Advances in molecular simulations and bioinformatics

Because of the complexity of biological molecules, at times it may be convenient not to use the same description of the interatomic potential. Our group develops and applies hybrid methods spanning from quantum to coarse grain, assisted by machine learning.

Massively parallel QM/MM

Massively parallel Quantum Mechanics / Molecular Mechanics methods

MM/CG

Hybrid molecular mechanics/coarse grained approach

Enhanced sampling

Novel algorithms for metadynamics-based MD simulations

Kinetics calculations

Predicting dissociation rates associated with ligand unbinding events

Machine-learning-accelerated molecular simulations

Exploiting the representational power of neural networks to build accurate models of interatomic potentials and accelerate rare events

Structural neuroinformatics

Understanding the structure/function of proteins expressed in the nervous system by combining traditional bioinformatics and biophysics simulations

Computational mass spectrometry

Investigating the effect of charge in the absence of solvent, as experienced by the protein ions in the vacuum of the mass spectrometer

Graph-based algorithms for dynamic hydrogen-bond networks

Novel graph-based algorithms to identify and characterize interaction hydrogen-bond networks in complex bio-systems, based on large datasets from structural biology or from numerical simulations

Other applications

Additional applications based on molecular simulations and bioinformatics