Recent advances in deep learning enabled approaches for identification of molecules of therapeutics relevance

Deep learning, the subset of artificial intelligence, has ushered the renaissance of efficient drug screening program comprehensively. It consists of algorithms designed to extract high-level abstractions from the input using multiple nonlinear transformations. Having accomplished substantial achievements in speech recognition, computer vision, and natural language processing, it has triggered its growth in drug discovery programs as well, in a much reliance and efficient way. The speed and time complexity to train the models integrated with billions of data has been drastically ameliorated owing to the engagement of multiple hidden layers trained on GPUs and clusters. Therefore, to showcase their great predictive capability in modern drug discovery programs, we introduce this chapter highlighting the recent trends of deep learning in drug screening from the exorbitant of chemical space. Inclusively, it will also briefly go through mainstream deep learning approaches that changed the way in which new drugs were identified.