MTiOpenScreen: a web server for structure-based virtual screening

Abstract

Open screening endeavors play and will play a key role to facilitate the identification of new bioactive compounds in order to foster innovation and to improve the effectiveness of chemical biology and drug discovery processes. In this line, we developed the new web server MTiOpenScreen dedicated to small molecule docking and virtual screening. It includes two services, MTiAutoDock and MTiOpenScreen, allowing performing docking into a user-defined binding site or blind docking using AutoDock 4.2 and automated virtual screening with AutoDock Vina. MTiOpenScreen provides valuable starting collections for screening, two in-house prepared drug-like chemical libraries containing 150 000 PubChem compounds: the Diverse-lib containing diverse molecules and the iPPI-lib enriched in molecules likely to inhibit protein-protein interactions. In addition, MTiOpenScreen offers users the possibility to screen up to 5000 small molecules selected outside our two libraries. The predicted binding poses and energies of up to 1000 top ranked ligands can be downloaded. In this way, MTiOpenScreen enables researchers to apply virtual screening using different chemical libraries on traditional or more challenging protein targets such as protein-protein interactions. The MTiOpenScreen web server is free and open to all users at http://bioserv.rpbs.univ-paris-diderot.fr/services/MTiOpenScreen/.

Figure 1.

Workflow of MTiAutoDock and MTiOpenScreen. The user input is shown in blue. Calculations and chemical libraries provided by MTiOpenScreen are shown in red.

Figure 2.

MTiOpenScreen interactive results pages. 3D protein structure of the VEGFR2 kinase domain (PDB ID: 4ag8) and the best docking pose are shown in 3D. The predicted binding energies of the top 100 ranked ligands and physico-chemical properties of ligands taken from Diverse-lib are also visualized. Ligand names in Diverse-lib and iPP-lib correspond to the PubChem SID. ‘Accepted’ annotation indicates that the compound does not contain toxic or PAINS groups. ‘Intermediate’ annotation indicates that the compound does not contain PAINS groups but contains toxicophores that belong to the low risk toxicity category (more information on low and high risk toxicity groups can be seen at http://fafdrugs2.mti.univ-paris-diderot.fr/groups.html).

Figure 3.

Enrichment curves of virtual screening of known actives and 1000 decoys performed with MTiOpenScreen on three protein targets.