There is currently a push towards big data and data mining in materials research to accelerate discovery. Zeolites, metal-organic frameworks and other related crystalline porous materials are not immune to this phenomenon, as evidenced by the proliferation of porous structure databases and computational gas-adsorption screening studies over the past decade. The endeavour to identify the best materials for various gas separation and storage applications has led not only to thousands of synthesized structures, but also to the development of algorithms for building hypothetical materials. The materials databases assembled with these algorithms contain a much wider range of complex pore structures than have been synthesized, with the reasoning being that we have discovered only a small fraction of realizable structures and expanding upon these will accelerate rational design. In this Review, we highlight the methods developed to build these databases, and some of the important outcomes from large-scale computational screening studies.