Investigation of seq2seq neural networks applied to prediction of novel organic Chemical reactions

• Grant number: 21/02584-6
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: June 01, 2021
• End date: May 31, 2022
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Paula Dornhofer Paro Costa
• Grantee: Ingrid Valverde Reis Zreik
• Host Institution: Faculdade de Engenharia Elétrica e de Computação (FEEC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil

Abstract

Research guided by large volumes of data and machine learning has shown great potential in revealing novelties and bringing significant advances in several areas of basic and applied sciences. Organic Chemistry and, more specifically, Organic Synthesis has benefited from this new approach concerning the prediction of organic reactions and retrosynthesis. Several advances are described in the literature, but there is still ample room for progress. This project is a multidisciplinary experience involving Computing and Chemistry with the main objective of gaining mastery over the techniques of artificial intelligence applied to the prediction of organic reactions and consequent detection of new synthetic methods not yet described in the literature. The focus of the project is the understanding and application of the template-free seq2seq model in the prediction task, together with the study of its limitations and potential. In addition to exploring this model, the project proposes, through its partnership with a co-supervisor in the specialist area, to test some of the predictions generated in the computational context in the laboratory for experimental chemical confirmation. (AU)