Reaxys Predictive Retrosynthesis: Redesigning the approach to synthetic chemistry
Discover an award-winning innovation in synthesis route design and prediction
GHP has chosen Reaxys Predictive Retrosynthesis as its 2020 Life Sciences Award winner. They state, “By combining Reaxys’ content with cutting-edge AI and machine learning technologies developed by PAI, the team is able to offer solutions unparalleled by competitors in the market.”
In pharmaceutical and chemical development, synthesis of molecules can be a bottleneck. Getting from a desired novel molecule via an optimal synthesis route takes considerable time, effort and expertise.
Reaxys Predictive Retrosynthesis drives innovation, increases success rates in synthetic chemistry and provides faster answers for a broader range of molecules than any existing solution. Developed on the basis of ground-breaking work published in Nature, this new solution leverages cutting-edge deep learning technology and high-quality reaction data of Reaxys to produce blueprints for the sequences of reactions needed to create small organic molecules.
Get access to Reaxys Predictive Retrosynthesis?
Why choose Reaxys Predictive Retrosynthesis
Reaxys Predictive Retrosynthesis is faster, broader, more flexible and smarter than any existing tool. Its novel approach auto-designs synthetic routes for known and novel substances. This new decision-support solution delivers multiple benefits for researchers and companies:
- Drive innovation: Find innovative routes to synthetize novel compounds and find novel, optimized routes for existing chemical compounds
- Support decision making: Make informed decision about which chemical compounds to make and how to make them
- Save time: Predicts route 30 times faster and for twice as many molecules compared to other computer-aided route design solutions
- Customizable: Flexibility to integrate users own custom building block libraries and proprietary chemistry reaction data
Reaxys Predictive Retrosynthesis solution’s development
This revolutionary innovation in retrosynthesis was designed on the basis of ground-breaking work published in Nature and adapted to meet the needs of the chemists and make the interaction between the chemist and AI powered retrosynthesis easy and intuitive. AI and machine learning was applied to high quality Reaxys data to build a solution that empowers scientists to advance drug discovery, agro-chemicals, fine chemicals and development of other chemical compounds.
Reaxys Predictive Retrosynthesis leverages:
- The high-quality reaction data in Reaxys to train the predictive model
- Three deep neural networks to auto-derive and apply reaction rules to predict the best synthesis routes
- Monte Carlo tree search approach to eliminate non promising synthesis routes early to speed up predictions
How Reaxys Predictive Retrosynthesis is different
The deep learning technology underlying Reaxys Predictive Retrosynthesis has digested data on over 15 million reactions to auto-derive up to 400,000 reaction rules. It uses these instead of hand-encoded rules to design and predict synthesis routes for known and novel molecules.
Reaxys Predictive Retrosynthesis is unmatched by existing solutions in multiple ways:
- Non-reliant on hand-encoded rules: Thousands of reaction rules are auto-derived from the Reaxys data, enabling it to be non-dependent on manual effort, which are traditionally used in other solutions
- Tailored: Reaxys Predictive Retrosynthesis can be tailored to meet the requirements of a company by being flexible to allow the integration of propriety custom building blocks and/or reaction data
- Self learns: Model creation and update is fast allowing to ‘self-learn’ from the rapidly ever-growing chemistry knowledge
- Tested: Predicted synthetic routes were double-blind tested with 45 organic chemists from around the world and have proven to be scientifically sound and robust
Would you like to get access to Reaxys Predictive Retrosynthesis?
Contact our Consultants