# Exploring genetic signatures of zoonotic influenza A virus at the swine–human interface with phylogenetic and ancestral sequence reconstruction

## TL;DR

This study identifies genetic markers in influenza A viruses that may help predict zoonotic transmission between swine and humans.

## Contribution

The paper introduces a novel approach combining phylogenetic analysis and machine learning to identify genetic signatures linked to interspecies transmission of IAV.

## Key findings

- Specific protein regions and amino acid positions in internal gene segments are more important for interspecies transmission.
- Analyses revealed complex mutational patterns within and across viral proteins associated with host adaptation.
- The study suggests potential genetic signatures for early-warning genomic surveillance systems.

## Abstract

Influenza A viruses (IAVs) in swine have zoonotic potential and pose a continuous threat of causing human pandemics, as demonstrated by the H1N1 pandemic in 2009. Despite increased genomic surveillance, we have limited knowledge of the IAV evolutionary dynamics leading to such zoonotic events and no clear understanding of genetic markers associated with interspecies transmission of IAV between humans and swine. To explore this, we analysed a comprehensive publicly available whole-genome dataset of human and swine IAV sequences. We conducted phylogenetic analyses and inference of ancestral host and sequence states for each IAV segment to map inferred mutations associated with hypothetical representative transmissions within and between swine and human hosts. We developed a custom python library to combine information from host and ancestral sequence annotated trees and applied statistical models to identify genetic markers associated with intra- or interspecies transmissions between swine and humans. This included analysing mutation rates and the selective pressures acting on the viral proteins following intra- and interspecies transmissions and using a scalable, gradient-boosted decision tree machine learning approach to predict key amino acid positions critical for different transmission types. Our analyses not only indicated complex mutational patterns within and across viral proteins, but also suggested that specific protein regions and amino acid positions of especially several of the internal gene segments were more important for interspecies transmission. Our findings identify potential genetic signatures across the IAV proteins associated with host adaptation and zoonotic potential, offering valuable markers for early-warning genomic surveillance systems to enhance animal health and minimize the potential for zoonotic transmission of IAV.