Joshua Opolko

VR in Drug Discovery: How Pharma Companies Use Virtual Reality to Find New Medicines

By Joshua Opolko · Published July 20, 2026

Short version: Pharmaceutical companies use VR to visualize molecular structures at human scale, letting chemists and biologists walk inside a protein binding pocket (the cavity where a drug molecule attaches to its target) and rotate a candidate drug molecule with their hands. This is not replacing computational chemistry, but it is changing how researchers communicate and make decisions about candidate molecules. Nanome, the leading platform, is used by Novartis, Roche, and over 30 pharma and biotech companies. The main claim: immersive 3D interaction surfaces insights that flat-screen molecular viewers miss, particularly in fragment-based drug design, where researchers must visualize exactly how a small molecular fragment fits inside a complex protein cavity.

Key takeaways

What problem does VR solve in drug discovery?

Drug discovery is slow and expensive. The Tufts Center for the Study of Drug Development estimates the average cost of bringing a new drug to market at $2.6 billion, with timelines of 10 to 15 years from initial target identification to approval (Tufts CSDD, 2014). A meaningful share of that time is spent in the early research phase, where scientists try to find molecules that bind tightly to a specific protein target and produce a desired biological effect.

Proteins are three-dimensional objects. They fold into complex shapes, and the sites where drug molecules attach, called binding pockets or active sites, are cavities with precise spatial geometry. A drug candidate needs to fit into that cavity in a specific orientation. Getting the shape wrong means the drug does not work, or causes side effects by also binding to unintended targets.

The problem is that nearly all molecular visualization has happened on flat screens. Software tools like PyMOL, UCSF Chimera, and Schrodinger Maestro display protein structures in 2D, requiring researchers to mentally reconstruct a three-dimensional object from a rotating image. This mental reconstruction is something scientists train themselves to do over years. It is also an area where spatial misreadings happen, particularly when examining tight, enclosed binding pockets or when comparing how two different candidate molecules occupy the same site.

Multi-site collaboration adds another layer. A medicinal chemist in Basel reviewing a binding pose with a structural biologist in Cambridge means screen-sharing a PyMOL session and verbally describing spatial relationships that both people are trying to visualize simultaneously from a 2D image. VR changes that specific interaction: both people are in the same 3D environment, pointing at the same atoms.

This same spatial collaboration challenge appears in other sectors. The patterns that drive enterprise VR adoption in growing teams map closely onto what pharma research groups describe: the value is not in the rendering quality but in the shared spatial reference frame.

How does molecular VR work?

The core workflow is straightforward. A researcher downloads a protein structure file, typically in PDB format (Protein Data Bank format, a standard text file describing the positions of every atom in a protein), from a public repository like the RCSB Protein Data Bank at rcsb.org. They import it into a VR platform like Nanome. The platform renders the molecule and scales it up to human size, so what was a nanometer-scale object becomes something a person can walk around and reach into.

Inside the headset, the researcher can rotate and translate the structure using hand controllers, toggle between different visual representations (ribbon diagrams that show the protein backbone, ball-and-stick models that show individual atoms, surface representations that show the outer shape of the binding pocket), and place annotations on specific atoms or residues (the individual amino acid units that make up the protein chain).

Candidate drug molecules, saved as SDF or MOL2 files from a computational chemistry workflow, are imported alongside the protein. The researcher can then examine how the drug molecule sits inside the binding pocket: which atoms are close to which protein residues, where there might be steric clashes (places where atoms from the drug and protein overlap and repel each other), and where there is room to add chemical groups to improve binding.

Multi-user sessions work through a shared server. Researchers in different physical locations join the same molecular environment, see each other's avatars or hand representations, and can take turns grabbing and rotating the structure. Nanome's platform supports this natively, which is the feature pharma companies describe using most for cross-site medicinal chemistry reviews.

Connection to computational chemistry software is increasingly direct. Schrodinger announced a partnership with Nanome in 2021 to allow import of Maestro format files (the native format for Schrodinger's computational chemistry platform) directly into the VR environment. Researchers using MOE (Molecular Operating Environment, from Chemical Computing Group) or OpenEye's tools can export standard file formats that Nanome and similar platforms accept.

Citation capsule: Nanome is a VR platform for molecular visualization used by over 30 pharmaceutical and biotech companies, including Novartis and Roche. It supports import of PDB and SDF files, multi-user collaborative sessions, and, through a 2021 partnership with Schrodinger, direct import of Maestro format files. Nanome raised a $14 million Series A in 2022. (Source: Nanome, nanome.ai; Schrodinger partnership announcement, 2021.)

