У оквиру семинара Центра за изучавање комплексних система Института за физику у Београду, у четвртак, 19. децембра 2024. године у 14 часова у читаоници библиотеке „Др Драган Поповић“, др Стефан Миленковић (Универзитет у Каљарију, Италија) одржаће предавање:
Enhancing accumulation prediction in E. coli by Integrating MD simulations and interpretable ML methods
САЖЕТАК:
Understanding and predicting the accumulation of molecules in Escherichia coli is critical for optimizing the development of effective antimicrobial agents and bioengineering applications. The complexity of membrane permeability and intracellular transport demands approaches that combine biophysical rigor with computational efficiency. In this talk, we will present a novel workflow that integrates Molecular Dynamics (MD) simulations with interpretable Machine Learning (ML) models to predict molecular accumulation in E. coli.
Our approach leverages MD simulations to capture the dynamic behavior of molecules in physiological environments. This behavior is summarized in a set of statistical descriptors, which are systematically integrated into the training procedure to identify the optimal combination of descriptors. The significance of these statistical descriptors, including those with a priori relevance to the problem, was examined relative to conventional molecular descriptor models. Our results demonstrate the immense potential of MD simulation-generated data for enhancing ML-based modeling of microbiological properties.
