Abstract
Drug-induced long QT syndrome (LQTS) that may lead to sudden cardiac death has become one of the key reasons for which some drugs fail to enter market, while others have been withdrawn from the market. Early identification of chemical entities causing LQTS is of extreme importance relevant to the production of safer drugs as well as to the direct reduction of attrition rate in drug development. In the present study, we have employed fourteen classification methods to develop a prediction model for cardio-toxicity. The analyses have been carried out on 127 drugs inducing LQTS and 250 cardio-safe drugs. These compounds have been randomly divided into two sets, namely a training set (consisting of 2/3) and a test set (consisting of 1/3). CONCLUSIONS: When models from different algorithms are combined using the proposed method, quality compared to the individual models is consistently and significantly improved in both training and test sets. The accuracy of our approach has 10-25% improvement over the best result obtained by individual classification techniques. The proposed strategy could be employed to infer cardio-toxicity or -safety for current and potential drugs. Certainly, it will also have important impact on decision making in the fields of screening molecules for drug development, biological activity, and other applications as well.
Keywords: Cardio-toxic drug, Long QT Syndrome, Molecular modeling, Molecular descriptors, Classification
4