What VR platforms do pharmaceutical researchers use?

Several platforms target molecular VR, each with different technical depth, hardware support, and pricing. Nanome is the most widely deployed in pharmaceutical settings. The University of York's iMD-VR is the most notable research prototype for a specific advanced use case. Others occupy more specialized or earlier-stage positions.

Nanome

Nanome is the leading commercial platform for pharmaceutical molecular VR. It supports Meta Quest headsets (standalone, no PC required for basic sessions) as well as PC-tethered devices for higher-fidelity rendering. Features include multi-user shared sessions, PDB and SDF file import, Schrodinger Maestro integration (2021 partnership), in-VR measurement tools, and annotation. A free academic tier is available at nanome.ai. Enterprise pricing is negotiated per organization. Nanome raised a $14 million Series A in 2022 and publicly lists Novartis and Roche among its pharma users, with over 30 pharma and biotech companies using the platform.

iMD-VR (University of York)

iMD-VR stands for interactive Molecular Dynamics in Virtual Reality. Researchers at the University of York published a paper in Science Advances in 2018 describing a system that streams a running molecular dynamics simulation (a physics simulation of atoms moving over time) directly into a VR headset, allowing a researcher to physically grab atoms and steer the simulation in real time (Molecular dynamics in VR, Science Advances, 2018). This is a more technically complex use case than static structure viewing: it requires a computing cluster running the simulation, a network connection to the headset, and software to handle the real-time rendering. The iMD-VR approach is a research tool, not a commercial product, but it represents the most advanced form of VR-molecular dynamics integration published to date.

Molecular Rift

Molecular Rift is a research prototype developed to explore VR molecular visualization with early Oculus hardware. It demonstrated the core concept of importing PDB files and examining protein structures in a head-mounted display, but has not evolved into a maintained commercial product. It is cited frequently in the academic literature as an early proof-of-concept.

SAMSON (SAMSON Software)

SAMSON is a modular molecular design platform from SAMSON Software (Institut Pasteur and associated developers) that supports VR as one of its visualization modes. It is aimed at structural biologists and computational chemists who want a programmable platform for molecular design, with VR as an optional output. It occupies a different niche from Nanome: more customizable, less turnkey.

Foldit

Foldit is a citizen science game developed by the University of Washington that lets non-expert players manipulate protein structures to solve folding puzzles. It has experimental VR extensions. It is relevant to drug discovery primarily as a source of human spatial intuition about protein shapes, not as a professional workflow tool.

Which pharmaceutical companies use VR for molecular visualization?

Nanome publicly names Novartis and Roche as customers, and documents use cases including structure-based drug design (SBDD) and fragment-based drug design (FBDD) review sessions. The company reports over 30 pharma and biotech companies in its user base as of its 2022 fundraising announcement.

AstraZeneca and Merck research groups have been cited in industry conference presentations and VR-in-pharma roundups as organizations that have piloted or adopted molecular VR tools, though peer-reviewed outcome publications from these programs are limited. The bulk of documented pharma VR work exists in conference presentations, vendor case studies, and a small number of academic papers.

The Schrodinger-Nanome partnership announced in 2021 is significant because Schrodinger is used by a large share of the pharmaceutical industry for computational chemistry. The partnership allows researchers already working in Maestro to bring their docking results and protein-ligand complexes (the protein and candidate drug molecule together, as predicted by the docking software) directly into a VR session without a separate file conversion step.

For context on how VR technology adoption works across different healthcare applications, the patterns in VR in surgical training are instructive: technology usually enters through a research or academic partnership before commercial deployment, and the leading platforms become known through a combination of peer-reviewed publications and industry conference demos.

Citation capsule: The AlphaFold Protein Structure Database, developed by DeepMind and hosted by EMBL-EBI, contained over 200 million predicted protein structures as of 2023. Researchers can download structures in PDB format and import them directly into VR platforms like Nanome for immersive 3D inspection. (Source: DeepMind/EMBL-EBI AlphaFold Database, 2023; alphafold.ebi.ac.uk.)

What do researchers say VR changes about their work?

Researchers describe several specific changes, and it is worth separating the ones that come up repeatedly from the promotional claims that appear only in vendor materials.

