Researchers at CMS College, Kottayam, have developed a computational framework capable of predicting the transmission of viruses from animals to humans.

The model, named SPHAK (Sequence-based Prediction of Host spillover by Analysis of k-mers), analyses viral protein sequences to identify signals indicating a virus’s ability to cross species barriers. It examines short amino acid patterns, or k-mers, within viral proteins to detect host-linked signatures and generates a `spillover prediction’ (SP) score. The finding has been published in Nature Scientific Reports.

According to the researchers, SPHAK differs from traditional methods that rely on whole-genome analysis by focusing on viral proteins, which function as molecular “keys” interacting with host cell receptors. By working with the 20-amino-acid protein alphabet, the system captures subtle evolutionary changes that may indicate a virus adapting to a new host.

An SP score above 0.5 is used to flag viruses with a higher likelihood of cross-species transmission, enabling early prioritisation of samples for further study without the need for complete genome sequencing.

Over 97% accuracy

The model was tested on more than 950 animal virus species and achieved over 97% accuracy in identifying viral families. It also detected early warning signals consistent with known zoonotic sources, with viruses from birds, bats and pigs showing strong spillover potential. In validation tests, SPHAK identified all known human-infecting influenza strains with 100% sensitivity and flagged the bat-derived coronavirus HKU5 as a potential zoonotic threat.

The framework has also been extended to plant viruses to assess the risk of crop disease spread. Focusing on capsid proteins, which protect viral genetic material, the system was able to identify high-risk signatures that could help in early containment of plant infections.

The researchers say the tool could serve as an early warning system for both animal and plant disease outbreaks, helping prioritise surveillance and preventive action. The study was led by Vibin Ipe Thomas, Assistant Professor of Chemistry, and involved a team including Assistant Professor of Biotechnology Prisho Mariam Paul, student researchers Vinny N. G., Ananya Prakash, S. Kavya, and C. Rajalakshmi.