Beyond Classical Physics: The Power of QM/MM Approaches

By Marta Devodier

Hello everyone! I am Marta, and today I want to share a bit about my work as a computational biophysicist. My field combines principles of physics, biology, mathematics, and computer science to study biological systems. Using computational tools and simulations, we can model, analyze, and predict the behavior of biological molecules and processes. Let’s dive together into some key aspects of this fascinating discipline!
Proteins, the tiny building blocks of our bodies, act as molecular machines with highly specific jobs. They are primarily composed of hydrogen atoms, which make up roughly 50% of their atomic composition. Hydrogen atoms play key roles in maintaining the structure and function of proteins:

• Hydrogen bonds contribute to the stability of protein folds, facilitate interactions with other molecules, and enable biochemical processes such as enzyme-substrate interactions and signal transduction.
• The collective behavior of nonpolar molecules or regions containing hydrogen atoms bonded to carbon atoms (C-H bonds) and the way they interact with water molecules leads to hydrophobic interactions.

Understanding these interactions is key to unraveling the molecular mechanisms underlying biological processes. From a theoretical physics perspective, in most cases, the structure, dynamics and energetics of these interactions can be described by classical molecular dynamics (MD) simulations. Classical MD is based on the solution of Newton’s equations of motion and on ad hoc potential energy functions (or force fields), which describe the interactions between the particles of a system.

However, specific, highly relevant phenomena, (such as proton and/or electron transfer processes (Fig. 1) or photoabsorption) require inclusion of electronic degrees of freedom that cannot be described using classical physics methodologies.

Since biological systems typically contain more than a few thousand atoms, the treatment of the whole system using first principles quantum mechanical (QM) approaches (typically density functional theory, DFT) is far too costly. To overcome this problem, one of the most powerful tools in my research usually used is a hybrid Quantum Mechanics/Molecular Mechanics (QM/MM) approach. Here, the system (for instance a protein and its binding site) is split into two parts: a smaller part (QM region) that is treated at the QM level of theory (for instance using DFT), and the rest of the system (MM region), in which atoms interact according to classical force field (Fig. 2).

By bridging these two worlds, QM/MM approaches make it possible to investigate large, complex systems, achieving a level of accuracy that classical methods alone can’t provide, while still keeping the computational cost manageable. To underline the importance of this method it will suffice to mention that it was awarded the Nobel prize in Chemistry in 2013, so far the only Nobel prize in the field of molecular simulations. However, even these approaches face the issues of sampling (that is, they cover too short timescales), severely limiting the domain of applications of this approach. For this reason, the current exascale revolution in high performance computing (HPC) represents an exciting opportunity to overcome these limitations by developing scalable software able to take full advantage from modern massively parallel architectures to tackle the time scale problem! Thus, currently the computational physicist is called upon using these powerful tools which can lead to unprecedented insight into biophysical systems.

Are you curious about how QM/MM approaches work, or have any questions about computational biophysics? Feel free to leave a comment—I’d love to hear your thoughts!