No doubt, cheminformatics and structure-based molecular design come of age. Basically every company performing active research in pharmacy, crop science, or biotechnology in general substantially invest in a reasonable cheminformatics and modeling research infrastructure. The inclusion of computational techniques in molecular design even got further impetus by the recent successes in machine learning in various application areas. The key problems to solve are well-known: Ranging from topological and geometric (3D) searching in chemical databases via conformer generation to molecular docking, scoring and binding affinity estimation on the one hand and all kinds of machine-driven models to predict ADME and toxicological effects on the other.<br/><br/>Surprisingly, the underlying algorithms of most methods applied on a daily basis are at least 15 years old, many of them much older. Advancements are made by optimal hardware exploitation (like GPU computing) and better parametrization on larger data collections. So, are all problems solved or do we have found the optimal algorithms for these problems? By looking at the performance on real data, one gets the impression that there is certainly room for improvements. In this talk, a few examples for new algorithms addressing new and old problems are shown. Furthermore, a route to more innovation in cheminformatics and structure-based drug design is envisioned.

All problems solved? On the quest for algorithms in cheminformatics and structure-based design

255th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.