| Resumo: | Artificial intelligence has become an indispensable resource in chemoinformatics. Numerous machine learning algorithms for activity prediction recently emerged, becoming an indispensable approach to mine chemical information from large compound datasets. These approaches enable the automation of compound discovery to find biologically active molecules with important properties. Here, we present a review of some of the main machine learning studies in biological activity prediction of compounds, in particular for sweetness prediction. We discuss some of the most used compound featurization techniques and the major databases of chemical compounds relevant to these tasks.
|
|---|