Knowledge of atomic-level structures of ligand-protein complexes is key for basic research and structure-based drug design. Computational methods have become a valid complement to experiments, but accuracy of predictions generally degrades with the extent of the structural changes associated to binding. Accurate description of ligand flexibility is equally crucial, particularly in a virtual screening (VS) whereby initial structures are often generated without accounting for structural adaptation in binding. To address this issue on the proteins side, we recently introduced gEDES (generalized Ensemble Docking with Enhanced-sampling of pocket Shape), a computational method based on metadynamics to generate bound-like conformations of proteins by only exploiting their unbound structure. In this talk, I will first introduce the gEDES protocol; next, I will talk about SHAPER, an algorithm aiming to generate ligand structures snugly fitting into the binding pocket of a generic receptor, adapting to its geometry. Preliminary results demonstrate that using this dynamic shape-matching algorithm improves the accuracy of VS campaigns compared to flexible docking calculations.