Chemoinformatics and proteochemometrics

The chemoinformatics and proteochemometrics group is dedicated to the identification of pharmacological modulators of macromolecular interactions such as protein-protein interactions (PPI) by small molecules using in silico approaches. It combines chemoinformatics and structural bioinformatics techniques to characterize the structural properties of binding cavities present at the core of PPI interfaces and the most suitable physicochemical profile of the small molecules meant to modulate them. An important aspect of our work relies on the rationalization of the chemical space of protein-protein interactions’ inhibitors by analyzing the properties of successful examples of pharmacological PPI modulations. To this end, we are driving the iPPI-DB initiative project (http://ippidb.pasteur.fr/), a database of PPI modulators (only small molecules). This database is a great source of pharmacological data that help us to derive some trends about the PPI chemical space using machine learning techniques. Another aspect of our project is also to get some insight into the PPI interfaces’ properties themselves that obviously condition the binding of small compounds. To this end, we are mapping the whole PPI pocketome  in order to assess the ligandability of all binding pockets present within PPIs. By crossing both types of information, i.e by crossing the target and the chemical spaces of PPIs, we aim at determining which chemotypes should be associated with which types of PPI targets. The final goal is therefore to facilitate the identification of quality chemical probes on PPI targets and more generally on macromolecular interactions by the mean of complementing chemical biology approaches. This expertise is used to design chemical libraries using chemoinformatics approaches that are tuned to specific projects and that can be employed to do either virtual ligand screening (VLS) within the team, or high throughput screening (HTS) campaigns.