Virtual Reality Puts Molecular Dynamics in Students’ Hands

Imagine reaching into a protein’s binding pocket with your own hands, feeling the resistance of a ligand as it squeezes past a flexible amino acid side chain, or grabbing a carbon nanotube and watching graphene sheets deform in real time under your grip. What sounds like science fiction is now a documented teaching and research tool. A new review published in Frontiers of Digital Education by Jinyu Ma, Huawei Cao, Ding Nie, and Dongping Chen of Beijing Institute of Technology examines how virtual reality (VR) can be fused with molecular simulations to let students and researchers see, touch, and manipulate molecular structures, reactions, and behaviors across chemistry, materials science, biology, and physics. The paper, published on 22 September 2025, centers on an emerging platform called Manta and argues that immersive, interactive molecular dynamics could reshape how the microscopic world is taught in higher education.

The core problem the authors tackle is a familiar one in science education: molecules are three-dimensional, dynamic, and invisible, yet most students encounter them as static two-dimensional drawings or passive screen animations. Decades of chemistry education research, which the review cites extensively, show that weak spatial skills correlate with poor performance in general chemistry, and that three-dimensional representations can significantly improve learning outcomes. Traditional visualization tools such as PyMOL, VMD, and VESTA made molecular graphics interactive on a flat monitor, but the user remained an observer outside the system. VR changes that relationship fundamentally. By rendering the molecular world at immersive scale and tracking the user’s head and hands, VR turns the learner into a participant who can push, pull, and steer atoms while the underlying physics engine computes the consequences of every intervention.

Technically, the approach rests on interactive molecular dynamics, or iMD, a scheme in which the simulation does not merely play back a precomputed trajectory but responds continuously to user-applied forces. In a conventional molecular dynamics calculation, atoms move under a force field, a set of functions describing bonded interactions, electrostatics, and van der Waals forces, integrated forward in time with a fixed timestep. In the interactive version, the user’s hand controller applies an additional external force to selected atoms, and the integrator incorporates that perturbation into the equations of motion. The result is a physically plausible response: tugging on a ligand stretches bonds, distorts angles, and samples conformational states that a passive viewer would never see. The review situates Manta within a lineage of such tools, including the multi-user frameworks developed by the Glowacki and Mulholland groups, which demonstrated collaborative molecular manipulation in VR and even drug docking to the SARS-CoV-2 main protease.

Under the hood, the fidelity of any such system depends on the quality of the potential energy surface that drives the atomic motion. Classical force fields such as CHARMM, implemented in engines like GROMACS and OpenMM, are fast but cannot describe bond breaking. Reactive force fields such as ReaxFF, developed by van Duin and Goddard, allow chemical bonds to form and dissociate, making them suitable for combustion, decomposition, and catalysis problems. At the higher-accuracy end, density functional theory and semiempirical methods such as DFTB provide quantum mechanical descriptions at greater computational cost, while GPU-accelerated codes like TeraChem have made ab initio interactive dynamics feasible. The review also highlights the newest layer of this stack: machine-learned potentials, including the Deep Potential and DeePMD-kit frameworks and large atomic models such as DPA-2, which approach quantum accuracy at a fraction of the cost and are increasingly practical for real-time interactive simulation.

Against this theoretical backdrop, the authors walk through a series of case studies that showcase Manta’s versatility across disciplines. The first involves aluminum and graphene, a materials-science system in which users can explore interfacial structure and deformation dynamics at the atomic scale. By grabbing and shearing the graphene layer within the VR environment, students observe how the metal and the two-dimensional carbon sheet interact, how strain distributes, and how defects nucleate, experiences that static textbook figures cannot convey. The case demonstrates how molecular deformation dynamics, a topic usually confined to research seminars, can be made tangible for undergraduates in materials science and mechanical engineering programs.