The most consistent report is about catching binding pose problems earlier. A binding pose is the predicted position and orientation of a candidate drug molecule inside the protein binding pocket. Computational docking software generates dozens or hundreds of these poses, ranked by predicted binding energy. Medicinal chemists review the top poses to decide which to pursue in the lab. In VR, researchers report noticing steric clashes and spatial mismatches that were not obvious when reviewing the same pose on a flat screen, because the immersive view makes depth relationships clearer.

The second common report is about explaining structures to non-chemist stakeholders. A biologist, a clinical pharmacologist, or a project manager who is not trained in reading PyMOL ribbon diagrams can walk into a VR session and understand the physical relationship between a drug molecule and its target protein in a few minutes. This changes how medicinal chemistry teams communicate with the rest of a drug discovery program.

Onboarding is a third area. New medicinal chemists, even those with strong computational chemistry training, need time to build intuition about protein structures in three dimensions. Experienced researchers describe using VR sessions to show new team members how a binding pocket works and why certain chemical changes are more promising than others. The claim is that the immersive format accelerates this spatial learning.

Collaborative annotation is distinct from what screen-sharing a PyMOL session provides. When two researchers share a VR molecular environment, they can each interact with the structure, point at specific atoms with their controllers, and place persistent annotations that remain visible to both people. The interaction is more like being in the same room at a physical model than like watching someone else navigate a 3D viewer on their screen.

What are the limitations of VR in drug discovery?

VR interaction is not computation. This is the most important limitation to state clearly. A researcher wearing a headset and examining a binding pose cannot improve the predicted binding affinity of a candidate molecule. They can observe the spatial arrangement, form a hypothesis about a chemical modification, and then return to a computational tool to test that hypothesis. The value is in the observation and communication step, not in replacing the computational step.

No VR platform currently performs force-field optimization, molecular dynamics simulation, or binding free energy calculation inside the headset at a level useful for drug design decisions. The iMD-VR system from the University of York streams a running simulation to the headset, which is the closest existing approach, but it requires a separate compute cluster and is a research prototype, not a production workflow tool.

Hardware cost has decreased significantly with standalone headsets like the Meta Quest series, but enterprise VR deployments at pharmaceutical companies still involve IT management costs, user training, and headset maintenance that add up. The free academic tier from Nanome lowers the barrier for academic groups considerably.

Learning curve varies. Experienced PyMOL or Chimera users have strong 2D-to-3D mental models already. Some find VR redundant; others find it genuinely accelerates certain types of spatial reasoning. New researchers or non-chemist stakeholders often get more relative value from the immersive format than experts who have already trained their 3D spatial intuition on flat screens.

Haptic feedback is limited. Current VR controllers give no meaningful tactile sensation when virtually grasping a molecular model. The sense of physical texture or resistance that chemists associate with physical molecular model kits is absent. This matters most for workflows where physical manipulation intuition is important, such as manually docking a small molecule into a binding site.

Controlled outcome studies are sparse. The peer-reviewed literature on molecular VR outcomes, specifically studies that compare decision quality or time-to-insight between VR and flat-screen molecular viewers for the same tasks, is small. The iMD-VR Science Advances paper (University of York, 2018) is the most rigorous published study on immersive molecular dynamics interaction. Most other evidence comes from conference presentations, vendor case studies, and qualitative researcher reports.

For a comparison case where VR outcomes have been studied more rigorously, the use of VR in anatomy and biomedical education has a somewhat larger evidence base, partly because educational outcomes are easier to measure than drug discovery decision quality.

Where is VR in drug discovery heading?

The most significant recent development is the AlphaFold database. DeepMind and EMBL-EBI released AlphaFold 2 predictions for essentially the entire human proteome in 2021, and expanded the database to over 200 million structures from across biology by 2023 (DeepMind/EMBL-EBI, 2023). Before AlphaFold, many proteins of pharmaceutical interest had no experimentally determined structure. VR platforms import PDB files, so every AlphaFold structure is immediately available for immersive inspection. The practical effect is that the number of protein targets a research team can examine in VR has expanded enormously.

Real-time molecular dynamics streaming to headsets is the direction the iMD-VR research points toward. The University of York system demonstrated in 2018 that it is technically possible to stream a running simulation to a VR headset and allow interactive steering. The challenge is compute cost and latency. As cloud GPU costs decline and network latency improves, this type of real-time interactive simulation may become more accessible outside research prototype settings.

