Computer-aided prediction of inhibitors against STAT3 for managing COVID-19 associate cytokine storm
Abstract It has been shown in the past that levels of cytokines, including interleukin 6 (IL6), is highly correlated with the disease severity of COVID-19 patients. IL6 mediated activation of STAT3 is responsible to proliferate proinflammatory responses that leads to promotion of cytokine storm. Thus, STAT3 inhibitors may play a crucial role in managing pathogenesis of COVID-19. This paper describes a method developed for predicting inhibitors against the IL6-mediated STAT3 signaling pathway. The dataset used for training, testing, and evaluation of models contains small-molecule based 1564 STAT3 inhibitors and 1671 non-inhibitors. Analysis of data indicates that rings and aromatic groups are significantly abundant in STAT3 inhibitors compared to non-inhibitors. In order to build models, we generate a wide range of descriptors for each chemical compound. Firstly, we developed models using 2-D and 3-D descriptors and achieved maximum AUC 0.84 and 0.73, respectively. Secondly, fingerprints (FP) are used to build prediction models and achieved 0.86 AUC and accuracy of 78.70% on validation dataset. Finally, models were developed using hybrid features or descriptors, achieve a maximum of 0.87 AUC on the validation dataset. We used our best model to identify STAT3 inhibitors in FDA-approved drugs and found few drugs (e.g., Tamoxifen, and Perindopril) that can be used to manage COVID-19 associated cytokine storm. A webserver “STAT3In” (https://webs.iiitd.edu.in/raghava/stat3in/ ) has been developed to predict and design STAT3 inhibitors.