MolecularWebXR: Browser-Based Multiuser VR

• MolecularWebXR is a free, browser-based XR platform that enables immersive, multiuser virtual collaboration for molecular visualization and discussion across diverse devices.
• It leverages WebXR technology with lightweight state synchronization and built-in audio communication to facilitate natural 3D interaction and reduce bandwidth demands.
• The platform is widely used for education, outreach, and scientific collaboration, offering accessible immersive experiences in classrooms, conferences, and remote sessions.

MolecularWebXR is a free, browser-based, multiuser virtual reality platform for chemistry and biology education, science communication, and scientific peer discussion, built on WebXR and delivered with no installation barrier through ordinary web access. It is designed to be deeply immersive on high-end VR headsets while remaining usable from smartphones, tablets, laptops, and desktops, including smartphone viewing through cardboard goggles. Its defining combination is WebXR-based immersion, multiuser shared sessions, built-in audio communication, browser delivery, and cross-device inclusivity; subsequent ecosystem-level descriptions place it as the shared immersive collaboration layer of the broader MolecularWeb platform, positioned between earlier commodity AR tools and later work toward physics-informed molecular modeling (Rodriguez et al., 2023, Abriata, 4 Sep 2025).

1. Conceptual scope and historical position

MolecularWebXR was developed from the premise that chemistry and biology are intrinsically three-dimensional, yet most teaching, communication, and discussion still depend on flat displays and indirect mouse-and-keyboard interaction. The platform addresses not only the representational loss imposed by 2D viewing, but also the practical difficulty of shared concurrent manipulation of scientific 3D objects. In the 2023 preprint, its stated aim is to make immersive, collaborative molecular and structural-biology visualization practical for education, outreach, and scientific discussion by combining natural 3D interaction, multiuser sessions, built-in communication, zero-install browser delivery, and accessibility from both XR and non-XR devices (Rodriguez et al., 2023).

Within the MolecularWeb ecosystem, later accounts describe MolecularWebXR as a room-based shared immersive venue rather than a content-generation tool or a full molecular simulation engine. In that framing, moleculARweb supplies commodity AR activities, PDB2AR supplies custom molecular assets, and MolecularWebXR supplies the multiuser virtual rooms in which those assets are discussed, inspected, and manipulated. The same later source is explicit that current MolecularWebXR handles mostly static 3D models, whereas HandMol is presented as the route toward dynamic, physics-informed interactive modeling (Abriata, 4 Sep 2025).

This positioning distinguishes MolecularWebXR from earlier browser-native molecular systems. HTMoL established that browser delivery can handle remote visualization of raw MD trajectories, but it was centered on streaming and visualizing molecular dynamics data rather than immersive XR interaction (Carrillo-Tripp et al., 2017). Earlier concurrent JSmol-based work showed that browser peers could synchronize rotations, zoom, and commands in near real time, but it did not introduce WebXR, immersive presence, avatars, or natural hand-based interaction (Abriata, 2017). Conversely, native VR systems for protein inspection and interactive molecular dynamics demonstrated much richer immersive interaction, but they relied on Unity/SteamVR or dedicated iMD-VR infrastructures rather than inclusive browser delivery (Ratamero et al., 2017, O'Connor et al., 2019).

2. System architecture and synchronization model

The platform relies on WebXR, described by its authors as an API and specification that lets web content interface with mixed-reality hardware while standardizing different devices’ input capabilities into browser events. On the server side, MolecularWebXR uses a backend running in Node.js that centralizes room creation and room management. One user creates a room or joins an existing room using a unique code; the room creator becomes the Admin and can configure aspects of the session such as whether object grabbing is enabled (Rodriguez et al., 2023).

Its networking model is deliberately lightweight. The paper states that no video is transmitted between users; instead, the system sends only the quaternions describing object positions, object orientations, object scales, and the poses of head and hand avatars. This places MolecularWebXR in a state-synchronization regime rather than a streamed-graphics regime, which is technically important because it reduces the bandwidth burden during live collaboration. The same paper notes that bandwidth consumption is high when VR objects are downloaded upon room entry, but once the room has loaded, very high-speed Wi‑Fi is not required; a stable internet connection is more important for fluid updates during interaction (Rodriguez et al., 2023).

