Machine Learning Tool Detects Bacterial Invasion and DNA Damage in Human Cells

Researchers at the HUN-REN Biological Research Center, Szeged, and the Hungarian Center of Excellence for Molecular Medicine have developed MALVINA, a Machine Learning-Based Virulence Interaction Analysis tool. Published on October 1, 2026, in *Nature Communications*, MALVINA detects fluorescently labeled bacteria inside human cells and uses machine learning to analyze microscopy images.

This method allows scientists to measure, within the same experimental system, how efficiently bacteria enter cells, how strongly they accumulate, and what kind of DNA damage they induce. MALVINA provides cell-by-cell details, differentiating between scenarios where a few bacteria infect many cells versus many bacteria accumulating in fewer cells. Co-first author Bence Bognár noted its strength in observing and quantifying these individual cell differences.

Initial tests on four *Escherichia coli* strains revealed distinct virulence profiles, including a strain from a colorectal tumor showing strong invasive capacity and DNA damage. The study also found bacteria can reshape each other's behavior; an invasive *E. coli* promoted entry of a harmless strain, while a genotoxic strain reduced rivals' invasion. Szilvia Juhász, head of the Cancer Microbiome Group at HCEMM, highlighted MALVINA's precise picture of infection consequences.

What to watch: Further applications of MALVINA in microbiology and antibiotic research.

Editor's note: The draft is a comprehensive and accurate summary of the provided scientific article, capturing all key findings and methodology.

This article is AI-generated and fact-gated. Original reporting: Phys.org