Skip to main content
ARS Home » Plains Area » Clay Center, Nebraska » U.S. Meat Animal Research Center » Meat Safety and Quality » Research » Publications at this Location » Publication #330575
Title: Support vector machine applied to predict the zoonotic potential of E. coli O157 cattle isolates

Author
item LUPOLOVA, NADEJDA - Roslin Institute
item DALLMAN, TIMOTHY - Public Health Agency
item MATTHEWS, LOUISE - University Of Glasgow
item Bono, James - Jim
item GALLY, DAVID - Roslin Institute

Submitted to: Proceedings of the National Academy of Sciences (PNAS)
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/2/2016
Publication Date: 10/4/2016
Publication URL: https://handle.nal.usda.gov/10113/5695356
Citation: Lupolova, N., Dallman, T.J., Matthews, L., Bono, J.L., Gally, D.L. 2016. Support vector machine applied to predict the zoonotic potential of E. coli O157 cattle isolates. Proceedings of the National Academy of Sciences. 113(40):11312-11317. doi:10.1073/pnas.1606567113.

Interpretive Summary: Infections with Escherichia coli O157:H7 originating from animals have emerged as a serious threat to human health in the last three decades. Conventional genome sequence-based analyses predict that the majority of E. coli O157 strains from cattle can cause infections in humans, especially if they belong to one of two established groups. In this study, we used genome sequence with computer modeling to successfully predict the human infection potential of cattle E. coli O157 isolates. We demonstrate that only a small subset of strains (<10%) from the bovine reservoir are likely to cause human disease, even within the previously defined groups. This approach was tested across E. coli O157 isolates from both the UK and USA. Also, we validated this approach using isolates from previous outbreaks to correctly classify those isolates responsible for causing disease from all the isolates recovered during the investigation. This finding has potentially important implications for public health as it means that cattle infected only with these virulent strains could be targeted for control. This approach also should be broadly applicable to other disease-causing bacteria passed from animals to humans.

Technical Abstract: Methods based on sequence data analysis facilitate the tracking of disease outbreaks, allow relationships between strains to be reconstructed and virulence factors to be identified. However, these methods are used postfactum after an outbreak has happened. Here, we show that support vector machine analysis of bovine E. coli O157 isolate sequences can be applied to accurately predict their zoonotic potential including identification of the cattle strains most likely to be a serious threat to human health. Notably, our analyses indicate that only a minor subset (less than 10 percent) of bovine E. coli O157 isolates analysed in our datasets have the potential to cause human disease; this is despite the fact that the majority are within previously defined pathogenic lineages I or I/II and encode key virulence factors. The major differences between human and bovine E. coli O157 isolates were due in the relative abundances of hundreds of predicted prophage proteins. This finding has profound implications for public health management of disease, as more accurate prediction of pathogenic potential can focus preventative measures and means that vaccination or targeted eradication may now be a more realistic option for control of zoonotic EHEC O157 in the reservoir host.