IMI COMBINE Preclinical Bacterial Strain Repository
The IMI COMBINE Preclinical Bacterial Strain Repository consists of well-characterised and virulent strains of Gram-negative bacteria Klebsiella pneumoniae and Pseudomonas aeruginosa. The repository is part of a collection of reference materials hosted on the website of the Paul-Ehrlich-Institut, the German Federal Institute for Vaccines and Biomedicines. The strains have been deposited to the German Collection of Microorganisms and Cell Cultures (DSMZ), where they can be ordered for distribution.
The strains have been characterised and successfully evaluated in the COMBINE pneumonia model and offers a resource for the reproducible and comparable evaluation of the efficacy of new antibiotics.
COMBINE Pneumonia Model
The COMBINE pneumonia model is a standardised, reproducible and well-characterised mouse model to evaluate the efficacy of new small molecule antibiotics. The protocol has been developed based on an extensive literature review and stakeholder discussions. The model standardises key methodological procedures that can influence study outcomes, such as bacterial isolates used; sex and breed of mice, immunosuppression procedures, and inoculation techniques.
The COMBINE pneumonia model has been published in the ASM Journal “Microbiology Spectrum”. It was developed using a consensus lung infection protocol, where researchers at Statens Serum Institut (SSI) in Denmark, GSK in the USA and the Paul-Ehrlich-Institut in Germany tested Klebsiella pneumoniae and Pseudomonas aeruginosa isolates against predefined virulence criteria and followed up by independent sites to confirm virulence with minimal variability in bacterial growth. A total of 7 isolates is now validated in the model and are available through the German Collection of Microorganisms and Cell Cultures GmbH (DSMZ) through the IMI COMBINE Preclinical Bacterial Strain Repository (PBSR).
Read the publication on the COMBINE pneumonia model: The COMBINE pneumonia model: a multicenter study to standardize a mouse pneumonia model with Pseudomonas aeruginosa and Klebsiella pneumoniae for antibiotic development
Learn more about the literature review on which the COMBINE pneumonia model is based on: Variability of murine bacterial pneumonia models used to evaluate antimicrobial agents
Here you find the summary of the stakeholder discussion: Expert workshop summary: Advancing toward a standardized murine model to evaluate treatments for antimicrobial resistance lung infections
Bioassay Protocol Ontology (BPO)
A new bioassay protocol ontology enables the conversion of unstructured bioassay protocol data into structured, machine-readable formats that promote reusability and allow an accurate description of in vivo efficacy study metadata for antibiotic agents.
To enhance the reproducibility and usability of bioassays in antimicrobial resistance (AMR) associated antibacterial drug discovery and development, there is an increasing need for standardisation of bioassay metadata into machine-readable formats. Such a standardisation process requires mapping bioassay data to standard ontologies and can be performed at the study result output and protocol levels. At the result output level, general ontologies such as the BioAssay Ontology (BAO) already exist, but antibacterial drug discovery and AMR-specific ontologies that can aid in the standardisation process at the protocol level are missing.
The development of BPO is a first attempt to standardise and organise information within in-vivo AMR-related drug discovery bioassay protocol data in a structured manner. BPO will allow researchers to capture information regarding experimental details such as the type of mouse model, the bacterial strain, and the sex and growth phases of mouse and bacteria respectively from the protocol.
Find out more and download the ontology from GitHub
FAIR Data Templates
A tailored application of a “FAIRification framework” facilitating the practical implementation of FAIR principles. By showcasing the feasibility and benefits of transitioning to FAIR data practices, this work aims to encourage broader adoption and integration of FAIR principles within a research lab setting. Tools for standardisation of data sets to allow for FAIR working and data reuse.
Download the Data Survey, the Lab Data templates, the Data dictionary, and the FAIR assessments from Zenodo
Machine Learning Model to Identify New Antibiotics (AntiMicrobial-KG)
Advanced computational methods and machine learning could help reduce the high costs and complexity of antibiotic drug discovery. The COMBINE project has developed an Antimicrobial Knowledge Graph with models that can scan compound libraries to identify new antimicrobial compounds. The database and a machine learning model built using the knowledge are now published in the Journal of Chemical Information and Modeling. The model is trained on the AntiMicrobial-KG database, representing an aggregation of public bioassay datasets in a FAIR-compliant format. It has generated the largest training set hitherto used for ML applied to antimicrobial activity. However, both model and code can be trained on external datasets that drug developers already have access to, and to expand the applicability and confidence of model predictions for research and development.
The AntiMicrobial-KG was developed within the framework of the Innovative Medicines Initiative (IMI) AMR Accelerator program’s Scientific Interest Group on Machine Learning, coordinated by the COMBINE project. The AntiMicrobial-KG is a repository for collecting and visualizing public in-vitro antimicrobial assays. Utilizing this data, the AMR Accelerator projects have built ML models to efficiently scan compound libraries to identify compounds with the potential to exhibit antimicrobial activity. Using Random Forest and XGBoost algorithms, Antimicrobial-KG has developed classification models on four classes of microorganisms (i.e. gram-positive, gram-negative, acid-fast, and fungi) that outperform existing models. The ML model was tested on the EU-OPENSCREEN screening library to demonstrate its applicability in a laboratory setting. Antimicrobial-KG uses Python scripts for model training, exploratory analysis, and KG generation, available from GitHub. The data collected with the AntiMicrobial-KG and models are available on Zenodo. The AntiMicrobial-KG website at SciLifeLab allows users to search the database and use the pre-trained models for compound activity prediction.
Explore the experimentally validated antibacterial chemicals on the AMR-KG Database.
Read more about the source code and data repository on GitHub.
Find out more about the ML-model: Predicting Antimicrobial Class Specificity of Small Molecules Using Machine Learning
Recommendations
Recurring issues and mitigation strategies for the (clinical) development of vaccines and monoclonal antibodies against ESKAPE infections.
Coming in 2026.
