Explainable Artificial Intelligence in Structure Based Drug-Design
Recently Artificial Intelligence (AI) has lead to great progress in a variety of fields, including Drug Design. Protein Structures represent some of the most valuable information available in Drug Design, thus their incorporation into AI methods can significantly improve these models. A major challenge in modern AI methods is their so called Black-Box character. The lack of understanding why a model made a certain prediction means that no mechanistic insights can be gained, thus severely limiting the usefulness in a rational drug design process. Thus the development of methods for incorporating structural data into AI methods, as well as for explaining models, will bring significant improvements to the capabilities of AI in Drug Design.