Sequencing of the human genome along with developments in combinatorial synthesis and high-throughput biological screening provide unparallel opportunities to drug discovery. It has been noted that the increased number of synthesized and annotated compounds did not yield the expected increase in number of viable drug candidates. To address this problem, several novel computation technologies have emerged for making combinatorial library design cost-effective. Of particular interest for the modern drug discovery are the structure-based or target-based methods that use structural information about target proteins and their small molecule ligands. In this work, we provide an overview of selected advances in computational algorithms for the rational selection of molecule libraries for the synthesis, with emphasis on structure-based approaches. These include a fusion of scaffold-linking method and combinatorial library design, pharmacophore matching and informative library design, and search by 3-D tree topological descriptors.