In this study, we aim to expand upon the current utilization of Deep Mutational Scanning. In nature, α-syn toxicity typically involves more than just a point mutation. To explore more "life-like" situations, multiple simultaneous mutations need to be considered. Fulfilling this via DMS is both laborious and significantly more challenging to select for all variants. Instead, we propose using computation techniques to bridge this gap.
Our goal is to achieve a highly accurate model that can not only predict the impact of simple point mutations but also generalize to more complex scenarios on multiple mutations. Upon successful training and validation, we will in silico build new sequences with multiple mutations. Using our trained model, we will predict the effects of these mutations, aiming to uncover potential sequences with enhanced or decreased fitness. This project will contribute to 1) insights into protein engineering, where we are expecting to find specific regions of the protein that are more prone to mutation and the chemical nature of those tolerated changes. 2) understanding how multiple mutations impact protein function.