Combination of whole genome sequencing and supervised machine learning provides unambiguous identification of eae-positive Shiga toxin-producing Escherichia coli - Original Research

Aug 5, 2026

Genome-wide supervised machine learning was used to build an unambiguous predictive model for detecting highly pathogenic STEC in E. coli read assemblies derived from complex samples that may contain multiple strains. The study accounts for E. coli genomic plasticity and uses STEC/E. coli pathogroup stratification by serotype and virulence factors to select biomarker combinations. From 1,493 curated genomes and 1,178 CDS, eight algorithms reduced features to six CDS. All eight models were trained with hyper-parameter tuning and cross-validation. Using only these six genes, EHEC can be clearly identified in silico mixtures and milk metagenomes, enabling marker genes for unambiguous EHEC characterization across diverse strain mixtures and raw milk metagenomes.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages13

Reading Time33 min

AccessFree to download

Introduction

Methods

Results and discussion

Keywords

TYPE Original Research PUBLISHED 12 May 2023 DOI 10. 3389/fmicb.2023.1118158 OPEN ACCESS EDITED BY Abani Kumar Pradhan, University of Maryland, College Park, United States REVIEWED BY Zachary R. Stromberg, Paciﬁc Northwest National Laboratory (DOE), United States Patrick Murigu Kamau Njage, Technical University of Denmark, Denmark *CORRESPONDENCE Fabien Vorimore [fabien.vorimore@anses.fr](fabien.vorimore@anses.fr) †These authors have contributed equally to this work RECEIVED 07 December 2022 ACCEPTED 21 April 2023 PUBLISHED 12 May 2023 CITATION Vorimore F, Jaudou S, Tran M-L, Richard H, Fach P and Delannoy S (2023) Combination of whole genome sequencing and supervised machine learning provides unambiguous identiﬁcation of eae-positive Shiga toxin-producing Escherichia coli. Front. Microbiol. 14:1118158 . doi: 10.3389/fmicb.2023.1118158 COPYRIGHT © 2023 Vorimore, Jaudou, Tran, Richard, Fachand Delannoy. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Combination of whole genome sequencing and supervised machine learning provides unambiguous identiﬁcation of eae-positive Shiga toxin-producing Escherichia coli Fabien Vorimore1*†, Sandra Jaudou1,2†, Mai-Lan Tran1,2 , Hugues Richard3 , Patrick Fach1,2 and Sabine Delannoy1,2 1ANSES, Laboratory for Food Safety, Genomics Platform IdentyPath, Maisons-Alfort, France, 2ANSES, Laboratory for Food Safety, COLiPATH Unit, Maisons-Alfort, France, 3 Bioinformatics Unit, Genome Competence Center (MF1), Robert Koch Institute, Berlin, Germany Introduction: The objective of this study was to develop, using a genome wide machine learning approach, an unambiguous model to predict the presence of highly pathogenic STEC in E. coli reads assemblies derived from complex samples containing potentially multiple E. coli strains. Our approach has taken into account the high genomic plasticity of E. coli and utilized the stratiﬁcation of STEC and E. coli pathogroups classiﬁcation based on the serotype and virulence factors to identify speciﬁc combinations of biomarkers for improved characterization of eae-positive STEC (also named EHEC for enterohemorrhagic E.coli) which are associated with bloody diarrhea and hemolytic uremic syndrome (HUS) in human. Methods: The Machine Learning (ML) approach was used in this study on a large curated dataset composed of 1,493 E. coli genome sequences and 1,178 Coding Sequences (CDS) . Feature selection has been performed using eight classiﬁcation algorithms, resulting in a reduction of the number of CDS to six. From this reduced dataset, the eight ML models were trained with hyper-parameter tuning and cross-validation steps. Results and discussion: It is remarkable that only using these six genes, EHEC can be clearly identiﬁed from E. coli read assemblies obtained from in silico mixturesand complex samples such as milk metagenomes. These various combinations of discriminative biomarkers can be implemented as novel marker genes for the unambiguous EHEC characterization from di􀀀erent E. coli strains mixtures as well as from raw milk metagenomes. KEYWORDS machine learning, Shiga toxin-producing Escherichia coli, food safety, metagenomics, raw milk 1. Introduction Shiga toxin-producing Escherichia coli (STEC) are important zoonotic pathogens comprising more than 400 serotypes (Beutin and Fach, 2015) . Pathogenic STEC strains such as enterohemorrhagic E. coli (EHEC) may cause hemorrhagic colitis (HC) and hemolyticuremic syndrome (HUS) in humans. However, it remains di􀀔cult to fully de􀀂ne human pathogenic STEC or i

FAQs

What goal does the study pursue for identifying STEC in complex samples?

The study develops a genome-wide supervised machine learning model to unambiguously predict the presence of highly pathogenic STEC in E. coli assemblies from complex, multi-strain samples.

How many genomes and coding sequences were used, and how were features selected?

The training used 1,493 curated E. coli genome sequences and 1,178 coding sequences (CDS). Feature selection across eight classification algorithms reduced the CDS to six.

What do the results show about EHEC identification using the selected biomarkers?

EHEC can be clearly identified using only the six selected genes in in silico mixtures and in complex samples such as milk metagenomes, supporting use of these marker combinations for unambiguous characterization.

Aug 5, 2026

Upload by Emma Wilson