Center for Genomic Epidemiology

Info: Change your bookmarks! We have a new base URL that is more secure and shorter - https://genepi.dk. Here you will also find our new landingpage, listing all available services and the status of our servers.

PathogenFinder 2

Welcome to the web application PathogenFinder2.

PathogenFinder2 is a novel deep learning model able to predict pathogenic capacity on humans from bacteria, only considering its genome. PathogenFinder2 is also able to report the proteins that has mattered the most for the prediction, as well as report the embeddings that can locate a bacterial genome in the Pathogenic Bacteria Landscape.
For more a more detailed information, please consider the article "Whole-genome prediction of bacterial pathogenic capacity on novel bacteria using protein language models, with PathogenFinder2".

BETA version: Note that the current version of PathogenFinder 2 is still in beta and you may encounter issues. Please let us now in via the contact page if you encounter any issues.

Known issues:

* Users upload genes or plasmids, instead of whole genomes, which the service expects and this causes the job to crash.
* Users upload non bacterial genomes, which might have more cds than bacteria, causing the job to crash.

Input data type:

Run extra phenotyping analysis:

This will delay the results notably.

Upload and submit job:

Citations

If you use and/or publish results obtained by the service, please cite the article below.

 

  • Ferrer Florensa, A., Almagro Armenteros, J. J., Kaas, R. S., Clausen, P. T., Nielsen, H., Rost, B., Aarestrup, F. M.
    (2025). Whole-genome prediction of bacterial pathogenic capacity on novel bacteria using protein language models, with PathogenFinder2. bioRxiv, 2025-04.

Supported by