Machine Learning in Neurobiological Systems
MACHINE LEARNING IN NEUROBIOLOGICAL SYSTEMS
LEAD PI (S): DRS. Paolo Carloni & MICHELE PARRINELLO
MENTOR: DR. ANDREA RIZZI
WEBSITE: https://www.fz-juelich.de/ias/ias-5/EN/Home/home_node.html
The recent diffusion of robust machine learning algorithms is opening up opportunities in molecular simulations that were unthinkable until only a few years ago. This project aims at exploring and developing techniques at the interface between machine learning and physics-based molecular simulations for the prediction of free energies and kinetics in neurobiologically-relevant systems.
In particular, the student will contribute to the development and testing of a neural network model of the interatomic potential energy with the goal of enabling the study of, for instance, enzymatic reactions and drug-protein binding at the density functional theory (DFT) quantum mechanical level of accuracy in hybrid quantum mechanics/molecular mechanics (QM/MM) settings.
Given the interdisciplinarity of the project, the student will have the opportunity to contribute in several ways. Depending on their interests, the student will acquire experience in one or more of the following topics:
- Development and training of neural networks potentials.
- Molecular dynamics with hybrid quantum mechanics/molecular mechanics models.
- Enhanced sampling methods for accelerating molecular simulations.
- Free energy and/or kinetics methods.
- Parallel computing algorithms in high-performance computing settings.
The project is part of a larger initiative on “Innovative high-performance computing approaches for molecular neuromedicine” funded by the Helmholtz European Partnering programme (PIs Paolo Carloni and Michele Parrinello). We anticipate that the work will lead to a publication in a high-impact journal, and opportunities for future follow-ups and longer-term collaborations between the groups will be abundant.
REQUIRED SKILLS
Students applying to this project are expected to have:
- Basic knowledge of programming languages (Python is preferred)
- Experience with molecular simulations (molecular mechanics and/or quantum)
COMPUTATIONAL RESOURCES
During the internship, the student will have the possibility to access the computational resources at Jülich Forschungszentrum and RWTH Aachen University IT Center, which include CPU and GPU cluster computing systems among the fastest and most energy-efficient in the world.