Comparing the Performances of Force Fields in Conformational Searching of Hydrogen-Bond-Donating Catalysts

Abstract

Here, we compare the relative performances of different force fields for conformational searching of hydrogen-bond-donating catalyst-like molecules. We assess the force fields by their predictions of conformer energies, geometries, low-energy, nonredundant conformers, and the maximum numbers of possible conformers. Overall, MM3, MMFFs, and OPLS3e had consistently strong performances and are recommended for conformationally searching molecules structurally similar to those in this study.

New star-shaped ligands generated by evolutionary fitting the Omicron spike inner-cavity

“…The AutoDockVina (ADV) program written in Python vs3.8 and included in the PyRx-098/PyRx-1.0 package was used as described before https://pyrx.sourceforge.io/). Briefly, ADV docking was performed after *.pdbqt file conversion of the trimer protein and ligands 68 using the mmff94s (Merck) forcefield for energy minimization 67,69,70 . Comparison of ligand InChiKeys before and after docking monitored for possible alterations of molecular geometries 67,[69][70][71][72] .…”

“…Briefly, ADV docking was performed after *.pdbqt file conversion of the trimer protein and ligands 68 using the mmff94s (Merck) forcefield for energy minimization 67,69,70 . Comparison of ligand InChiKeys before and after docking monitored for possible alterations of molecular geometries 67,[69][70][71][72] . ADV generates many possible conformers using the rotatable bonds of the input ligands and selects those with the lower docking-scores for output 73 ; 74 .…”

Evolutionary-docking targeting bacterial FtsZ

“…Additionally, further in silico predictions of the viability of chemical synthesis of each out-ligand chemotype may help to filter those candidates before chemical synthesis for experimental assessment. The results remain to be confirmed by new docking algorithms (additional consensus docking) with improved force-fields [53][54][55][56] , and in vitro/in vivo binding to FtsZ before physiological and pharmacological tests. The parent molecules were selected from fragments of PC190723 from Figure 3.…”

“…However, to predict experimental binding is challenging due to the limitations of present in silico docking algorithms, such as DW and ADV. Known limitations include fixing of the docking-cavity amino acid side-chains, eliminating water molecules, unreliable calculation of docking-scores, molecular geometry changes by force-field energy minimizations [53][54][55][56] , and unreachable chemotype/chemical spaces 38,39 . Specially the force-field failures to correctly recognize different atom types, would greatly affect the conservation of molecular geometries after minimization [47][48][49][50] .…”

“…Since at present no scoring algorithm has proven yet to be reliable for every chemotype, a consensus between at least two different algorithms seem to be necessary to select the many leads generated by explorations of the vast chemotype space 66, 67 . In this work, a minimal consecutive consensus docking by ranks has been introduced using DW and ADV to theoretically increase prediction accuracies, as suggested in other systems [53][54][55][56] . Therefore, it may be expected that combining low docking-scores, high ligand efficiencies and consensus docking by ranks, helped to improve binding predictions.…”