ASAP: <em>A</em>ccelerating antimicrobial peptoid <em>S</em>equence discovery through <em>A</em>I:<em>P</em>olicy based reinforcement learning approach
발표자
()
초록
내용
De novo drug discovery is vital in the pharmaceutical industry, with AI playing a key role in identifying species with potent antimicrobial properties. Researchers can expedite the discovery process and focus their efforts on synthesizing promising candidates. In recent days, peptoids have emerged as a compelling alternative to peptides for antimicrobial activities. However, applying conventional machine learning face a limitation as it requires vast amounts of sequence data. To alleviate the issue, we employ policy-based Reinforcement Learning for antimicrobial peptoid generation. Our approach involves training an agent to explore the embedding space of peptoid sequences, providing it with high rewards when generating antimicrobial candidates. This study represents the first AI application to discover antimicrobial peptoids, potentially revolutionizing the discovery process. In future research, we will validate our approach by synthesizing the candidates in a wet laboratory setting.