Design and development of a versatile graphical user interface for assisting consensus docking analyses on therapeutic targets for cancer treatment.

Molecular Docking, and its application in high-throughput virtual screening (HTVS), has arisen over the years as one of the most used tools in Drug Discovery, which is important to eliminate much of the effort and to decrease costs of experimental assays. A comprehensive understanding of the system that is subject to docking and of its underlying structural principles, may need a parallel usage of Molecular Graphics Visualization Software to visualize, at first bio-molecules, but also, interactions and search spaces. The steps that must be carried on to complete a docking process, from the preparation of input files to the analyses of the results, usually take advantage of more than one platform. Furthermore, although docking algorithms are usually implemented for a command-line usage only, just for some of them is currently available a graphical user interface (GUI) that may facilitate their usage. Nevertheless, the already developed GUIs are limited to the implementation of a single docking algorithm but, it is a general opinion, that in some cases it is not enough to rely on a single one to reach accurate results, making it necessary the usage of more docking algorithms in parallel. To overcome these drawbacks, I’ve developed a versatile, user-friendly and cross-platform GUI for assisting the process in all its steps, which works as a plugin of the molecular visualization system PyMOL. The plugin’s peculiar feature, which is not found in any other GUI currently developed, is to be suited for the implementation and usage of more docking algorithms in parallel. Therefore, in order to truly reflect its purposes, we have both tested and implemented a new scoring scheme, which is based own on this feature and which is termed as “Consensus Scoring”. This approach re-scores the results combining the output of different algorithms and it is primary useful to have a much more accurate ranking of the real bio-active molecules. Indeed it becomes necessary when dealing with Virtual Screening that, too often, lead to False Positives (FP) results. The biological interest was focused on Aurora-A protein, a Serine-Threonine Kinase involved in different Cell-Cycle related (CCR) processes, for which have been reported several mis-regulations that lead to the appearance of a neoplastic phenotype. By now, diverse types of compounds able to bind to the active site, impeding its catalytic activity, have been developed but, when talking about kinases, off-target events and cell toxicity may arise; meaning that it is very challenging the identification of selective and potent inhibitors. Therefore, it is needed the development of new methods to reach this aim. On these bases I’ve decided to evaluate the accuracy of “Consensus Scoring” method on Aurora-A, exploiting the information coming from crystallographic experiments about the real bio-active conformation adopted by different types of ligands in the interaction with this kinase. When applied to Aurora-A, this method revealed both an improvement of accuracy and some scoring issues in the docking algorithms used. Thus, concluding, we have identified the combination of algorithms and protocols to be used to reach considerable accurate predictions on Aurora-A system.