Drug design is crucial for the effective discovery of anti-cancer drugs. The success or failure of drug design often depends on the leading compounds screened in pre-clinical studies. Many efforts, such as in vivo animal experiments and in vitro drug screening, have improved this process, but these methods are usually expensive and laborious. In the post-genomics era, it is possible to seek leading compounds for large-scale candidate small-molecule screening with multiple OMICS datasets. In the present study, we developed a computational method of prioritizing small molecules as leading compounds by integrating transcriptomics and toxicogenomics data. This method provides priority lists for the selection of leading compounds, thereby reducing the time required for drug design. We found 11 known therapeutic small molecules for breast cancer in the top 100 candidates in our list, 2 of which were in the top 10. Furthermore, another 3 of the top 10 small molecules were recorded as closely related to cancer treatment in the DrugBank database. A comparison of the results of our approach with permutation tests and shared gene methods demonstrated that our OMICS data-based method is quite competitive. In addition, we applied our method to a prostate cancer dataset. The results of this analysis indicated that our method surpasses both the shared gene method and random selection. These analyses suggest that our method may be a valuable tool for directing experimental studies in cancer drug design, and we believe this time- and cost-effective computational strategy will be helpful in future studies in cancer therapy.