As part of the efforts to improve their solutions, CAS will integrate Molecule.one's deep learning models for better synthesis predictions, which will enable scientists in exploring potential routes for small molecule synthesis, thereby speeding up scientific advancements
CAS joins forces with Molecule.one to accelerate drug discovery. (Credit: National Cancer Institute on Unsplash)
American Chemical Society’s scientific information solutions unit CAS is teaming up with Molecule.one to expedite drug discovery by jointly developing computer-aided synthesis design technologies.
Molecule.one is a provider of artificial intelligence (AI) based solutions for pharmaceutical chemistry.
By speeding up scientific breakthroughs in early-stage drug discovery, the technologies to be developed by CAS and Molecule.one are expected to help chemists in discovering novel small molecules.
Drawing on their expertise and technologies, such as Molecule.one’s deep learning models and synthesis planning platform, and CAS’ chemical reactions collection, the partners will collaborate to improve chemical synthesis planning.
Molecule.one co-founder and CEO Piotr Byrski said: “We are thrilled to bring to market the first generative models for chemistry trained on CAS’ content, as CAS’ strategic partner for synthetic accessibility and retrosynthesis deep learning enhancements.
“Generative AI has demonstrated amazing feats across multiple fields when trained on large datasets and I believe that there’s a clear need for bringing this to drug discovery.”
The initial joint offering, named M1 RetroScore powered by CAS, is part of Molecule.one’s suite. It replaces the current tool, RetroSAS, for evaluating synthetic accessibility.
M1 RetroScore employs Molecule.one’s advanced generative models and CAS’ chemical reactions data through machine learning to forecast synthesis feasibility of new molecules.
It is said to furnish all users with accessibility scores and corresponding top synthesis pathways. For CAS SciFindern users, the tool links with reference reactions in the platform.
As part of the efforts to improve their solutions, CAS will integrate Molecule.one’s deep learning models for better synthesis predictions. This will assist scientists in exploring potential routes for small molecule synthesis, thereby fast-tracking scientific advancements.
CAS chief product officer Tim Wahlberg said: “By collaborating with Molecule.one, we will bring a unique combination of capabilities to the market to solve challenges in synthesis planning – both early-stage screening for synthesisability as well as laboratory and scale-up synthesis design. These solutions will continue to reduce time spent in the discovery process for scientists, empowering them with efficiencies throughout their innovation journey.”