Seminar, 06. February 2020, Matthias Rarey

Thursday, 06. February 2020, 16.15 p.m. 

Ernst-Abbe-Platz 2, Seminarraum 3423

Algorithmic Challenges in Structure-Based Molecular Design and Cheminformatics

Prof. Dr. Matthias Rarey
(Center for Bioinformatics, University of Hamburg)

Drug discovery and design is heavily interlinked with modern computational approaches related to protein-structure-based modeling and cheminformatics. In this talk, I will survey our recent method developments and their implementations in tools and modern web services. Structure-based design is characterized by graphic-intense, interactive applications. Due to the increase in available structure data on the one hand and machine learning approaches on the other, methods replacing interactive modeling tasks by reliable automated procedures are required. The ultimate aim is a fully automated protein-ligand docking and screening process making use of all available data from the Protein Data Bank PDB and ChEMBL. There are lots of problems along this line and I will just survey a few like estimating electron density support in structure models or protein binding site ensemble generation.

In Cheminformatics, the rising amount of compounds readily available for synthesis or testing causes big challenges. Analyzing and prefiltering of compound collections are typically done with chemical patterns (SMARTS). To support chemists in developing appropriate filter sets, a series of tools automating pattern generation and comparison have been developed. Currently, our research focus is on the mastery of chemical spaces like EnamineREAL representing up to 11 billion purchasable compounds. Since these spaces cannot be screened one by one, completely new approaches for navigation making use of the combinatorial nature of the spaces are requested.

Web resources:
https://uhh.de/amd     https://uhh.de/naomi
https://proteins.plus     https://smarts.plus