How IBM, Cleveland Clinic, and RIKEN are exploring chemistry research with quantum-centric supercomputers
New infectious diseases emerge annually, threatening millions of people around the world. Meanwhile, existing pathogens evolve resistance to treatment. Medical science can struggle to keep up.
Researchers turn to high-performance computers to identify promising chemical compounds for medicine by studying their interactions with the human body. However, the medical industry faces challenges as it works to develop new drugs and fight disease.
At present, the average cost to bring a new drug to market is ~$2 billion.
New regulations, cost constraints, and expiring patents lead to fewer new drugs for the same amount of R+D effort.
Preclinical studies may fail to reveal toxicity, wasting resources if flagged later in the development arc.
Today’s chemistry simulations require vast computational resources to accurately simulate large molecules.
Developing new drugs amid these challenges requires more efficient research and development methods.
When researching molecules, scientists need to isolate them and model their underlying composition. They also must anticipate how they will interact with other compounds in the human body.
Click on the tiles below to learn about different key molecular properties of interest for medicinal chemistry.
Supramolecular interactions:
Supramolecular interactions are those between different molecules, such as interactions between molecules that make up the drug and a target protein found in pathogens or the human body.
Researchers and industry leaders worldwide are developing quantum-centric supercomputing techniques for challenging problems. For example, startup QunaSys (opens in a new tab) offers a function (opens in a new tab) to users of IBM Quantum Platform, that implements a quantum-centric technique for chemistry called quantum-selected configuration interaction, or QSCI.
Partnering with RIKEN, IBM explored one method for simulating chemistry with quantum-centric supercomputers called sample-based quantum diagonalization (opens in a new tab), or SQD. It uses quantum to generate measurements corresponding to electronic configurations, then uses classical to process those configurations into an answer robust to quantum computing noise. This technique has the potential to outperform what classical approximation methods could do alone.
Take an iron-sulfur cluster [4Fe-4S], for instance. On today’s pre-fault-tolerant quantum computers alone, it would take 3 million years to model. SQD offers a way to study this compound before the emergence of fault-tolerant quantum computing.
A single quantum computer running a VQE algorithm
An error-corrected quantum computer running phase estimation
IBM Heron processor + RIKEN’s Fugaku supercomputer running SQD
As you can see, we don't have to wait for fault-tolerant quantum systems to start exploring chemistry applications with quantum. Cleveland Clinic is extending the RIKEN work to run molecular simulations relevant for drug discovery. New work combines quantum and classical using SQD to obtain information on the energies of molecules in a way that is robust to the noise inherent to quantum computation.
Explore the demos below to see how quantum-centric supercomputing tackles two critical chemical research tasks: modeling supramolecular interactions and conformational energies. They compare SQD with state-of-the-art classical methods used to approximate these properties today.
A classical computational chemistry method that approximates the energies of complex systems by focusing on all of the possible excitations from a set of predetermined orbitals
A quantum-centric supercomputing approach that first calculates the energy of a system on a quantum computer, and then produces a more accurate answer with an iterative classical correction.
Average distance between points
2.54e-08 kcal/mol
The small average difference between the energies from each method demonstrates excellent agreement between the quantum and classical methods for this example.
These demos show that the output of quantum-centric supercomputing methods can already match the precision of leading classical methods for certain use cases.
Quantum-centric supercomputing can go beyond the limitations of either quantum or classical computing alone and has the potential to reduce computational load and the costs required for analyzing drug compounds and interactions. These advances will help accelerate the industry and usher in a new era of computing.
Want to explore the potential benefits of quantum for your industry?
Explore IBM Quantum Platform to get access to cloud-based quantum hardware, news, and world-class learning material. The Qiskit Functions catalog helps developers accelerate their workloads with a suite of abstracted services for key research applications.
