AlphaMut: A Deep\nReinforcement Learning Model to Suggest\nHelix-Disrupting Mutations

Helices are important secondary structural motifs within\nproteins\nand are pivotal in numerous physiological processes. While amino acids\n(AA) such as alanine and leucine are known to promote helix formation,\nproline and glycine disfavor it. Helical structure formation, however,\nalso depends on its environment, and hence, prior prediction of a\nmutational effect on a helical structure is difficult. Here, we employ\na reinforcement learning algorithm to develop a predictive model for\nhelix-disrupting mutations. We start with a model to disrupt helices\nindependent of their protein environment. Our results show that only\na few mutations lead to a drastic disruption of the target helix.\nWe further extend our approach to helices in proteins and validate\nthe results using rigorous free energy calculations. Our strategy\nidentifies amino acids crucial for maintaining structural integrity\nand predicts key mutations that could alter protein structure. Through\nour work, we present a new use case for reinforcement learning in\nprotein structure disruption.