Project

acetyl choline receptors; cheminformatics; computer-aided drug design; Virtual screening; databases; Chemical space

Lead
Développment de nouveaux médicaments par l'utilisation de GDB, une base de données de tout l'espace chimique.
Lay summary
la découverte de nouvelles molécules thérapeutiques est essentielle pour améliorer les traitements existants ainsi que pour combattre les maladies émergentes. L'un des aspects fondamentaux des efforts de découverte est l'exploration de l'espace chimique, c'est-à-dire l'ensemble de toutes les molécules possibles. Mon groupe de recherche a contribué à la compréhension de cet espace chimique en produisant GDB-17, qui est une base de données contenant 166 milliards de molécules représentant l'ensemble de toutes le molécules possibles jusqu'à 17 atomes. La taille de cette banque de donnée, qui est de plusieurs ordre de grandeur supérieure à toute autre base de donnée de molécules organiques déjà existante, démontre que la synthèse chimique n'a jusqu'à maintenant presque rien exploré. Notre projet continue avec l'amélioration et l'expansion de cette collection, ainsi que son application au développment de nouveaux médicaments et à la synthèse de nouvelles molécules.

Direct link to Lay Summary

Last update: 19.04.2016

Self-organised

Communication	Title	Media	Place	Year

Number	Title	Start	Funding scheme

The continuous discovery of new small molecule drugs is essential to the success of modern medicine, in particular to address unmet medical needs, to combat emerging infectious diseases, and to improve existing therapies. Over the last few years the use of big data has gradually established itself as a strategic resource for drug discovery. One aspect concerns the use of very large compound databases to open new opportunities for discovery. The unique contribution of my research group has been the production of the Chemical Universe Databases GDB-11, GDB-13 and GDB-17, which enumerate all possible organic molecules up to 11, 13 and 17 atoms following simple rules for synthetic feasibility and chemical stability, but taking the complete range of possible scaffold topologies into account. With 166.4 billion structures, GDB-17 is the largest small molecule database available today. Due to its size essentially all of its content represents new molecules. This proposal aims at further developing methods to exploit this unique resource for drug discovery, and at extending the database to larger molecules. Project A: The fragment database GDB-17F. A fragment subset of GDB-17, named GDB-17F and containing 4.6 billion structures, has been produced by selecting molecules with reduced complexity and complying with the rules of fragments by imposing constraints on the number of asymmetric and quaternary centers, molecule polarity and flexibility. An ongoing project to discover biologically active fragments by coupling virtual screening with organic synthesis and assays will be pursued. Project B: Adrenergic ligands from GDB-17. A virtual screening workflow has been established to identify analogs of any query molecule by a sequence of fingerprint similarity searches in GDB-17. In collaboration with Prof. P. Kolb in Marburg, we have performed high-throughput docking on a limited set of 360,000 structures identified by a search of the entire 166.4 billion molecules in GDB-17 for new analogs of ß-adrenergic receptor ligands. The workflow is entirely innovative and comprises visual selection from interactive color-coded maps of the database in pharmacophore fingerprint space. A series of new ligands will be synthesized and tested to identify new modulators of the ß-adrenergic receptor, providing an important proof-of-concept for the exploitation of GDB-17 in drug discovery. Project C: Expanding the GDB. In this project we will continue our effort to expand GDB to larger molecules by completing the assembly of GDB-30, which is an interactive resource to explore organic molecules up to 30 atoms based on a graph classification. The underlying data is a database of 55 billion graphs which can be searched by topological shape similarity. A molecule enumerator is connected to this database such that shape/pharmacophore analogs of any query molecule are generated covering a broad range of scaffold topologies. Statistical sampling will be carried out to obtain a representative overview of the full database content, which is estimated at 10E33 molecules. Project A: The fragment database GDB-17F. A low complexity fragment subset of GDB-17, named GDB-17F and containing 4.6 billion structures, has been produced by selecting molecules with reduced complexity and complying with the rules of fragments by imposing constraints on the number of asymmetric and quaternary centers, molecule polarity and flexibility. An ongoing project to discover biologically active fragments by coupling virtual screening with organic synthesis and assays will be pursued. Project B: Adrenergic ligands from GDB-17. A virtual screening workflow has been established to identify analogs of any query molecule by a sequence of fingerprint similarity searches. In collaboration with Prof. P. Kolb in Marburg, we have performed high-throughput docking on a limited set of 360,000 structures identified by a search of the entire 166.4 billion molecules in GDB-17 for new analogs of the ß-adrenergic receptor. The workflow is entirely innovative and comprises visual selection from interactive visualization maps of the database in pharmacophore fingerprint space. A series of new ligands will be synthesized and tested to identify new modulators of the ß-adrenergic receptor, providing an important proof-of-concept for the exploitation of GDB-17 in drug discovery. Project C: Expanding the GDB. In this project we will continue or effort to assemble GDB-30, which will take the form of an interactive resource to explore organic molecules up to 30 atoms based on a graph classification. The underlying data is a database of 55 billion graphs which can be searched by topological shape similarity. A molecule enumerator will be connected this database such that shape/pharmacophore analogs of any query molecule can be generated covering an extremely broad range of scaffold topologies. Statistical sampling will be carried out to obtain a representative overview of the full database content, which is estimated at 10E33 molecules.