Department of Computer, Control and Management Engineering

Despite the drug approval process consists of extremely rigorous clinical and preclinical studies, not all side effects are identified before its marketing, posing a significant risk to public health. Furthermore, considering the huge use of economic and human resources, in-silico predictive approaches for the identification of side effects are essential. In this study, we introduce a new method based on random walk with restart algorithm to delineate previously unidentified links between drugs and side effects, and we apply it on the drug-induced Asthma and long QT syndrome. We identified the genes potentially involved in the development of the analyzed side effect by comparing side-effect-related drugs with drugs not known to induce side effects. Analyzing the sets of genes most likely influenced by the perturbation of each individual drug, we observed that, on average, side-effect-related drugs perturb a higher percentage of genes involved in the development of side effects compared to side-effect-unrelated drugs. Based on this finding, we developed a classifier to explore all possible unknown associations between drugs and side effects. This method can be extended to the analysis of other side effects as well.