Predicting mammalian hosts in which novel coronaviruses can be generated

Area of Science:

• Virology
• Zoonotic Diseases
• Computational Biology

Background:

• Novel pathogenic coronaviruses, including SARS-CoV and SARS-CoV-2, originate from homologous recombination between co-infecting viruses within a single host cell.
• Identifying hosts that harbor multiple coronaviruses is crucial for understanding the emergence of novel viruses, yet most coronavirus-host interactions remain uncharacterized.

Purpose of the Study:

• To predict which mammalian species serve as hosts for multiple coronaviruses.
• To identify potential reservoirs for the generation of novel coronaviruses, including SARS-CoV-2.

Main Methods:

• A meta-ensemble of similarity learners was employed, integrating viral, mammalian, and network perspectives.
• Predictive models were developed to identify coronavirus-host associations.

Main Results:

• The study predicts a significantly higher number of coronavirus-host associations (11.5-fold increase) than currently documented.
• Over 30-fold more potential hosts for SARS-CoV-2 recombination and over 40-fold more host species with diverse coronavirus subgenera were identified.
• These predictions hold true even at stringent probability cut-offs (>0.9821).

Conclusions:

• There is a substantial underestimation of the potential for novel coronavirus generation in both wild and domesticated animals.
• The findings underscore the importance of expanded coronavirus surveillance in identified high-risk animal species.