A novel approach to prediction of supramolecular motifs was applied to more than 6000 monomolecular structures containing 2-periodic H-bond patterns. It is shown that a number of topological descriptors allow one to rationalize supramolecular motifs, find the regularities in their structure, and store the information in a knowledge database. The knowledge database can then be used in an expert system to mimic the work of a human expert and to forecast the method of assembling molecules into supramolecular ensembles and into extended (periodic) architectures. The crystals of N-[(4-methylbenzene)sulfonyl]serine were synthesized, and the principles of the expert system were used to successfully predict the 2-periodic square-lattice H-bond pattern in this compound.