Data-driven classification of Escherichia coli using protein language model ascertains O-serotype determining genes

2026-03-19 · Source: Machine learning : nature.com subject feeds · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Bioinformatics · Depth: Advanced, medium

Summary

A new data-driven approach utilizes a Protein Language Model (PLM) to classify *Escherichia coli* O serotypes, addressing limitations of traditional methods that rely on predefined gene databases. Researchers applied the ESM-2 model to encode protein sequences into vector representations and trained a machine learning classifier to predict serotypes from genomic data. Analyzing 11,272 *E. coli* genomes, the study identified nine key marker genes—*wcaM*, *wcaL*, *wcaK*, *wzzE*, *wzxC*, *wecC*, *glmM*, *garR*, and *hisD*—that significantly contribute to O serotype classification. This PLM-based model, employing a Random Forest classifier, achieved 93% accuracy, outperforming traditional bioinformatics tools. It also demonstrated high recall for low-frequency serotypes, enhancing balanced performance and overall accuracy, offering a scalable framework for epidemiological surveillance.

Key takeaway

For AI scientists developing bacterial classification tools, this research indicates that integrating Protein Language Models like ESM-2 can significantly improve *E. coli* O serotyping accuracy and generalization, especially for novel or underrepresented variants. You should consider adopting PLM-based approaches to enhance high-throughput epidemiological surveillance and overcome the limitations of traditional reference-based methods, potentially accelerating vaccine development and clinical diagnostics.

Key insights

Protein Language Models can accurately classify *E. coli* O serotypes, improving upon traditional database-dependent methods.

Principles

• Data-driven models enhance serotype prediction.
• PLMs can generalize to novel bacterial variants.

Method

Encode protein sequences using ESM-2, then train a machine learning classifier (e.g., Random Forest) on genomic data to predict *E. coli* O serotypes, focusing on key marker genes.

In practice

• Use ESM-2 for protein sequence vectorization.
• Prioritize *wcaM*, *wcaL*, *wcaK* for O serotype analysis.

Topics

• Protein Language Models
• Escherichia coli Serotyping
• Genomic Classification
• ESM-2
• Bacterial Epidemiology

Best for: AI Scientist, AI Researcher, Data Scientist, Research Scientist

Related on AIssential

• Predicting phage–host specificity — Protein language models enhance phage–host specificity prediction by capturing biological information from viral sequences.
• Move over, AlphaFold: open-source model predicts shape of 1 billion proteins — ESMFold2, an open-source AI, predicts over a billion protein structures, outperforming AlphaFold3 and enabling novel protein design.
• Ensembles of Large Language Models for Identifying EQ-5D Studies in PubMed Based on Their Abstracts — Ensemble LLM frameworks reliably automate biomedical text classification, balancing performance and interpretability.
• 🔬ESM: The Bitter Lesson is Coming for Proteins - Alex Rives, BioHub — Scaling protein language models with vast, diverse data reveals emergent biological capabilities and enables programmable design.
• [R] Genomic Large Language Models — Genomic LLMs can identify functional similarities in DNA beyond sequence alignment, indicating learned regulatory patterns.
• Viral Proteins Reveal Geometry of Protein Language Models — Protein language models organize sequence embeddings by a "nativeness axis" while preserving specific biological group signals.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.