Data proliferation, reconciliation, and synthesis in viral ecology

Abstract

The fields of viral ecology and evolution have rapidly expanded in the last two decades, driven by technological improvements, and motivated by efforts to discover potentially zoonotic wildlife viruses under the rubric of pandemic prevention. One consequence has been a massive proliferation of host-virus association data, which comprise the backbone of research in viral macroecology and zoonotic risk prediction. These data remain fragmented across numerous data portals and projects, each with their own scope, structure, and reporting standards. Here, we propose that synthesis of host-virus association data is a central challenge to improve our understanding of the global virome and develop foundational theory in viral ecology. To illustrate this, we build an open reconciled mammal-virus database from four key published datasets, applying a standardized taxonomy and metadata. We show that reconciling these datasets provides a substantially richer view of the mammal virome than that offered by any one individual database. We argue for a shift in best practice towards the incremental development and use of synthetic datasets in viral ecology research, both to improve comparability and replicability across studies, and to facilitate future efforts to use machine learning to predict the structure and dynamics of the global virome.

Mammal virus diversity estimates are unstable due to accelerating discovery effort

“…The full database contains host-virus association data up to and including 2021; however, novel association records become notably sparser after 2018 (Supp. Figure 1), likely due to delays between viral sampling and full reporting, including taxonomic assignment (6).…”

“…Such work frequently investigates the number of viruses known to infect a given host species (i.e. viral richness), based on synthetic datasets of known host-virus associations that are often compiled for this kind of hypothesis testing (4,6,7).…”

The Global Virome in One Network (VIRION): an atlas of vertebrate-virus associations

“…VIRION aggregates information from a total of five static sources (referred to as HP3, GMPD2, EID2, Shaw, and PREDICT) and two dynamic sources (Global Biotic Interactions and GenBank, hosted by the National Center for Biotechnology Information [NCBI]). The first four of these sources are separately reconciled in a new version of the CLOVER database (Gibb et al, 2021), which previously only covered mammalian viruses, but is here expanded to all vertebrate hosts; the rest of the data sources are integrated into the same data architecture. There is a high degree of redundancy among these different sources, both because they draw on the same limited body of scientific literature, and because they may directly query or incorporate each other (e.g., many of the viral genome sequences generated by the PREDICT program have been deposited in GenBank).…”

“…Moreover, recent evidence shows that data on the network structure of the global virome can be used to improve predictions of viruses with zoonotic potential based on genome composition (Poisot et al, 2021). Despite the obvious value of host-virus network data, currently available datasets are notably limited by the challenges of data integration and reconciliation; manual re-compilation and reconciliation is too inefficient to consistently keep these datasets up to date, and many sources are effectively a decade behind current knowledge (Gibb et al, 2021). Therefore, open, reproducible data that are as complete and detailed as possible are desperately needed.…”

Zoonotic Risk Technology Enters the Viral Emergence Toolkit

“…However, prior work anticipating the level of predictive resolution that exists today has also historically been subjected to similar skepticism. As these technologies become powered by growing datasets cataloguing the global virome [2,[66][67][68][69][70], and more complex microbiological predictors that capture host-virus interactions, it is difficult to imagine today how accurate and valuable they might become. If their potential for global health manifests, we should prepare now to guard against potential misuse, including monopolization in high income countries, and to anticipate important matters of equity, including the equitable sharing of the benefits arising from their use.…”