STREAM · Statistical Tools for Reaction Efficacy AssessMent: Prediction and Understanding in Organocatalyst Discovery

Statistical Tools for Reaction Efficacy AssessMent: Prediction and Understanding in Organocatalyst Discovery

The need for chiral compounds, often as single enantiomers, has escalated abruptly in recent years, driven particularly by the demands of the pharmaceutical industry. For example, two-thirds of prescription drugs are chiral, with the majority of new chiral drugs being administered as single enantiomers. Moreover, chiral compounds have found use as agricultural chemicals, flavors, fragrances, and materials. In response to this widespread demand, chemists have recorded impressive successes for the preparation of stereochemically pure compounds. Both small molecule catalysts and enzymes have been developed for such purposes. However, it was only relatively recently that such asymmetric catalysis, with enantiomeric excesses approaching 100%, was achieved with synthetic catalysts. Therefore, innovations in asymmetric reaction/catalyst development represents a goal of considerable significance for world-wide health and would improve the lives of millions of people. Thus, the overarching objective of the action was to develop data science-driven workflows that enable the prediction of experimental enantioselectivity. In these studies, we concluded that multivariate linear regression can correlate the structure of every reaction component (substrate, catalyst, and conditions) to enantioselectivity data. The resulting statistical models can anticipate how changing the substrate and catalyst structure alters the reaction outcome. We showed that these general reaction models are accurate in predicting out-of-sample, and provide guidance in reaction application to include additional substrates.

Текст от CORDIS, на английски · Данни: CORDIS, © Европейски съюз