Metagenomic surveillance of viruses at the human-animal interface
Zoonotic viruses are a major contributor to emerging infectious diseases, and continuously burden public health systems. Early detection and effective response to viral emergence require an overview of what viruses are circulating in animal hosts, which of these can and do infect at-risk human populations, and which pose the greatest risk of further spread. However, knowledge of such epidemiological patterns is generally biased towards known pathogens of humans and of economically important livestock species. With metagenomic sequencing, one can begin to address these biases by generating a more representative picture of what viruses are present in different host species living in a shared environment. Vietnam is considered a high-risk setting for the emergence of zoonoses, due to its high population and livestock densities and the prevalence of socio-cultural practices involving frequent close contacts between humans, livestock and wildlife. The Vietnam Initiative on Zoonotic Infections (VIZIONS) was established to improve our understanding of zoonotic emergence in this context. Over 2000 faecal samples and rectal swabs were collected from humans and a variety of farmed animals, and subjected to metagenomic sequencing. In this thesis, I use viral taxonomic classification methods to identify and characterise the viruses present in these samples. I investigate any signals for (putative) zoonotic viruses, and assess whether they could represent emerging public health threats. I also evaluate the roles and challenges of metagenomic surveillance for emerging viruses at the human-animal interface. The first part of this thesis focuses on the development and testing of a viral taxonomic classification pipeline. I describe the basic steps of this pipeline, and the rationale behind the chosen methods. Next, I test the pipeline on a subset of samples and viruses for which diagnostic quantitative PCR (qPCR) data were available for comparison. Receiver operating characteristic (ROC) curve analysis showed that the pipeline accurately distinguishes qPCR positive from qPCR-negative samples, and read pair counts correlate well with qPCR cycle threshold values. Investigation of samples with discordant qPCR and metagenomic results indicated that taxonomic misclassification by the pipeline plays a minor role in these discrepancies. Additionally, I found that, for each of the tested viruses, negative samples have variable read pair counts (“background noise”) that correlate with the total number of read pairs assigned to the virus across all samples of the same sequencing run. I hypothesise that this is due to “index switching”, a form of cross-contamination, and model the association. The findings of these investigations allow me to incorporate additional steps into the pipeline to counteract misclassification, and to use signal thresholds that take into account the effect of index switching cross-contamination. In the second part of this thesis, I focus on the characterisation of viruses identified with the taxonomic classification pipeline. I present an overview of the mammalian viruses found in samples from humans, swine and rats from Dong Thap province. After removing likely contaminants, I categorize the remaining viruses according to their zoonotic potential. Seven of these viruses are known or generally presumed to be zoonotic; three are only found in the animal study populations, but four – Rotavirus A, Picobirnavirus, Human associated cyclovirus 8, and Mammalian orthoreovirus – are shared between human and animal populations. Comparison of signals suggests that viral chatter (Rotavirus A) and cross-species transmission within a more generalist ecology (Picobirnavirus, Human associated cyclovirus 8) are plausible in this setting. Additionally, three putative novel zoonoses are identified, but knowledge gaps hinder extensive interpretation. I evaluate the relevance of these 10 zoonotic and putative novel zoonotic viruses as potential emerging public health threats, and highlight the knowledge gaps that need to be addressed before the risks of these viruses can be properly assessed. Finally, I interpret my findings in the general context of disease emergence, and evaluate the roles and challenges of viral metagenomics as a tool in the surveillance for emerging infectious disease.