Combinatorial library optimization process

Before the actual production of a new library for JECL can begin, a validation document is prepared by the one of the five companies responsible for the synthesis of the library, together with Lead Discovery Center GmbH (LDC). LDC’s task is to  maximize the diversity and the novelty of the new compounds and to ensure that the compounds are in the medicinal chemistry property space. Diversity is not only assessed purely from a structural point of view but more importantly also the spread of physicochemical properties.

The first step in the library optimization is to search for all chemical building blocks that are applicable to the synthesis routes. After an initial price filtering, the affordable building blocks are filtered based on their physicochemical properties such as molecular weight and calculated logP. The remaining building blocks are then passed through several sub-structural filters that include both proprietary filters provided by the seven EFPIA partners and publicly available ones for example to exclude Pan Assay Interference Compounds (PAINS). After these filtering steps, there is usually a manageable number of building blocks left so that the synthetic accessibility of the building blocks can be evaluated manually by a chemist responsible for synthesis. Based on his/her feedback, the building blocks for production are selected so that the diversity of the produced compounds is maximized.

Before the library design is finalized, there are several quality controls in place to ensure that both novelty and diversity aspects of the new library are taken into account. The potential new library is compared against JECL, previously enumerated European Lead Factory compounds and public databases such as eMolecules and ChEMBL. The library design is adjusted if overly similar compounds have been included in the initial enumeration. The success rate for the library synthesis is estimated to be 50-70%, so usually around 1000 molecules are designed to allow the production of the final library with 500 compounds.

For more detailed information, please see the poster presented at The 10th International Conference on Chemical Structures (ICCS) and the 10th German Conference on Chemoinformatics (GCC), Noordwijkerhout, The Netherlands.