Is it possible docking and scoring new ligands with few experimental data? Preliminary results on estrogen receptor as a case study