Future of Drug Design and Reinforcement Learning
Ekta Narkhede, Bharti Ingle, Purva Pandey, Monali Gulhane, N. Mohankumar, Nitin Rakesh, Mandeep Kaur · 2024
This paper focuses on how reinforcement learning (RL) is applied to revolutionize the processes of drug discovery and development. Borrowing from behavioral psychology, RL enables the robotic language model to identify optimal solution to all kinds of tasks, which includes novo drugs design. Using the Myers-Briggs type indicator dataset, the research focuses on identifying personality traits which will, in turn ensure effectiveness in developing a new drug. Though traditional drug discovery methods are slow and expensive, developments improve a very promising avenue for the acceleration of the process. Using RL algorithms, we aim to improve the sample efficiency and policy optimization for better and efficient novo development of drug candidate. We design the model such that it not only makes the predictions for the compound of the high activity but along with it a continuous and diverge range of structures that will serve as base for the future exploration open to all. A detailed explanation of drug-design methodologies, demonstrating RL's efficiency in simultaneously optimizing molecule for multiple goals through Structure Based Drug Design (SBDD) and High-Throughput Screening (HTS) is very amicable, thus it is a stern solution to the challenges held in the whole process of drug discovery in the creation of the new therapeutic molecule, which at the end of the day bring better outcomes.