Hypernet Labs has begun work with scientists at the Open Force Field Initiative (OpenFF) to make foundational improvements to the tools used for pharmaceutical research and other applications of quantum chemistry.

OpenFF is a collaborative effort involving academic and industry researchers who seek to develop automated and systematic data-driven techniques to parameterize and assess new generations of more accurate force fields. The simulations that model interactions between proteins and small drug molecules rely on quantum chemistry calculations, and force fields are the models that guide the physics of these simulations. (Check out the medical section of our Magazine to read more about Galileo’s use in rational drug design, drug screening, and molecular docking simulations that are helping researchers find effective antivirals to fight COVID-19.) 

Significantly, OpenFF scientists are looking to perform this work in a way that allows other researchers to very easily use their methods and data. While it’s currently relatively easy to share data files, it’s more difficult to do the same with methods. At times, the calculations are so complex and computationally intensive that there is a limit to what you can expect other scientists to reproduce, even simply for the purpose of verifying results. The collaboration between the OpenFF and Hypernet Labs aims to solve some of the problems of reproducibility and the sharing of methods.

Jeffrey Wagner, postdoctoral researcher in the Department of Pharmaceutical Sciences at the University of California at Irvine and senior chemist and software scientist at Open Force Field Initiative explains: “One of the great things about working with Hypernet Labs and specifically the Galileo platform is that it ratchets up the scale of what we can expect people to reproduce.”

“One of the great things about working with Hypernet Labs and specifically the Galileo platform is that it ratchets up the scale of what we can expect people to reproduce.”

Specifically, Hypernet Labs is working with OpenFF on the QCFractal infrastructure. The goal is to facilitate the customization of the workflow developed by OpenFF so users can obtain an excellent model of their specific molecule of interest.  This will allow them to get the most accurate possible results from a drug candidate simulation. For everyone involved in OpenFF–researchers and pharmaceutical companies–, scientists’ ability to generate these results with their own input is extremely valuable.

As Dr. Wagner explains, QCFractal has two important roles:

1. It allows for the sharing of datasets. There’s a public QC archive and interface where data can be pulled in minutes using a simple python prompt.
2. It’s also a task distribution engine, which helps to “iron out the wrinkles” in all of the different academic supercomputers. This enables cooperation between labs at different universities as they share the work of calculating the energy of small molecules.

Collaboration with Hypernet Labs is helping OpenFF to “create an on-ramp” to the workflow they are building with QCFractal. The aim is to empower other researchers to use the workflow in a very simple way instead of going through the labor intensive process of setting up the infrastructure from scratch.  Instead, OpenFF can point them to Galileo, which will include pre-made components allowing researchers to spin up a server, send the work, retrieve results, and shut down. Wagner explains, “The user could have results at basically maximum velocity with zero complexity.”