For over five decades now, chemists have strived to “teach” computers the art of synthetic planning. However, despite numerous and often ingenious approaches being pursued, there have been no examples of synthetic routes first planned by the machine and then successfully executed in the laboratory. This state of affairs might have reflected the incompleteness of the chemical knowledge encoded in the programs or the inadequacy of algorithms and hardware available in these pioneering efforts. With enormous improvements in computers’ performance, with the modern methods of machine learning and graph theory and, admittedly, with a tedious process of humans encoding various reactivity rules, our team has been able to develop the so-called Chematica platform that is finally capable of  completely automated and chemically correct design of synthetic pathways leading to arbitrary targets. In my talk, I will first discuss the theoretical basis of Chematica and then describe an experiment where this software designed, without human supervision, syntheses leading to eight commercially valuable and/or medicinally relevant targets  – in all cases tested, Chematica offered significant improvements on previous approaches or identified efficient routes to targets for which previous synthetic attempts proved unsuccessful. I will also describe the first demonstration of Chematica planning a long, multistep route leading to a relatively complex natural product which our experimentalist are now completing in the laboratory. With these encouraging results at hand, I will also narrate the challenges for future research in this area and the improvements in the software (and the underlying algorithms) we envision.<br/>

Computational design and experimental validation of synthetic routes created by Chematica

256th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.