Integrating machine learning and robotic precision, scientists introduce an integrated approach for computer-augmented chemical synthesis, one that successfully produced 15 different medicinally related small molecules, they state.
Their new, AI-informed, robotically controlled platform has the ability to significantly enhance target-oriented continuously flowing chemical reactions and is a significant step toward completely automated and scalable production of complex molecules. The design and production of complex organic molecules are vital to the discovery of valuable new compounds, such as small-molecule pharmaceuticals.
However, in spite of developments in laboratory automation, the production of complex organic molecules mostly remains a manual process, where the chemists overseeing several labor-intensive steps spend substantial time and effort. For this reason, an automated platform for chemical synthesis that can chart synthetic pathways as well as carry out the flow chemistry needed to synthesize large numbers of new molecules is highly necessary.
Innovations toward such a system have greatly advanced along two parallel tracks; some methods have exhibited success in making the best of AI in compound design, while others exploit automated processes in reaction execution and production. Conner Coley and teammates explain an open-source experimental system that combines these two methods.
The retrosynthesis algorithm of the platform can generalize millions of formerly reported reactions and of doing in silico validation, to put forward successful synthetic paths. After being refined by human chemists, this data is compiled into reusable chemical “recipes,” which are operated on a modular flow chemistry platform that uses a robotic arm to automatically reconfigure and install the unit to conduct the reaction.
Coley et al. exhibit their computer-augmented process by successfully producing 15 different medicinally linked small molecules with increasing complexity.