A classical lay summary by Ragothaman M Yennamalli, from the SASTRA University.
Imagine you are at a park on a busy day. There’s activity everywhere: some people are chatting, others are listening to a busker, many are strolling slowly toward a food truck, while some are running or skating. Despite the hustle and bustle, there are no collisions or mishaps because everyone follows basic rules and etiquette.
Molecular dynamics (MD) simulation is similar to this scene, but instead of a park, the setting is at the atomic scale. Molecules—such as proteins, nucleic acids, or drugs—interact within their environment and follow the basic rules of physics and chemistry.
In this analogy, the people moving around represent atoms in an MD simulation. Just as some people hold hands, form small groups, and move together, atoms form chemical bonds and exhibit coordinated movements. At the atomic scale, molecules display harmonic motions, interact with other molecules, and exhibit dynamic behavior.
Dr. Aneesur Rahman, known as the father of molecular dynamics, conducted the first MD simulation using argon atoms in 1964. He solved the Lennard-Jones potential, which describes the attraction and repulsion between two nonbonding atoms or molecules, between 864 argon atoms in this setup. This groundbreaking work, conducted on a mainframe computer (CDC 3600 computer) with only 1.5 MB of memory, was a computational experiment to study the "space and time dependence of two-body correlations." [1]
Interactions between atoms and molecules can be described using quantum mechanical, statistical mechanical, or classical mechanical models. However, quantum and statistical models are computationally expensive and limited in their use for biological macromolecules. MD simulations rely on classical mechanics, solving Newton’s second law of motion to calculate how atoms move over time.
Like the park analogy, molecules in a given environment exhibit motions driven by temperature and other physical factors. Since we know an atom’s mass and the forces acting on it, we can calculate its acceleration. These motions occur at incredibly fast timescales, measured in femtoseconds (10⁻¹⁵ seconds). This is because atomic bonds, which act like tiny springs, vibrate very rapidly, undergoing stretching, bending, and twisting at femtosecond intervals.
To capture these rapid, transient interactions accurately, MD simulations divide time into tiny steps of 1–2 femtoseconds. This granularity allows us to understand the dynamic nature of molecules, which is directly tied to their function—such as how proteins fold or how a drug binds to a target. Every small positional change alters velocity and energy, requiring atomistic-level calculations of the potential energy function at each timestep. These computations are intensive and time-consuming, even for simulations lasting only a few picoseconds. Dr. Rahman’s original simulation lasted just 3 picoseconds.
Today, advancements in computational science allow us to run micro- and millisecond-scale simulations far faster than ever before, thanks to powerful graphical processing units (GPUs), improved storage capabilities, and widespread access to high-performance computing resources.
But what do we do with all the data generated from these simulations? Think of it like a flipbook: each page shows a slightly different position, and when flipped, it reveals the coordinated movement of a figure. Similarly, MD simulations produce trajectories—snapshots of a molecule’s movement and dynamics over time. By analyzing these trajectories, we gain insight into how biological molecules behave in their environments.
Expanding beyond single molecules, researchers have successfully simulated entire cells [2]. By replicating the composition of diverse molecules (proteins, metabolites, ions, lipids, etc.) in close proximity, they have uncovered intricate mechanisms, such as the complex processes involved in protein folding. MD simulations continue to push the boundaries of our understanding of molecular behavior, paving the way for new discoveries in biology and medicine.
MD simulations have wide applications ranging from material science to protein folding to drug design. The area got wider notice and recognition with the Drs. Martin Karplus, Michael Levitt and Arieh Warshel getting the 2013 Nobel Prize in Chemistry "for the development of multiscale models for complex chemical systems" [3].
References:
- Rahman (1964). "Correlations in the Motion of Atoms in Liquid Argon". Physical Review. 136: A405-A411. doi:10.1103/PhysRev.136.A405.
- Samuel Russell PP, Alaeen S, Pogorelov TV. In-Cell Dynamics: The Next Focus of All-Atom Simulations. J Phys Chem B. 2023 Nov 23;127(46):9863-9872. doi: 10.1021/acs.jpcb.3c05166.
- Development of Multiscale Models for Complex Chemical Systems From H+H2 to Biomolecules. Nobel Lecture, December 8, 2013 by Martin Karplus (https://www.nobelprize.org/uploads/2018/06/karplus-lecture.pdf)