Beyond the Snapshot: Molecular Ensembles in Prebiotic Chemistry and Biophysics using Enhanced Sampling and Machine Learning
Guillaume Stirnemann (Ecole Normale Supérieure & CNRS)
Abstract:
Biological function and chemical reactivity emerge from ensembles of molecular configurations, their statistical populations, and the transitions between them. Understanding these processes therefore requires to go beyond what is typically provided by structural biology or quantum chemistry, i.e., more than static structures or a few energy calculations: it demands exploration and statistical sampling. I will illustrate this perspective through two complementary directions. In biophysics, enhanced sampling and deep learning reveal hidden conformational states that explain enzyme activity and ribozyme function. In chemistry, machine-learning interatomic potentials, developed through our ArcaNN active-learning framework, enable extensive and reliable exploration of reaction pathways and free-energy landscapes, exemplified by phosphoester bond formation in water, a key process in prebiotic chemistry. These examples highlight how combining accurate molecular interactions with sufficient sampling connects microscopic configurations to experimental observables.
About AMCOOL:
AMCOOL is a network of Amsterdam-based researchers (UvA and VU) with a shared interest in origins-of-life studies. As such, it is an interdisciplinary endeavor including fields such as biology, chemistry, astronomy and geoscience. The network meetings are opportunities for researchers from different fields to interact, share ideas, and collaborate. If you want to stay up-to-date on the work of AMCOOL, events and networking opportunities, then sign up to our newsletter on our webpage.
This seminar will take place in room C0.05. There will be drinks afterwards.