Repurposing Therapeutics for COVID-19: Rapid Prediction of Commercially available drugs through Machine Learning and Docking

Sovesh Mahapatra; Prathul Nath; Manisha Chatterjee; Neeladrisingha Das; Deepjyoti Kalita; Partha Roy; Soumitra Satapathi

Here, we have at first taken the inhibitors of SARS-CoV-2, which doesn't allow them to replicate in the host. These are screened and collected in the bioassay AID 1706 which were used as the main component for the modeling of the training model using ML. The 914 attributes were taken under consideration for more than 200,000 compounds. We have not used unsupervised learning to filter out the dataset, because it would have made the dataset much weaker. As mentioned in the method section, we have used a classifying algorithm to train the model and the best among them was further used for the testing and predicting the drugs from the Drug Bank. The schematic of the process is shown in Figure 1 .