A new approach could help materials scientists identify the appropriate molecules to use in order to synthesize target nanomaterials. The method was developed by Daniel Packwood of Kyoto University’s Institute for Integrated Cell-Material Sciences (iCeMS) and Taro Hitosugi of the Tokyo Institute of Technology. It involves connecting the chemical properties of molecules with the nanostructures that form as a result of their interaction. A machine learning technique generates data that is then used to develop a diagram that categorizes different molecules according to the nano-sized shapes they form.