human virome network. It is interesting to note that, of the US$920 million invested in microbiome research from 2012 to 2014, only 3 % of research was dedicated to studies of viromes [4]. Extrapolating available viral diversity data leads to an estimate that there are approximately 320,000 mammalian viruses awaiting discovery [5]. This number is completely overshadowed by the estimates for the numbers of viruses that infect bacterial cells (bacteriophages), with estimates as high as 1031 bacteriophages on the planet, outnumbering their bacterial hosts at a 10:1 ratio [6]. In comparison to studies of the bacterial microbiome, the relative paucity of studies on viral diversity undermines our capacity to understand full biological networks.
Metagenomic analysis will continue to classify an extensive menagerie of both eukaryotic viruses and bacteriophages. This information will populate the nodes of the biological interaction network, but will be largely valueless for characterizing graph interactions. The generation of interaction networks will require classical experimental approaches, including mouse and tissue culture infection models for characterizing eukaryotic viruses, and bacterial host infection studies for bacteriophages. Both of these strategies are extremely challenging and rely on the isolation of pure virus and the availability of a susceptible model host.
The selection of a model host organism to study eukaryotic virus–host interactions is largely determined by the availability of laboratory models, with mouse being the most prevalent. The lack of replication in mice would severely hamper the study of virus–host interactions. Bacteriophage–host determination is equally challenging. Computational predictions of bacteriophage hosts are largely unsuccessful, with the capacity to assign 1 bacteriophage to 1–4 possible bacterial hosts in only 10–40 % of cases [7]. While computational methods will certainly mature, mechanistic studies also require the growth of individual bacteriophages in susceptible bacterial hosts. While challenging, these mechanistic studies will incrementally populate certain portions of the overall interaction network (virus–gene, bacteriophage–virus interactions).