A Structure-Based Platform for Predicting Chemical Reactivity

Despite their enormous potential, machine learning methods have only found limited application in predicting reaction outcomes, because current models are often highly complex and, most importantly, are not transferable to different problem sets. Here, we present a structure-based machine learning platform for diverse applications in organic chemistry. Therefore, an input based on multiple fingerprint features (MFFs) as a versatile molecular representation was developed that was shown to be applicable over a range of diverse problem sets. First, molecular properties across a diverse array of molecules could be predicted accurately. Next, reaction outcomes such as stereoselectivities and yields were predicted for experimental datasets that were previously evaluated using (complex) problem-oriented descriptor models. As a final application, a systematic high-throughput dataset was investigated as a real-world problem, and good correlation was observed when using the structure-based model.