DREAMM is a novel machine learning tool that predicts the protein-membrane interfaces of peripheral membrane protein, and optionally predicts binding sites near the predicted membrane-penetrating residues in protein conformational ensembles. As an output, the user can retrieve the membrane-penetrating residues in a .csv file and if the user's choice was to predict binding sites, a .zip file will be downloaded including the abovementioned .csv file, the binding pocket predictions, the visualizations, and summarized the binding site clustering results.