Computer simulations have long been used to provide a conceptual framework to understand biophysical phenomena. Because of the high degree of complexity arising from the large number of interacting components, they constitute an essential class of tools in interpreting experimental studies, connecting them to fundamental physics, organizing our knowledge and asking new questions based on an ever improving picture.

While data-based models have provided powerful tools to generate hypotheses of biomolecular structures and interactions, current AI-models provide neither full transferability nor insights regarding emerging complex phenomena like structure formation in molecular assemblies from the  known laws of physics.

A physics based model of biomolecular processes in contrast is fully transferable and allows to zoom in to the level of individual atomic interactions to make sense of, for instance, protein folding and the interaction of proteins with their environment under different physical conditions. It can be directly used to explain the behaviour of intrinsically disordered proteins, as well as supra-molecular assembly processes such as peptide aggregation and coacervation.

Such processes typically involve thousands of interacting entities, complex interactions and inconveniently long time scales. Markov-chain Monte Carlo provides an underexplored tool to circumvent the time scale barrier characteristic for Molecular Dynamics. The workshop will introduce MCMC based simulation compare it to MD, and show ways to combine both methods. This will enable researchers to circumvent the limitations of the individual methods and to extract useful insights over a large span of spatial and temporal dimensions.

In this one-week CECAM Flagship School you will

• understand the principles of MCMC simulations and how they compare to MD simulations

• perform unbiased protein folding simulations on a supercomputer yourself

• generate and analyze free energy landscapes of protein folding and learn how to interpret them

• perform different Monte Carlo techniques and learn when and how to apply them

• practice parametrizing, monitoring, and optimizing Monte Carlo simulations

• get an overview of the ProFASi software, its tools, and its easy customization via plugins

• discuss pros and cons of different simulation approaches as well as your own use cases

Prerequisites

A basic understanding of molecular simulation of proteins.

A basic understanding of Linux commands, Jupyter notebooks and Python is helpful.

Fees and support

There are no course fees. Travel and accommodation need to be organized and paid by the participant. Unfortunately we cannot provide any travel grants.

The course is organized by the Simulation and Data Laboratory Biology at the Jülich Supercomputing Centre, Forschungszentrum Jülich, 52425 Jülich, Germany.