Life’s sustainability crucially depends on the presence of enzymes, pivotal for accelerating chemical reactions within a biologically compatible temporal framework. Enzymes exhibit advantageous characteristics, including high specificity and selectivity, operating effectively under very mild biological conditions.
Despite the potential advantages, the general application of enzymes in industry is quite limited, as most industrial processes lack a natural enzyme able to perform the desired transformations. Current enzyme design efforts mostly rely on experimental trial-and-error attempts, as none of the available enzyme design strategies can rapidly design tailor-made enzymes exhibiting Nature-like catalytic activities.
The objective is to develop fast but accurate computational methodologies for designing industrially relevant enzymes with catalytic activities as those found in Nature. To that end, protocols integrating computational chemistry, deep learning such as the neural network AF2, graph theory, and computational geometry are used to properly describe the complex nature of enzyme catalysis
- These approaches can be used to study enzymatic systems,
- to design industrially relevant enzymes such as squalene hopene cyclases for the industrial production of the fragrance Ambroxide
- , and also to design non-enzymatic synthetically useful catalysts (4).
