Project start: 1.10.2025.

PepNanoDisco aims to advance the discovery of supramolecular peptide materials by integrating neural networks (NN), molecular dynamics (MD) simulations, and experimental approaches. The project builds on the results of our previous research and focuses on addressing key challenges in applying artificial intelligence (AI) to peptide properties predictions, such as the lack of integration across these fields, data reporting inconsistencies, and difficulties in reproducibility. Generative AI based on neural networks will be used to propose peptide sequences with high self-assembly potential, which will undergo validation through MD simulations and experimental characterization. This iterative process will refine AI models taking into account not only the sequence but also the context of self-assembly. Coarse-grain simulations will then be used to gain a better understanding of peptide aggregation and its key parameters. These results will be cross-validated with experimental measurements, such as circular dichroism, to create a feedback loop that refines both models and simulations. The most promising peptide candidates, identified by AI and refined through MD simulations, will be experimentally characterized under different conditions, enabling a deeper understanding of the self-assembly process in different environments and varying conditions to understand material formation. Ultimately, a framework will be established for discovering supramolecular peptide-based materials, providing experimentalists with smart exploration of the chemical space and more sustainable peptide material design. This approach will uncover new insights into the relationship between sequence, experimental conditions, and self-assembly, thereby improving the ability to predict and control peptide aggregation and enabling the discovery of new functional materials for applications in therapy or catalysis.