Practical implementation of Predictive Retrosynthesis in Reaxys Chemistry Database

The Department of Chemistry Presents:  Dr. Irakli Samkurashvili PhD, Application Scientist, Life Sciences

Dr. Samkurashvili will discuss the practical implementation of predictive retrosynthesis in Reaxys Chemistry Database. Synthetic Chemists would embrace a tool which accelerates both new reaction discovery and the synthesis of new molecule. Numerous attempts have been made to develop such tools, generally called Computer Aided Synthesis Planning or CASP, for 40 years. Although the CASP idea has been very attractive conceptually the practical implementation of CASP has been burdened with technical challenges. One of the biggest problems was to define rule set on which predictive algorithm is based. Traditional approach where manually coded rules have been used proved to be less than ideal solution for predictive retrosynthesis. The latest advances in the AI research and the significant increase in computational power contributed in the development of new neural network based algorithms. The combination of these algorithms with the high quality training datasets seem to produce promising predictive retrosynthesis tools (RT) with very high prediction accuracy. Novel AI based Predictive retrosynthesis engine has been successfully integrated in the Reaxys chemistry database. We will discuss typical examples of how to use Reaxys Synthesis Planner and Retrosynthesis Tool to quickly identify synthesis plans with high degree of accuracy. The discussed scenarios would be very useful for chemists and can give valuable insight how predictive retrosynthesis works.

BIO

Irakli Samkurashvili has been an Application Scientist at Elsevier since 2012. He is supporting Elsevier’s Life Sciences Portfolio products including Reaxys, Embase, Pharmapendium, and EmBiology. Irakli is passionate about helping Chemists, Biologists and Information Professionals to effectively utilize specialty databases in chemistry and the life sciences to advance their research. Prior to joining Elsevier Irakli worked in bioinformatics field at Ariadne Genomics and Informax Inc. His academic background includes post-doctoral fellowship at NIH and Graduate degree in Molecular Genetics from University of Cincinnati.