A Novel One‐Click Automated Drug Discovery tool: A New Approach Based on Binding Efficiency Indices

Abstract

IntroductionModern drug discovery involves the identification of screening hits, medicinal chemistry and optimization of these hits to increase the affinity, selectivity, efficacy/potency, metabolic stability, and oral bioavailability. Once a compound that fulfills all of these requirements has been identified, it will start the process of drug development prior to clinical trials. The objective of this study is to design a one‐click drug discovery tool that is able to do the whole procedure of drug discovery for a large number of ligand that can be understandable and usable by pharmaceutical industries.MethodsTo run this algorithm, only a bash command under Linux platform is needed and depends on the number of compounds of interest, the running time would be different. While finishing, a CSV‐formatted Microsoft Excel file containing a sorted list of hundreds or thousands of compounds is saved. As a prerequisite, researchers provided a reference folder containing the computer‐generated files of chemical structures or html files that are generated by the PubChem database and the pdb‐formatted files of the target. The reference folder should also contain a grid parameter and docking parameter files. For the current project, four crystals of Smoothened receptors have been selected and the number of run was set to 30 (N=30). As a result, the algorithm starts to run 120 of all 485 ligands and takes averages and other attributes in the results file. The algorithm uses the AutoDock modeling tool, which automatically downloads and installs the MGL tool if it is not already installed..ResultsThis algorithm automatically processes four steps as follows; (1) At first step, downloading of structure files of each compound based on the PubChem CIDs provided in the reference folder will be started and then each sdf‐formatted file will automatically be converted to pdb‐formatted file and will receive a number and saved in a different folder. (2) At second step, AutoGrid executive file will be run. (3) Then, AutoDock executive file will be run. The time of this step depends on the number of run and the speed of the CPU and differs from several hours to a few days. (4) Having finished the dockings for all input ligands and crystals, the algorithm will automatically start computations of parameters demonstrated in Figure 1. The druglikeness properties such as topological polar area surface (TPSA), molecular weight (MW), LogP and so on will be predicted based on the online database, and then the binding efficiency indices nBEI, NSEI, and LLE will be calculated. These formula previously described by Abad‐Zapatero have been used for these calculations. After receiving the final result file, the results are depicted as a Cartesian‐plot to identify the best ligands located at the top of the plot (Figure 2a). The Figure 2b demonstrates pKi and binding energies.ConclusionAs the traditional drug discovery procedures do not consider the ADME/tox properties of each ligand, this algorithm finalizes the results based on the binding efficiency indices which are calculated based on the both thermodynamics and physico‐chemical properties of each ligand. Thus, while considering these properties, the drug discoverer will likely have more reliable results than the traditional and this algorithm not only increases the number of ligands and crystals in a single click process, but it also completes the results so that it can be easily depicted for further discussion. The selected ligands may be considered for further in vitro evaluation.