New Virtual Reality Tool Simplifies Structure-Based Drug Design
By Deborah Borfitz
March 27, 2020 | A virtual reality (VR) platform developed by researchers at the University of Bristol is allowing chemists, computational scientists and structural biologists to collaboratively design molecules like engineers create airplane components. “It’s bringing human intuition to bear in a new way,” according to Adrian Mulholland, a professor in the university’s Centre for Computational Chemistry.
The newest task for the interactive molecular dynamics simulations in VR (iMD-VR) tool is to help a seven-member team at the University of Bristol predict how potential drugs will bind to the coronavirus disease 2019 (COVID-19) targets, Mulholland says. “It’s amazing how quickly the international scientific community has responded to the COVID-19 outbreak,” he adds. “Structures of proteins from the virus have been solved by X-ray crystallography and Google DeepMind has predicted the structure of other [culprit] proteins, and people are starting to try to assemble models of the virus.”
VR has been touted as a drug development tool for two decades but dismissed by some scientists as little more than a toy for looking at molecules and proteins, “not a genuine research or education tool,” Mulholland says. Sentiments have been changing with the maturation of VR and its application in training everyone from surgeons to architects.
Gaming technology has been the big game changer when it comes to using VR for structure-based drug design, he continues, with the development of VR technology such as Facebook’s Oculus. “We are now able to not just look at a molecule but also to manipulate it.” Many small-molecule drugs work by binding to proteins, thereby stopping a virus from reproducing, but they need to be precisely positioned to fit snugly to their biological targets, he explains.
The cost of setting up a VR system has also come down dramatically from a million-dollar investment in large, three-dimensional immersive spaces to about $5,000 for an iMD-VR setup to cover the cost of the hardware (one headset and a set of paddles) and graphics card to run simulations. The system is based on free, open-source software (Narupa interface), described last year in the Journal of Chemical Physics, and uses easily accessible room-scale VR headsets, such as the HTC Vive, says Mulholland.
What makes the iMD-VR tool unique is that it enables multiple users to be in the same virtual environment, enabling collaboration across labs and geographies, he says. Its lead developer is David R. Glowacki, a research fellow at the University of Bristol with a joint position in chemistry and computer science. Hundreds of people have done user tests of the technology, many at industry conferences.
Use of VR is taking off at universities around the world. Nanome has a VR docking program it describes as a “Minecraft for matter” that researchers are using to visualize and directly manipulate atoms, molecules and proteins. The lab of Jacob Durrant, an assistant professor of computer-aided drug design at the University of Pittsburgh, is developing VR software enabling students and researchers to walk around, and through, their favorite proteins, Mulholland says. The University of Tromsø in Norway has a dedicated VR lab and one is being built at the University of Copenhagen, he cites as examples.
It won’t be long before VR is the customary way drug designers work on many drug development problems, Mulholland predicts. Being able to view small molecules as three-dimensional objects simplifies the task of modifying them so they more tightly fit within the “keyhole” of a protein binding site.
In a recent study published in PLOS One, University of Bristol researchers demonstrated that even in the hands of nonexperts the iMD-VR system could be used to easily dock a small drug molecule to influenza neuraminidase and HIV protease, recreating the binding poses hypothesized by X-ray crystallography.
In previous work, Mulholland, Glowacki and coworkers carried out user tests that showed that iMD-VR was better for molecular modelling tasks than traditional computer-and-mouse methods. It’s likely that the newer technology performs much better because it’s the more intuitive way to deal with a molecule, he adds. “You really get a sense of [a molecule’s] three-dimensional structure and… literally reach out and grab each end and move it with your hands, and you can also walk around the problem.”
iMD-VR users in the study were able to accurately unbind and rebind drugs from protein targets in less than five minutes of real time, he notes. It’s the combination of VR with interactive molecular dynamics that makes all the difference—a drug design process he likens to playing the Nintendo game 3D Tetris. “Using human intuition and a bit of chemistry knowledge [anyone] can do a real good job in VR of predicting how a drug binds.”
Educational Value
Using iMD-VR requires “almost no training at all,” says Mulholland. After a brief demonstration, undergraduate students in a teaching lab at the University of Bristol can effectively use a VR headset and paddles to manipulate molecules and learn how they bind to proteins. Even high school students have completed sophisticated molecular modeling tasks “relatively easily and well.”
Plastic models of molecules traditionally used in chemistry class are quite informative but can break and don’t move properly, says Mulholland. With iMD-VR, students can “physically and correctly see how molecules behave, how they move and how they react” either by using the tool themselves or watching a live demonstration by their instructor. “In principle, everyone could be manipulating the same molecule at the same time.”
In Mulholland’s lab, iMD-VR is currently being used for binding molecules to proteins involved in antibiotic resistance as well as to find drugs for COVID-19. In collaboration with BP, researchers are also modeling catalysts in VR trying to understand how to make molecules in cleaner processes, consuming less energy and generating less waste (aka “green chemistry”), he says.
Two years ago, with support from Oracle, University of Bristol researchers were using VR cloud-based tools that allowed several people to interact with molecules in the same virtual space at the same time, says Mulholland. It successfully demonstrated that a simulation run in Frankfurt on Oracle Cloud Infrastructure at the company’s German center could be visualized in Bristol, allowing geographically dispersed scientists to collaborate on molecular modeling tasks.
The University of Bristol subsequently did some interactive data sharing with other universities, he adds, and is looking to build on this model. Such collaborations benefit from cloud mounting of the framework and sharing of computational resources.