A Unified Framework to Prioritize RNA Virus Cross-Species Transmission Risk Across an Expansive Host Landscape

Area of Science:

• Virology
• Genomics
• Epidemiology

Background:

• RNA viruses evolve rapidly, posing public health risks due to potential cross-species transmission.
• Identifying novel viruses with spillover potential is challenging with current prediction methods that lack precision.
• Existing methods predict hosts at broad taxonomic levels, limiting accurate risk assessment for zoonotic diseases.

Purpose of the Study:

• To develop a unified framework, UniVH, for accurate virus-host association prediction across diverse viral and host families.
• To improve the identification of novel RNA viruses with zoonotic potential.
• To enable robust prediction for data-scarce hosts and phylogenetically distant viruses.

Main Methods:

• Developed UniVH, a unified virus-host association prediction framework.
• Trained the model on a broad spectrum of 90 viral families and 240 host families.
• Validated UniVH performance on novel viruses discovered after 2020, comparing it to BLASTp homology approaches.

Main Results:

• UniVH achieved 71.2% host prediction accuracy for novel viruses, a 15.3% improvement over BLASTp.
• Identified viral structural genes and host immune/metabolism genes as key predictors.
• Predicted significant host-range expansion in 20 mammalian virus families, increasing zoonotic potential.

Conclusions:

• UniVH offers a significant methodological advance for evaluating RNA virus spillover risks.
• The framework enables more precise identification of potential zoonotic hosts and emerging viruses.
• UniVH's interpretability aids in understanding virus-host interactions and prioritizing future research.