Revitalizing old drugs: uncovering potential dengue antivirals through in silico target prediction and in vitro validation / Zafirah Liyana Abdullah

Dengue virus (DENV) infection is a rising health concern worldwide. Despite the alarming situation, there is no effective antiviral for DENV infections. With the increasing number of NS3 viral inhibitors developed for other diseases, the development of drugs targeting dengue NS3 protein is an interesting venture. Thus, this study is set forth to identify potential DENV inhibitors by building a prediction model of NS3 dengue antiviral. Initially, the models were built using bioactivity data of 62,354 compounds using ligand-based (L-B), and proteochemometric (PCM) modelling approaches. For the L-B approach, a Random Forest (RF) one-vs-one classification model was utilized while the PCM model employed the Parzen-Rosenblatt Windows (PRW) algorithm. Subsequently, the validated predictive models were used to screen marketed drugs, and in vitro assays were conducted to validate the drug’s viral inhibitory potential. Finally, the interactions that are responsible for the observed in vitro results were validated using molecular docking. The in silico studies revealed that both L-B and PCM models performed well in the internal and external validations. However, the L-B model showed better accuracy in the external validation, in terms of its sensitivity (0.671).