Integration with AI-generated binding pose visualization is an emerging area. Large language models and structure prediction tools increasingly generate candidate binding poses and protein-small molecule interaction hypotheses. Visualizing these AI-generated results in VR, rather than in a flat-screen viewer, is a natural extension that several platform developers have described in public roadmap discussions.

The risk of over-claiming is worth naming. VR in drug discovery is a genuine tool for a specific set of spatial visualization and communication tasks. It is not going to shorten the 10-to-15-year drug discovery timeline by itself. The Tufts CSDD estimate of $2.6 billion per approved drug reflects the cost of clinical trials, regulatory processes, and late-stage failures that VR does not address. The technology's realistic contribution is in the early research phase: helping teams make better spatial decisions, faster, about which molecules are worth advancing to the lab.

Frequently asked questions

How is VR used in drug discovery?

VR is used in drug discovery to visualize protein structures and candidate drug molecules at human scale. Researchers import protein data files (PDB format) from sources like the RCSB Protein Data Bank or the AlphaFold database, scale the structure up to room size, and physically walk through the binding pocket (the cavity where a drug molecule attaches) to examine spatial relationships that are difficult to judge on a flat screen. Platforms like Nanome support multi-user collaborative sessions, so teams across different sites can annotate and discuss the same molecular structure simultaneously. VR does not replace computational chemistry: it is a visualization and communication layer on top of docking and simulation results from tools like Schrodinger Maestro or OpenEye.

What pharma companies use VR for molecular visualization?

Novartis, Roche, AstraZeneca, and Merck research labs are among the pharmaceutical companies that have publicly documented or partnered around VR molecular visualization. Nanome, the most widely deployed dedicated platform for this purpose, reports over 30 pharma and biotech companies as users. Schrodinger announced a partnership with Nanome in 2021 to allow direct import of Maestro file formats, connecting computational chemistry workflows to VR review. Academic and government research groups at institutions including the University of York have also published peer-reviewed studies on immersive molecular dynamics.

What is Nanome and how do pharmaceutical researchers use it?

Nanome is a VR software platform built specifically for molecular visualization and drug design review. Researchers use it to import protein structures in PDB or SDF format, scale molecules to a walkable size, inspect binding pockets in 3D, and annotate structures with labels and measurements during shared multi-user sessions. Nanome supports headsets including the Meta Quest series and PC-tethered devices. It offers a free academic version and enterprise licensing. Nanome raised a $14 million Series A in 2022 and reports use by over 30 pharmaceutical and biotech companies. More information is available at nanome.ai.

Can VR replace computational chemistry tools?

No. VR does not perform force-field optimization, docking scoring, or molecular dynamics calculations. It has no force field engine and cannot predict how tightly a candidate drug binds to its target protein. What VR changes is how researchers inspect and communicate the results from computational tools like Schrodinger Maestro, MOE (Molecular Operating Environment), or OpenEye OMEGA. The combination researchers describe is: run the computation, bring the top-ranked binding poses into VR for spatial inspection and team discussion, then return to computation to refine. VR is a visualization and collaboration layer, not a simulation engine.

How do researchers import protein structures into VR?

The most common path is downloading a structure file in PDB format from the RCSB Protein Data Bank (rcsb.org) or from the AlphaFold Protein Structure Database (alphafold.ebi.ac.uk), which as of 2023 contains over 200 million predicted protein structures (DeepMind/EMBL-EBI, 2023). That file is imported into a platform like Nanome, which handles the rendering and scaling. Nanome also accepts SDF files for small molecules (candidate drug compounds) and, through its Schrodinger integration announced in 2021, can import Maestro format files directly from a computational chemistry workflow.

Is VR useful for fragment-based drug design?

Fragment-based drug design (FBDD) involves screening very small molecular fragments that bind weakly to a target protein, then growing or linking those fragments into a larger, tighter-binding drug candidate. The spatial reasoning involved in deciding how to grow a fragment inside a binding pocket is exactly where immersive 3D visualization adds value. Researchers examining a fragment's position relative to nearby amino acid residues (the building blocks of the protein) benefit from seeing that geometry at human scale rather than rotating a flat image on a monitor. Nanome documentation and pharmaceutical use-case descriptions cite FBDD as one of the primary workflows where VR interaction surfaces insights that screen-based molecular viewers miss.

Written by Joshua Opolko. I track VR and XR adoption across healthcare and enterprise technology sectors. Statistics sourced to linked references and platform documentation. Verified July 2026.