s MIA-QSAR Modeling of the Anti-HIV-1 Protease Activities and Cytotoxicities of Ritonavir Analogs

Drug-likeness and toxicity prediction of compounds are so important as to estimate their bioactivities. In rational design of drugs, looking for safe rather than only highly active synthetic targets has increasingly became mandatory. In this context, structure-based methods to model toxicities of drug-like compounds arise as fundamental tasks to achieve safer drugs. Accordingly, the MIA-QSAR method, which has been widely applied to model bioactivities of several classes of compounds, can also be used to predict toxicities of drug-like compounds. In fact, the MIA-based approach has shown to be accurate to model bioactivities, boiling points, NMR chemical shifts and electrophoretic profiles, but it has been used to model cytotoxicities for the first time in this work, in order to contribute for studies to develop safer drugs. The QSAR modeling of bioactivities (pEC50) and cytotoxicities (CCIC50) of a series of HIV-1 protease inhibitors, some ritonavir derivatives, is reported in this work using the MIA-QSAR approach. The statistical quality of both models indicates that pEC50 and CCIC50 of ritonavir analogs can be reliably predicted using this method; therefore, improved drugs can be designed.