Berlin 2024 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 45: Modeling and Simulation of Soft Matter IV

CPP 45.1: Talk

Thursday, March 21, 2024, 15:00–15:15, H 0111

Multiscale simulations to understand pairing and stacking at the origin of life — •Laurie Stevens¹, Riccardo Martina², Alberta Ferrarini², and Marialore Sulpizi¹ — ¹Physics Department, Ruhr Universität Bochum, Germany — ²Chemical Science Department, Universita di Padova, Italy

Our research focus on life’s beginnings by examining nucleotide interactions, a critical point in RNA formation. Using Molecular Dynamics, we aim to understand how self-assembly of these components can promote the synthesis of long polynucleotides. However, due to the system’s complexity and scale, traditional ab initio methods are too slow for our targeted timescales of several nanoseconds.

To bypass these limitations, we turn to the Machine Learning approach, employing DeepMD. We train neural networks potentials (NNPs) to mimic the intricate behaviours of nucleotides. As a first step we consider a single AMP molecule in water, where we aim to reproduce the complex free energy landscape as function of the relevant degrees of freedom.

The trained NNPs are able to accurately reproduce the solvation structure around the different chemical groups, as well as the conformational changes associated to the torsional angles around the sugar and the phosphate groups.

Keywords: Machine Learning; Molecular Dynamics; Neural Network Potential; Nucleic Acids; Self-assembly