The second case study brings organic chemistry to life through the Diels–Alder reaction, the classic cycloaddition between a conjugated diene and a dienophile that forms a six-membered ring. In the VR environment, students do not simply watch a reaction coordinate diagram; they steer the two reactants toward one another, feel the energetic landscape through haptic feedback, and watch new sigma bonds form as the geometry approaches the transition state. Because reactive force fields or semiempirical quantum methods can describe the bond-making event, the simulation captures genuine chemistry rather than a canned animation. The authors argue that this embodied encounter with a pericyclic reaction helps students internalize concepts, orbital overlap, stereochemistry, and activation energy, that are notoriously abstract when presented on a whiteboard.

A third, more dramatic application concerns the thermal decomposition of energetic materials, a specialty of the Beijing Institute of Technology’s State Key Laboratory of Explosion Science and Safety Protection, where all four authors are based. Energetic materials such as CL-20/TNT cocrystals and hydrazine compounds release enormous energy on decomposition, and understanding the elementary reactions involved is critical for safety and design. ReaxFF-based reactive dynamics and neural-network potentials have been used to simulate these processes, and the review shows how VR visualization lets researchers and students watch decomposition cascades unfold atom by atom, identifying which bonds rupture first and which small molecules evolve. For a field where experiments are hazardous and often impossible to observe directly, immersive simulation offers both a research instrument and a safe educational window.

Perhaps the most pedagogically compelling case is the virtual docking experiment aimed at computational biology students. Molecular docking, predicting how a small ligand binds into a protein pocket, is a staple of drug discovery, yet conventional docking software presents results as ranked poses with scoring functions that students often accept on faith. In the interactive VR version, students grasp a flexible ligand and manually guide it into the binding site while the protein’s side chains relax around it, sampling binding pathways that rigid-receptor docking misses entirely. Prior work cited in the review showed that interactive iMD in VR achieves accurate flexible protein–ligand docking and can even estimate free energies along binding pathways. For undergraduates learning structure-based drug design, physically feeling steric clashes and electrostatic attraction transforms docking from a black box into an intuitive exploration.

The review also addresses computational chemistry education directly, noting that tools like Gaussian and GaussView have long been used in analytical and physical chemistry teaching but impose a steep learning curve: students must master input files, basis sets, and convergence criteria before seeing any result. By contrast, the VR-based approach described in one case study quantifies structural changes, bond lengths, angles, and energies, as the user manipulates the molecule, delivering quantitative insight through direct interaction. The authors argue that this lowers the barrier to entry for quantitative structural analysis and aligns with constructivist learning theory, in which learners build understanding through active engagement rather than passive reception. Earlier studies of VR learning environments cited in the paper support this view, reporting positive learner attitudes and improved engagement in chemistry courses that adopted immersive platforms.

The authors conclude that VR-enhanced molecular simulation has the potential to revolutionize both research practice and education across molecular dynamics, materials science, and beyond, and their timing is persuasive. Interactive iMD in VR has already been named among emerging technologies in chemistry, and the explosion of machine-learned potentials is removing the computational bottleneck that once made real-time quantum-accurate simulation unthinkable. Challenges remain, including headset cost, motion sickness for some users, the difficulty of assessing learning gains rigorously, and the need for curated curricula that pair immersive experiences with solid theoretical grounding. But the trajectory is clear: as Manta and its predecessors demonstrate, the days of teaching chemistry from flat pictures may be numbered. The next generation of scientists may learn the shape of a molecule the way they learn the shape of a room, by walking around inside it, reaching out, and giving it a push.

Subject of Research: Interactive molecular dynamics in virtual reality for multidisciplinary science education

Article Title: Interactive Molecular Dynamics in Virtual Reality for Multidisciplinary Education: Theory and Higher Education Applications

Article References: Ma, J., Cao, H., Nie, D., & Chen, D. (2025). Interactive Molecular Dynamics in Virtual Reality for Multidisciplinary Education: Theory and Higher Education Applications. Frontiers of Digital Education, 2(4), Article 36. https://doi.org/10.1007/s44366-025-0073-8

Image Credits: AI Generated

DOI: 10.1007/s44366-025-0073-8

Keywords: virtual reality, molecular dynamics, interactive simulation, Manta, higher education, chemistry education, Diels-Alder reaction, molecular docking, ReaxFF, machine-learned potentials, materials science, immersive learning