This architectural choice aligns MolecularWebXR with a broader class of web-native collaborative scientific systems that synchronize scene state instead of pixels. The resemblance is especially strong to the earlier collaborative JSmol prototype, which also exchanged local state and commands rather than screen streams (Abriata, 2017). A plausible implication is that MolecularWebXR’s performance envelope depends less on continuous media transport than on object download latency, transform-update stability, and browser/device support for WebXR itself. More generally, WebXR performance studies argue that browser XR can benefit from pairing WebXR with WebAssembly for heavy computation, although MolecularWebXR itself is not described as using such a stack (Kim et al., 2021).

3. Participation modes, avatars, and interaction design

MolecularWebXR is explicitly designed around heterogeneous participation. The richest mode is headset-based immersive VR with six degrees of freedom, natural head movement, and either hand tracking or controllers. In this mode, users can grab objects, translate them, rotate them, scale them, pass them around, point naturally with their hands, and walk around the scene while seeing other VR users as simplified head-and-hands avatars. The 2023 paper reports primary development and testing on Oculus Quest 2 and Oculus Pro with hand tracking enabled, also verified with handheld controls, and additional verification on Oculus Quest 1, Meta Quest 3, and HTC Vive Pro (Rodriguez et al., 2023).

The same system remains accessible from simpler devices. Smartphones support a non-immersive mode with a virtual joystick and touch-based looking, and an immersive cardboard-goggle mode when the phone supports WebXR. The latter yields 360-degree viewing but only three degrees of freedom, with no hand tracking and no ability to grab objects. Tablets support touch-based navigation without true immersive VR. Desktop and laptop users navigate with arrow keys or $W,A,S,D$, look with the mouse, and can observe avatars and objects. In the user-role model described by the paper, Admins manage the room, VR-active users can actively manipulate objects and talk, and passive users can follow the session and listen but cannot grab objects and cannot talk (Rodriguez et al., 2023).

The interaction model is intentionally embodied rather than command-driven. A recurring claim in the paper is that users manipulate scientific objects with their bare hands, and that this is more natural than conventional 2D interaction. For co-located sessions, the authors align guardians or safe spaces so that real and virtual relative positions correspond, allowing physically co-present participants to feel proximity naturally inside the virtual room. This differs sharply from earlier Unity/SteamVR molecular inspection pipelines, which relied on controller-based grabbing in installed native applications, and from iMD-VR systems such as Narupa, where interaction applies biasing forces to live simulations rather than moving prebuilt shared objects (Ratamero et al., 2017, O'Connor et al., 2019).

4. Scientific content and custom scene generation

MolecularWebXR organizes content as preset rooms plus an empty room for ad hoc material. At launch, the preset rooms were Symmetry, Orbitals, Isomerism, Materials and Crystals, Structural Biology, Cryo-Tomography, and an Empty room. These rooms target concepts that the authors regard as especially difficult to teach or discuss through 2D media, including molecular symmetry, molecular orbitals, conformations, crystal structures, protein structure, large biomolecular assemblies, and cryo-electron tomography–derived cellular landscapes (Rodriguez et al., 2023).

Custom content is introduced through PDB2AR, a VMD-based web tool from the same ecosystem. The 2023 paper states that users can prepare content following previously published procedures, receive a GLB file link by email, and paste that link into MolecularWebXR to populate the empty room. The broader 2025 chapter gives a more explicit pipeline: PDB2AR can start from raw PDB input, structures fetched from the Protein Data Bank, models from AlphaFold DB by UniProt entry, or VMD-exported WaveFront objects; it then generates a VMD script, processes the content on the server, and produces a compressed GLB model that can be used in standalone AR/VR or inserted into a MolecularWebXR room (Rodriguez et al., 2023, Abriata, 4 Sep 2025).

