Sometimes the complexity of a system, or the properties derived from it, do depend neither on the individual characteristics of the components of the system nor on the nature of the physical forces that hold them together. In such cases the properties derived from the “organization” of the system given by the connectivity of its elements can be determinant for explaining the structure of such systems. Here we explore the necessity of accounting for these structural characteristics in the molecular descriptors. We show that graph theory is the most appropriate mathematical theory to account for such molecular features. We review a method (TOPS-MODE) that is able to transform simple molecular descriptors, such as logP, polar surface area, molar refraction, charges, etc., into series of descriptors that account for the distribution of these characteristics (hydrophobicity, polarity, steric effects, etc) across the molecule. We explain the mathematical and physical principles of the TOPS-MODE method and develop three examples covering the description and interpretation of skin sensitisation of chemicals, chromosome aberration produced by organic molecules and drug binding to human serum albumin.