Back in June 2023, Microsoft announced an aggressive plan to leverage the combined capabilities of high-performance computing, quantum computing, and AI, and make them readily accessible via the Azure cloud to the chemical and material science sector to help accelerate new discoveries and innovations.
The platform behind the plan was called Azure Quantum Elements, and Microsoft CEO Satya Nadella said at the time that the company was looking “to compress the next 250 years of chemistry and materials science progress into the next 25.” With the first of those 25 years now complete, Microsoft recently added two new tools to Azure Quantum Elements to further help scientists in this arena.
One of those new tools is an end-to-end workflow called Generative Chemistry, which puts generative AI capabilities to work to help scientists looking for the right molecules and molecular compounds to use for a particular application, a process which normally is like looking for a needle in a haystack, given that there are hundreds of millions of such compounds and substances available, with many more waiting to be discovered, Microsoft said. The other tool, which can be used in combination with Generative Chemistry, is Accelerated Density Functional Theory (DFT), which enables HPC simulations of a molecule’s electronic structure to make it easier and faster for scientists to identify the various properties of those molecules.
The complexity of these processes and the notion that new combinations of substances could result in new drugs that have broad societal impact makes this an ideal scenario for leveraging HPC, quantum computing, and AI to speed things up.
“Many problems facing the world today boil down to chemistry and material science problems,” a Microsoft spokesperson told Fierce Electronics. “Innovations in chemistry and material science are estimated to have an impact on 96 percent of all manufactured goods, which impacts everyone.”
Generative Chemistry allows scientists to enable that change in a handful of steps. First, researchers submit information on their desired molecular characteristics to be used for a specific application. The Generative Chemistry tool then generates seed molecules from a dataset, which are then used to initiate the guided AI generation of candidate molecules for your application. Users have the choice of selecting the most relevant generative AI model for their purposes, as well as the ability to screen compounds for toxicity, among other options.
“AI-based screening models predict properties of the candidate molecules that are important for real-world applications—such as boiling point, density, or solubility,” Microsoft said in a blog post. “A feedback loop sends this information back to the guided AI generation [stage] to modify the selection of candidate molecules. In this step, you also have the option to fine tune the AI models for your specific application.”
AI-guided synthesis planning is then used to determine the feasibility of making the molecules in a laboratory. Filtering candidates based on how easy they are to make will allow scientists and the labs and companies they work for to figure out whether or not they can be efficiently produced.
HPC simulations are then performed on the top candidates, and at this point Accelerated DFT to quickly–within hours–determine electronic properties of these candidates, such as dielectric constant, ionization potential, and polarizability. Lastly, Microsoft said, “You are presented with the final candidate molecules from which you can select the most promising for laboratory synthesis and testing.”
The blog post added, “This entire process takes only days, shaving months or even years off trial-and-error laboratory experiments that were previously required to arrive at this point. Generative Chemistry suggests entirely new compounds and gives scientists the freedom to focus on only those molecules that are fit for the desired purpose—saving time, money, and effort.”