Regensburg 2025 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

BP: Fachverband Biologische Physik

BP 32: Computational Biophysics II

BP 32.12: Vortrag

Freitag, 21. März 2025, 12:30–12:45, H46

MolecuTas: an ML platform for refining quantum properties and bioactivity of complex molecules — •Vicente Domínguez Arca^1,2, Jannis Krüger², Álvaro Vallejo Bay³, Thomas Hellweg², and Luis Taboada Antelo¹ — ¹Biosystem and Bioprocesses Engineering, IIM-CSIC, Spain — ²Physical and Biophysical Chemistry, Bielefeld University, Germany — ³Applied Physics, University of Santiago de Compostela, Spain

The integration of machine learning (ML) and computational chemistry enables efficient prediction of quantum properties for complex molecules, crucial for advancing drug discovery and materials science. Our ML platform leverages Graph Convolutional Neural Networks (GCNNs) and the "sliding window" methodology to predict quantum mechanical parameters like partial atomic charges, overcoming traditional ab initio constraints. This approach scales to larger, biologically relevant molecules, enhancing molecular dynamics simulations and rational drug design.

Focusing on marine saponins -complex thalassochemicals with unique sulfated glycoside structures- our platform improves charge distribution predictions, enabling precise simulations of bioactive interactions. These advances highlight the therapeutic potential of marine saponins in oncology, lipid metabolism, and immune modulation. By applying our platform to marine saponins, this research bridges computational and experimental workflows, fostering the discovery of novel thalassochemicals for applications in functional foods, pharmaceuticals, and sustainability.

Keywords: MolecuTas; Quantum Property Prediction; Marine Saponins; Machine Learning in Drug Discovery; Thalassochemicals