How much time do you lose focusing on molecules that cannot be synthesized?
M1 RetroScore powered by CAS is a synthetic accessibility scoring tool created to save your time by using deep learning models trained on CAS chemical reaction content to predict the likelihood of synthesis for novel small molecules.
How does it work?
M1 RetroScore powered by CAS enables you to estimate how hard it is to synthesize a given compound allowing you to focus on drug candidates with the lowest synthesis costs and the highest probability of success.
The underlying generative model does this by looking for an actual synthesis pathway that must be executed to obtain the target compound. It considers many different synthetic pathways, choosing the most promising one for target synthesis. These various pathways are evaluated and assigned a score on a scale from 1 (easiest) to 10 (hardest) based on four key factors:
Number of steps in the synthesis pathway
Availability and cost of starting materials
Feasibility of reactions in the pathway
Order of the reactions
(risky reactions ideally take place at the start of a given pathway)
Create batch synthetic accessibility scoring of target compounds
Search and score hundreds of thousands of compounds using our generative or template-based ML models trained on CAS chemical reaction content via UI or API, and easily export results to other applications
View the top synthetic pathway provided to back up score
Reactions are further supported by literature evidence from CAS SciFinder^n
Choose from multiple starting materials datasets
Select from the default CAS starting materials dataset of ~60M compounds or upload your own
Secure platform and complete data ownership
Safely engage with proprietary data with a dedicated deployment, isolated cloud resources, and multiple layers of security
Adjust search parameters based on specific needs
Select from different deep learning models for diverse approaches to reaction generation, define the expected synthesis pathway length and more
Leverage the API to process hundreds of thousands of compounds a month and evaluate candidate drug structures.
Reduce your workload by quickly prototyping synthesis pathways for many molecules in a short amount of time.
Save time and make informed decisions using models trained on very large datasets to screen targets for synthesis.