Many powerful open-source programs for computer-aided drug discovery (CADD) are designed with expert computational chemists in mind. Because such tools often prioritize functionality over usability, experimentalists and educators are sometimes reluctant to engage with them. Server applications encourage broad application by running calculations on remote resources ("in the cloud"), accessible via browser-based graphical user interfaces (GUIs). But these excellent community resources have some notable disadvantages: changes in funding can complicate long-term maintenance, impacting reproducibility; lengthy queues are often required to accommodate many users; and users must often upload their proprietary molecular structures to a third-party server. “Browser apps” address these challenges by leveraging users’ web browsers as secure computing environments. These apps are also accessed through a browser, but rather than accept user data, they provide the software required to run calculations locally in the browser’s secure "sandboxed" environment. Thanks to this de facto distributed approach, browser apps do not require a difficult-to-maintain remote computer infrastructure, and users never need to wait in queues or send sensitive intellectual property to a third-party server. We here present three of our lab’s proof-of-principle browser apps: Webina, an implementation of the AutoDock Vina docking software; ProteinVR, a system for visualizing molecular structures in virtual reality; and DeepFrag, a deep-learning model to guide lead optimization. Each of these three examples illustrates different aspects of our approach, including: compiling binaries written in C/C++ to WebAssembly, a browser-compatible binary-code format; building on existing web APIs such as WebXR, WebGL, and WebGPU; and transpiling Python scripts directly to JavaScript. A number of recent consumer products (e.g., Office 365, Chrome OS) leverage the browser as a secure system for delivering and running software. By similarly implementing key CADD tools as browser apps with user-friendly GUIs, we aim to broaden the accessibility of these powerful computational-chemistry approaches. Aside from advancing research, these brower-based tools are also well suited to classroom use in that they do not require separate installation and are compatible with many devices.

Head out of the clouds: Computational-chemistry tools that run locally in a web browser

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.