Submission Details

Design Rationale:

The main goal of our project was the repositioning of accepted drugs as M-pro inhibitors. With this aim, we took from e-Drug3D (https://chemoinfo.ipmc.cnrs.fr/MOLDB/index.php) all the 1930 molecular structures approved between 1939 and 2019 by the FDA with a molecular weight ≤ 2000 Da and docked them at the binding site of 6LU7 structure by using three different docking programs (i.e., GlideSP, FRED and VINA). For each ligand and docking program, we have kept the top 10 docked poses (i.e., 10 for GlideSP, 10 for FRED and 10 for VINA). When considering 10 poses per ligand/program we assume that one of them will be the bioactive one (although the score function not always have ranked it as the first one). Finally the bioactive pose is assumed to be identified if it has been found simultaneously by the three programs (this means that the rmsd for each pairwise pose comparison is below 1.5 angstroms). Our idea here is that, in general, protein-ligand docking programs are really good when trying to look for the possible poses of a ligand in a binding site but usually fail to rank correctly such poses. Then, if three different protein-ligand programs find the same pose, this have a high probability to be the bioactive one. Then, we applied a second filter in order to keep only those poses that are not only identified by the three programs but also show high affinity for M-pro. In order to find which threshold can be used to discriminate high from low affinity poses, we docked the 1577 compounds from the OTAVA Machine Learning SARS Targeted Library (https://otavachemicals.com/targets/sars-cov-2-targeted-libraries?utm_medium=email&utm_source=UniSender&utm_campaign=229613803) into the binding site of 6LU7 with GlideSP, FRED and VINA (with the identical set up conditions than were previously used for the 1930 FDA approved compounds). For this docking, we kept only the top-ranked pose obtained for each ligand by each programs. Then, we make three different histograms for the docking results of the OTAVA library (one per program) and arbitrarily selected as a threshold for distinguish from low to high M-pro affinity the lowest score value of the interval that contains the top 30% poses with the highest affinity. The resulting threshold was -6.3 for GlideSP, -7.0 for FRED and -7.5 for VINA. Then, all the poses that had succeed the first filter but had scores more positive than these threshold values were removed. Then, only 5 FDA approved compounds simultaneously accomplished that: (1) the same pose is found independently by GlideSP, FRED and VINA; and (2) this pose has score values equal or more negative than the three mentioned thresholds. This submission corresponds to one of these five compounds: CELECOXIB

Other Notes:

This submission corresponds to the commercial drug Celecoxib (https://en.wikipedia.org/wiki/Celecoxib). Although this molecule has not been built from your fragments, I have had to select one of the (i.e. x0072) because filling the "Fragment ids" field is mandatory for submission.

Inspired By:

Download PDB File