Computational methods to model biomolecular systems have adopted a dominant position in tackling problems that would otherwise be impractical or impossible to solve via experimental techniques. In this thesis, three of such methods are used to model proteins: molecular dynamics, molecular docking, and deep learning. The first two are structure-based methods, where atoms are considered explicitly, while the latter is a data-driven method. Molecular modeling has yielded remarkable achievements in the last few decades, driven mainly by the increase in computer power and better algorithms. Deep learning is a newcomer to the study of biomolecular systems. Yet, it has the potential to spark a revolution in molecular simulations. Along the thesis, the three numerical methods are mainly used to build pipelines that can predict enzymatic activity. Accurately predicting the enzymatic activity can be a powerful tool in guiding efforts to tailor of the enzymatic activity toward a desired compound.
|Qualification||Doctor of Philosophy|
|Place of Publication||[Groningen]|
|Publication status||Published - 2022|