PythiaCHEM : a user-friendly machine learning toolkit for chemistry

Machine learning (ML) is currently transforming the field of chemistry by offering unparalleled efficiency in addressing complex challenges. Despite the progress made, a notable gap persists in the availability of user-friendly tools tailored to chemical problems involving small and sparse datasets. Here, we introduce PythiaCHEM, an ML toolkit designed to develop data-driven predictive ML models. It enables the use of various descriptors and ML frameworks for regression and classification tasks in an automated, flexible, and accessible manner through Jupyter Notebooks, making it easy to customize for specific tasks. We showcase the capabilities and versatility of PythiaCHEM in two distinct chemistry tasks: first, the evaluation of the transmembrane chloride anion transport activity of synthetic anion transporters, and second, the prediction of enantioselectivity in the Strecker synthesis of a-amino acids. Our results highlight the utility of PythiaCHEM as a powerful and user-friendly framework for developing predictive ML models applicable in different domains of chemistry.