Quantum-based startup QSimulate has announced a collaboration with Amgen focused on integrating large scale, high accuracy, quantum mechanics calculations (QM) into the drug discovery pipeline. The impetus for the new collaboration is the development by QSimulate of a novel implementation of density functional theory, a variant of QM, that allows these calculations to be run with discovery-relevant throughput on realistic ligand/protein models for the first time by taking full advantage of cloud resources.
According to Narbe Mardirossian, who is leading the effort for Amgen, “we have known that quantum mechanics has the potential to outperform existing computational methods when it comes to certain important questions in drug discovery. However, issues related to cost and speed have left accurate QM methods out of reach. QSimulate’s novel implementation has the potential to change that.”
“The QSimulate advantage,” explains Toru Shiozaki, QSimulate CEO, “is that we have entirely reenvisioned and reprogrammed how accurate QM calculations are performed, so that they can now be run efficiently in the cloud, taking advantage of the huge amounts of compute power and memory available there.” The result is the ability to practically perform QM calculations on thousands of processors, using terabytes of memory. This frees scientists to run these calculations with fast throughput on much larger systems than ever before. In the realm of drug discovery, this opens up the possibility of applying QM to full ligand/protein models consisting of thousands of atoms without needing the crude approximations used in the past. According to Shiozaki, “in our benchmarks, a computation of the ligand binding energy for an entire 2500-atom ligand/protein system, within a high accuracy QM representation, can be run on AWS in a fraction of an hour and costs less than $10.”
QSimulate and Amgen plan to initially apply the new approach to ligand/protein systems that have previously been characterized using other non-quantum mechanical approaches, to validate the advantages that QM brings to important questions in drug discovery. They plan to publish this work shortly thereafter.
Professor Garnet Chan of Caltech and QSimulate co-founder points out that “QSimulate plans to make their software and expertise available not only to the field of drug discovery, but also to such areas as material science and chemicals. Broadly, we believe we are just at the early cusp of where the integration of quantum calculations with huge cloud resources will help change the future.” They see their new approach useful for both direct application, as with Amgen, and as part of a broader computational plan. For example, QSimulate has agreements in place to facilitate the development of artificial intelligence models, and to help design the platforms that will be used in next-generation quantum computers.
QSimulate collaborated with Amgen as part of a research project within the Amazon Quantum Solutions Lab. The lab helps customers research the performance and viability of early-stage quantum computers and quantum applications, as well as address business problems through a combination of classical and quantum-inspired solutions using machine learning and high-performance computing resources on AWS. Eric Kessler, Head of Business Development for Quantum Computing at AWS, notes “We are proud to have facilitated QSimulate’s work with Amgen on solutions that could have broad impact on the drug discovery process. Their results illustrate how the Amazon Quantum Solution Lab can help customers combine the power of highly scalable classical computing resources with innovative quantum algorithms and, ultimately, prepare for the arrival of scalable quantum computing.”