Because the pipeline uses VMD conventions, it supports molecular representations such as ball-and-stick, cartoons, isosurfaces, and forms suitable for electronic orbitals and cryo-electron maps. Later descriptions of MolecularWebXR also emphasize that the rooms are not restricted to molecules in a narrow sense: they can contain molecular structures, surfaces, orbitals, electron maps, viruses, ribosomes, nucleosomes, protein–DNA complexes, and related scientific 3D objects. At the same time, the 2025 chapter is explicit that the models currently used inside MolecularWebXR are static in terms of their internal structural dynamics; users can move and resize them, but the platform is not yet an atomistically simulated molecular graphics workbench (Abriata, 4 Sep 2025).

5. Educational, outreach, and scientific uses

The reported deployment record is broad. The 2023 paper describes use in science outreach days, an EPFL open science day, school visits, student instruction, conference lectures, scientific collaborations, and remote sessions linking participants separated by more than 11,000 km between Rosario and Lausanne. It also reports short public demonstrations in which almost 150 people tried a 15-minute VR presentation and none had to stop early because of VR sickness or other problems (Rodriguez et al., 2023).

The same paper emphasizes age and experience range as a usability result. Users who tried VR-headset experiences ranged from 12 to 80 years old, more than 80% had absolutely no previous experience utilizing VR headsets, and all of them could seamlessly manipulate objects with their hands. In a practical scalability observation rather than a formal benchmark, the authors report sessions with up to 8 simultaneous VR-active users inside VR, plus an Admin on a laptop and two passive online users, with no apparent lag (Rodriguez et al., 2023).

The scientific content itself spans both chemistry and structural biology. Examples include symmetry elements for $\mathrm{CO_2}$, frontier molecular orbitals such as the LUMO of $\mathrm{O_2}$, conformations of butane, local coordination in calcite $\mathrm{CaCO_3}$, manually superimposing an atomically detailed alpha helix onto a cartoon helix in a transcription factor bound to DNA, viral assemblies such as SARS-CoV-2 and bacteriophages, and cryo-ET scenes such as SARS-CoV-2 spike bound to ACE2 aligned to EMDB 30430 density (Rodriguez et al., 2023). Later ecosystem documentation expands these examples into more structured rooms for periodic table exploration, VSEPR theory, structural biology coursework, and the “Physics to Biology” outreach room that connects body scale, cells, viruses, DNA, protein–DNA recognition, electron density, drug binding, and orbitals inside one guided immersive narrative (Abriata, 4 Sep 2025).

6. Limitations, misconceptions, and future trajectory

A central misconception would be to treat MolecularWebXR as either a headset-only system or a full immersive molecular modeling engine. The primary paper rejects the first interpretation by design: cross-device accessibility is one of its central claims, and users without headsets can still participate from smartphones, tablets, laptops, or desktops (Rodriguez et al., 2023). The later chapter rejects the second interpretation explicitly: current MolecularWebXR is mainly a shared visualization and discussion platform based on static 3D objects, not yet a full environment for live atomistic deformation, residue editing, direct volumetric-data handling as native immersive data, or real-time molecular simulation (Abriata, 4 Sep 2025).

Other limitations are operational. Smartphone immersive participation is restricted to three degrees of freedom and does not support hand tracking or object grabbing. Non-VR devices remain less immersive and depend on indirect controls. Object download at room entry can consume significant bandwidth, stable connectivity remains important, and mixed local/remote audio requires careful management. The validation evidence is substantial but mostly practical and qualitative: the papers do not report controlled comparative studies, formal learning-gain analyses, latency benchmarks, or rigorous task-performance measurements (Rodriguez et al., 2023).

The future direction presented in the MolecularWeb literature is expansion rather than replacement. The 2023 paper identifies the next step as a web app for immersive molecular simulations supporting both VR and non-VR devices. The 2025 chapter ties that trajectory to HandMol, described there as an upcoming WebXR software layer for concurrent multiuser immersive visualization and modeling with bare hands, real-time molecular mechanics, natural-language input via a LLM, and access from both headsets and consumer devices. This suggests that MolecularWebXR is best understood not as a final endpoint but as the browser-native shared-room substrate from which more dynamic molecular work environments are intended to emerge (Rodriguez et al., 2023, Abriata, 4 Sep 2025).