Salta al contenuto principale
Passa alla visualizzazione normale.

MARCO TUTONE

The Molecular dYnamics SHAred PharmacophorE (MYSHAPE) approacha new tool to arise docking and pharmacophore modeling performance: virtues and vices

Abstract

In a recent paper, we presented a new virtual screening workflow that addresses the arising issues of molecular docking and pharmacophore modeling when using a single set of coordinates and a single active ligand [1]. MD simulations were carried out and ligand-protein interactions were analyzed and collected together with their appearance frequency. A pharmacophore model was then created using only the common feature patterns that all the ligands exhibited during MD simulations. This ‘Molecular dYnamics SHAred PharmacophorE’ was then used for virtual screening on active and inactive molecules library. MYSHAPE was also used as constraints for the creation of the docking grid. The application of the MYSHAPE model showed an interesting increase of the screening capability both in terms of sensitivity of the model and specificity when compared to the PDB models. This work [1] was a first essay for a workflow that should be applied to different proteins. In the present study we tried to apply the MYSHAPE approach to other three different ligand-protein systems (ERα; RXRα, and MAPKp38) with the aim to optimize the method to each different biological target taking in consideration the early recognition. The obtained results for these new targets confirmed that it is mandatory, to optimize the virtual screening campaign, the selection of dynamic features and constraints for docking. In particular, the addition of the constraints derived from MD simulation leads to an improvement in the model selectivity for RXRα and ERα in standard precision docking mode. For MAPKp38, validation metrics such as ROC, BEDROC, and AUAC are higher in extra precision mode. For the pharmacophore modeling, the addition of the features derived from the common interactions in MD simulations guarantee an improvement in the AUC for RXRα (37%), and ERα (77%), but light improvement for MAPKp38. MD simulation derived common interactions revealed fundamental for docking selectivity, while they are applied to pharmacophore modeling only when the number of final features in the common and dynamic pharmacophore is higher than the starting static pharmacophore. The strength behind the protocol is the ease of use related to the improvement of results. It also could represent a valid alternative to use very time-consuming techniques such as XP docking with constraints. Reference: 1. Perricone, U., Wieder, M., Seidel, T., Langer, T., Padova, A., Almerico, A. M., & Tutone, M. (2017). A Molecular Dynamics–Shared Pharmacophore Approach to Boost Early-Enrichment Virtual Screening: A Case Study on Peroxisome Proliferator-Activated Receptor α. ChemMedChem, 2017. DOI: 10.1002/cmdc.201600526