Advances in Human-Protein Interaction - Interactive and Immersive Molecular Simulations

Molecular simulations allow researchers to obtain complementary data with respect to experimental studies and to overcome some of their limitations. Current experimental techniques do not allow to observe the full dynamics of a protein at atomic detail. In return, experiments provide the structures, i.e. the spatial atomic positions, for numerous biomolecular systems, which are often used as starting point for simulation studies. In order to predict, to explain and to understand experimental results, researchers have developed a variety of biomolecular representations and algorithms. They allow to simulate the dynamic behavior of macromolecules at different scales, ranging from detailed models using quantum mechanics or classical molecular mechanics to more approximate representations. These simulations are often controlled a priori by complex and empirical settings. Most researchers visualise the result of their simulation once the computation is finished. Such post-simulation analysis often makes use of specific molecular user interfaces, by reading and visualising the molecular 3D configuration at each step of the simulation. This approach makes it difficult to interact with a simulation in progress. When a problem occurs, or when the researcher does not achieve to observe the predicted behavior, the simulation must be restarted with other settings or constraints. This can result in the waste of an important number of compute cycles, as some simulations last for a long time: several days to weeks may be required to reproduce a short timespan, a few nanoseconds, of molecular reality. Moreover, several biomolecular processes, like folding or large conformational changes of proteins, occur on even longer timescales that are inaccessible to current simulation techniques. It can thus be necessary to impose empirical constraints in order to accelerate a simulation and to reproduce