AI and machine learning can deliver strong results in drug discovery, but a fundamental limitation appears when you push beyond known chemistry: the generation of truly novel, viable scaffolds. Most workflows learn from existing data and can reliably propose candidates that improve properties within familiar chemical space. But when you extrapolate into new regions, uncertainty rises quickly – predictions become less dependable and error rates rise. If the goal is genuinely new, patentable structures beyond known actives, these approaches typically need to be complemented with additional methods and concepts to reach the required confidence threshold.
In this webinar, we explore approaches for unlocking new chemical modalities and take a look at the pitfalls and myths surrounding de novo compound ideation.
