The development of cost-effective synthetic routes to target compounds is performed primarily through searches on large databases of chemical reactions, supplemented with the knowledge and expertise of synthetic chemists. As part of a program to automate all steps in the design and synthesis of small molecules, we desired designed a route-planning tool that could be easily interfaced with downstream automated synthesis systems.  Call SynRoute, it our route prediction software tool that combines rapid searching of 17 million published reactions with the creation of new computer-generated reactions to develop novel synthetic routes for target compounds, enabling the automated generation of high-quality synthetic routes. The searches incorporate minimization of the combined costs of starting materials, solvents, reagents, reaction implementation, while maximizing the overall yield of products to find routes that are efficient and cost-effective. The computer-generated reactions are created retrosynthetically using a set of 62 reliable chemical transformation types that can be applied during the search. We have created a classifier for each of these transformations by using machine learning techniques to determine the success probability of the created reactions, and only high-probability reactions are used in routes. A new codification of reactions was developed that is based on extracting almost all atom patterns from the 16 million molecules available in Reaxys and letting the machine learning algorithm identify the most decisive patterns. The classifiers have an average accuracy of 92% based on cross-validation studies with literature data. SynRoute can typically find routes in a few seconds to a few minutes using a single-core computer, and we have demonstrated its ability to produce new synthetic routes that have been reduced to practice by chemists in the lab. SynRoute offers an intuitive user interface where routes found are graphically displayed. This softwareSynRoute can be an effective tool for reducing the time and costs of synthetic route development and production of drug-like pharmaceutical compounds.<br/>

SynRoute: Synthetic route planning software that combines rapid search and machine learning

256th National Meeting (2018)

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