Computation 100,000 times faster! Quantum computing for accelerated drug discovery

The promise of quantum computing is to solve unsolvable problems. Related companies have made progress in hybrid approaches (methods that combine classical and quantum computing) to address challenges such as drug discovery for incurable diseases.

By using hybrid quantum computing to accelerate drug molecule simulation and modeling, startup Qubit Pharmaceuticals is significantly reducing the time and investment required to identify promising treatments in oncology, inflammatory diseases and antiviral drugs.

Qubit is using the NVIDIA QODA programming model to build a drug discovery platform for hybrid quantum-classical computers and Qubit's Atlas software suite. Atlas creates detailed simulations of physical molecules that are 100,000 times faster to compute compared to traditional research methods.

The Paris- and Boston-based company, founded in 2020, is a member of NVIDIA Inception, a program that provides go-to-market support, expertise and technology to cutting-edge startups. Qubit has one of the largest GPU supercomputers for drug discovery in France, powered by NVIDIA DGX systems. The startup aims to enable pharmaceutical companies to begin testing the first drug candidates discovered through its GPU-accelerated research in 2023.

"By combining NVIDIA's computing power and leading-edge software with Qubit's simulation and molecular modeling capabilities, we are confident in our ability to dramatically reduce drug discovery time and cut its cost by a factor of 10," said Robert Marino, president of Qubit Pharmaceuticals , "This unique collaboration should enable us to develop the first quantum physics algorithms for drug discovery applications."

01Tapping into unprecedented computational power

Computational drug discovery involves generating high-resolution simulations of potential drug molecules and predicting the extent to which these molecules bind to target proteins in vivo.

To obtain accurate results, researchers need to perform extensive sampling and simulate hundreds of different conformations - possible spatial arrangements of molecular atoms. They must also correctly model the molecule's force field, predicting the charge of the affinity, or how one molecule will bind to another.

Such simulations and modeling require high-performance computing, so Qubit chose an in-house supercomputer built from an NVIDIA DGX system and other NVIDIA acceleration servers with a total of 200 NVIDIA Tensor Core GPUs. the supercomputer runs Qubit's Atlas software, completing calculations in just a few hours, whereas traditional methods would take years.

Atlas models quantum physics at the microscopic level for maximum accuracy, and the Qubit team is using NVIDIA QODA to explore a mix of GPU-accelerated supercomputers and quantum computers, where quantum processors could one day accelerate key software kernels for molecular modeling.

Using the NVIDIA cuQuantum SDK, Qubit developers can simulate quantum circuits, enabling the team to design algorithms that are ready to run on future quantum computers.

02AI for every stage of drug discovery

Qubit estimates that while traditional research methods require drug developers to synthesize an average of 5,000 drug compounds before preclinical testing can bring a single drug to market, simulation-based approaches to drug discovery can reduce the number to about 200, saving hundreds of millions of dollars and years of development time.

The company's Atlas software includes AI algorithms for each stage of the drug discovery cycle. To support target characterization, where researchers analyze proteins that play a role in disease, Atlas supports molecular dynamics simulations on the microsecond time scale - helping scientists identify pockets where drug molecules bind to new proteins.

During drug candidate screening and validation, researchers can use AI models to help narrow down potential molecules and generate new compounds. qubit is also developing additional filters for predicting the drug-forming properties, safety and cross-reactivity of candidate molecules.

Reference link：

https://blogs.nvidia.com/blog/2022/11/30/qubit-pharmaceuticals-accelerates-drug-discovery-quantum-computing/