In quantitative structure-activity relationships (QSAR) modeling, the structure determines the modeled property/activity. There are many examples of successful QSARs developed for engineered nanoparticles (nano-QSARs), where the structural descriptors reflect the features of atoms/molecules constituting the nanoparticle or/and describe its supramolecular “nanostructure” (i.e., size, shape, coating). However, modern nanotechnology proposes more comprehensive nano-systems (e.g., nanoparticles with chemically different core, coating, and surface modifications). Moreover, from the recent progress of nanotoxicology, we have learned that the biological activity of nanoparticles can change during their lifetime because their structure varies depending on the external conditions (e.g., pH, ionic strength, interactions with proteins). Therefore, there is a question, how to appropriately represent nanostructure in more universal nanoinformatics models? In this talk, we will closely examine the currently proposed concept of nanoparticle representation. Then, we will discuss how to model multicomponent nanomaterials and implement various strategies for considering system dependency in the structure and modelled properties.