With the continuously increasing amount of available compound, bioactivity and structural data, machine learning (ML) gained momentum in drug discovery and especially in ligand-based virtual screening (VS, see also description of Workflow 4) to predict the activity of novel compounds against a target of interest. In this workflow, the following steps are shown: 1.Split dataset into active & inactive compounds (pIC50 cutoff = 6.3) 2.Train ML classifiers using random forests, support vector machines & artificial neural networks 3.Apply k-fold cross validation 4.Evaluate models with ROC curves Author: Dominique Sydow, Michele Wichmann, Jaime Rodríguez-Guerra, Daria Goldmann, Gregory Landrum, and Andrea Volkamer