iPPI-DB: A User-Friendly Web Application to Query a Database of Protein-Protein Interactions Inhibitors
Protein-protein interactions (PPI) are one of the next predominant classes of therapeutic targets. With about 130,000 binary PPI just in humans , the development of small molecule drugs targeting these systems represents a considerable challenge with unprecedented pharmaceutical and medical benefits. However, experimental screening techniques (e.g High throughput Screening – HTS) relying on conventional commercial chemical libraries, that usually allow chemists to identify innovative molecules, have clearly demonstrated their limits on PPI targets. Those difficulties are mainly due to the inadequacy of those chemical libraries that were historically designed for conventional targets such as G--Protein Coupled Receptors (GPCR) or enzymes but most importantly to our misunderstanding of the PPI chemical space. A paradigm shift is therefore essential in our way to design chemical libraries when aiming at PPI targets. Following this path, learning from the known successful examples of PPI modulation with small non--peptide inhibitors has been pinpoint as a critical step. This strategy should help to promote in a more systematic manner new chemical entities (NCE), and allow the scientific community to derive general trends such as sets of appropriate physicochemical properties and privileged chemotypes. To guide the chemists in addressing this issue, we have created iPPI--DB. In this database, we have so far collected the structures, the physicochemical characteristics, and the pharmacological data (biochemical and/or cellular binding data) of 1650 small non--peptide inhibitors across 13 families of PPI targets (v1.0 -- February 2013). Those data are extracted from the literature and manually curated by experts. As an online service to the PPI community, we wanted to propose a user--friendly web interface to access iPPI--DB . Thus, we have developed a web application that can be accessed by anyone from the website of our Inserm technological platform CDithem. The uniqueness of iPPI--DB resides in the combination of its manual curation and the nature of its querying and visualizing tools. The first way to query our database is by choosing pharmacological criteria such as the PPI target, the threshold for the activity of the compound and/or for some molecular descriptors’s thresholds (molecular weight, proportion of sp3 carbon, etc.). All compounds fulfilling the query criteria are displayed as a list with all annotated properties. Recently, we added the possibility for users to sketch their own molecule in an embedded Marvin Sketch applet and to submit this molecule as the query for iPPI--DB using a similarity search based on ECFP4 or FCFP4 fingerprints (powered by JChem v6.1). The results of such query are displayed similarly to those of pharmacological queries for the five most similar iPPI--DB compounds based on the type of fingerprints chosen by the user. These results are preceded by a reminder of the input molecule structure using the Marvin View applet, its physicochemical profile (radar chart showing nine molecular descriptors calculated with JChem), the molecule compliance with Lipinski’s, Veber’s and Pfizer’s 3/75 rules. The purpose of this development is to facilitate the access to the right information using appropriate criteria and to cross--reference the annotated data to visualize more efficiently the results. We think iPPI--DB and its web--application could assist chemists, biologists and clinicians to more rationally design the next generation of PPI modulators.