MELD4AMR
Molecular Embeddings via Low-Distortions Framework for Antimicrobial Resistance
MELD4AMR is a German–Serbian research project developing compact, interpretable, and privacy-aware molecular representations for antimicrobial resistance (AMR) research. We aim to make complex molecular data easier to analyse while retaining information relevant to biological activity. By bringing together expertise in artificial intelligence, mathematics, and computational biology, we seek to support antimicrobial discovery and public health research.
Efficient Molecular Embeddings
Developing compact molecular representations that aim to preserve meaningful relationships while reducing the computational demands of large-scale analysis.
Interpretability & Explainability
Investigating how molecular representations can retain biologically relevant information and provide insights that researchers can understand and evaluate.
Privacy-Aware Collaboration
Exploring privacy-preserving methods for molecular data analysis to support responsible research and collaboration across institutions and national borders.
Hattab, G., Anžel, A., Spänig, S., Neumann, N., & Heider, D. (01 2023). A parametric approach for molecular encodings using multilevel atomic neighborhoods applied to peptide classification. NAR Genomics and Bioinformatics, 5(1). https://doi.org/10.1093/nargab/lqac103
Weckbecker, M., Anžel, A., Yang, Z., & Hattab, G. (2024). Interpretable Molecular Encodings and Representations for Machine Learning Tasks. Computational and Structural Biotechnology Journal, 23, 2326–2336. https://doi.org/10.1016/j.csbj.2024.05.035
Dr. Aleksandar Anžel and PD Dr. Georges Hattab initiated the project concept, establishing the foundation for a collaborative research effort. They brought together Nevena Ćirić, Prof. Dr. Jovana Kovačević, and Dr. Nevena Veljković to contribute their complementary expertise and strengthen the scientific collaboration.
The official MELD4AMR GitHub repository and project website were established to provide a central platform for sharing project information, research outputs, software, documentation, and related resources. The website serves as the public home for the project, while the repository supports transparent and collaborative development. Visit the GitHub repository and the project website.
MELD4AMR brings together researchers from Germany and Serbia to develop efficient, interpretable, and privacy-aware molecular representations for antimicrobial resistance research. The project builds on established research in molecular encoding and machine learning to advance methods that support reliable, transparent, and collaborative scientific discovery.
The team is actively implementing the project objectives, translating the research concept into practical methods and tools. Through coordinated efforts across the participating institutions, the researchers are progressively developing and evaluating approaches to molecular representation that address the challenges of antimicrobial resistance research.
Publications
Explore the key publications of the MELD4AMR project.
First Foundational Publication
Second Foundational Publication
Project Lead · Germany
ZKI-PH, Robert Koch Institute
Project Lead · Serbia
Faculty of Mathematics, University of Belgrade
Scientific Mentorship · Serbia
Faculty of Mathematics, University of Belgrade
Scientific Oversight · Germany
ZKI-PH, Robert Koch Institute / Freie Universität Berlin
Industry Advisor · Serbia
Genial D.O.O. / Vinča Institute of Nuclear Sciences
Our team combines molecular representation research, mathematical foundations, computational biology, and industry expertise.
MELD4AMR is delivered through German–Serbian research collaboration and funding support.
