Enhancing autophagy is a potentially effective strategy for the treatment of several human disorders. Therefore, there is a great effort in developing drugs modulating autophagy, and various approaches have been taken towards this goal. Gene expression has been considered an important biomarker for drug activity for prediction of drug mode of action. However, the lack of efficient method of analysis has hampered recognition of drug mode of action based on the analysis of gene expression profiles. A novel and robust tool for prediction of drug mode of action and drug repositioning overcomes the limitations of previously available methods. This novel tool is based on a data set of expression profiles derived from a large number of drugs integrated into a "drug network" constructed by comparing the transcriptional responses induced in human cell lines. Automatic analysis of the topology of the drug network makes it possible to classify compounds and to predict unreported effects of well-known drugs. Using this tool, it was possible to identify fasudil as a new enhancer of autophagy.