Towards a comprehensive landscape and mechanism understanding of virus-disease associations

Viruses are linked to human diseases through complex molecular mechanisms, yet a comprehensive investigation of virus-disease associations and their molecular basis remains lacking. We curated 3,555 high-confidence virus-disease associations involving 267 viruses and 954 diseases through text-mining over 28 million PubMed abstracts followed by rigorous manual verification, establishing the most comprehensive resource in this field to date. We then developed AVD, a user-friendly online database to store, visualize, and analyze these associations and their molecular mechanisms. Our analysis revealed that virus-disease association patterns are significantly influenced by viral taxonomy, phenotype, and key biological traits such as transmissibility, transmission routes, and tissue tropism. Molecular analysis identified a core set of bridging genes involved in the majority of virus-disease associations. Notably, these bridging genes were extensively involved in viral infections and diseases, and exhibit distinct characteristics, including a high proportion of secreted proteins, elevated expression across human tissues, and high centrality in the human protein-protein interaction network. Furthermore, we developed a machine-learning model capable of accurately predicting virus-disease associations. Collectively, this study provides a valuable data resource and analytical framework for studying virus-disease relationships and offers novel insights into their molecular mechanisms. In this manuscript the authors analyze viral-disease associations and identified that virus-disease associations are significantly influenced by viral taxonomic, phenotype, and their biological traits. Genes that may mediate the viral-disease associations are also discovered. A database called AVD (Association of Virus and human Diseases) is also constructed to facilitate the storage and convenient analysis of the associations.