Using AMBER18 for Relative Free Energy Calculations

Abstract

With renewed interest in free energy methods in contemporary structure-based drug design there is a pressing need to validate against multiple targets and force fields to assess the overall ability of these methods to accurately predict relative binding free energies. We computed relative binding free energies using GPU accelerated Thermodynamic Integration (GPU-TI) on a dataset originally assembled by Schrödinger, Inc.. Using their GPU free energy code (FEP+) and the OPLS2.1 force field combined with the REST2 enhanced sampling approach, these authors obtained an overall MUE of 0.9 kcal/mol and an overall RMSD of 1.14 kcal/mol. In our study using GPU-TI from AMBER with the AMBER14SB/GAFF1.8 force field but without enhanced sampling, we obtained an overall MUE of 1.17 kcal/mol and an overall RMSD of 1.50 kcal/mol for the 330 perturbations contained in this data set. A more detailed analyses of our results suggested that the observed differences between the two studies arise from differences in sampling protocols along with differences in the force fields employed. Future work should address the problem of establishing benchmark quality results with robust statistical error bars obtained through multiple independent runs and enhanced sampling, which is possible with the GPU-accelerated features in AMBER.

Towards chemical accuracy for alchemical free energy calculations with hybrid physics-based machine learning / molecular mechanics potentials

“…We applied this nonequilibrium switching free energy correction scheme to a benchmark set of a wellstudied congeneric series of inhibitors for non-receptor tyrosine-protein kinase (Tyk2) from the Schrödinger JACS benchmark set ( Figure 2) [45]. This benchmark set is challenging for both commercial force fields (OPLS2.1 achieves a ΔG RMSE of 0.93±0.12 kcal mol −1 [45]) and public force fields (GAFF 1.8 achieves a ΔG RMSE 1.13 kcal/mol −1 and ΔΔG RMSE of 1.27 kcal/mol [50]). We consider a purely MM binding free energy baseline by using the ANI-2x [49] ML model (parameterized from DFT B97X/6-31G*) with AMBER14SB [47], TIP3P [48], and the Open Force Field Initiative OpenFF 1.0.0 ("Parsley") small molecule force field [46].…”

“…We consider a purely MM binding free energy baseline by using the ANI-2x [49] ML model (parameterized from DFT B97X/6-31G*) with AMBER14SB [47], TIP3P [48], and the Open Force Field Initiative OpenFF 1.0.0 ("Parsley") small molecule force field [46]. The OpenFF 1.0.0 ("Parsley") MM free energy calculations, shown in Figure 6 (A) and (C), achieve an accuracy that is statistically indistinguishable from other public and commercial MM force field benchmarks in terms of both root-mean squared error (RMSE; OPLS2.1 [45] and GAFF 1.8 [50]) and mean unsigned error (MUE; OPLS3.1, GAFF 2.1, and CGenFF 4.1 [64]). When the MM free energy calculation is corrected to ML/MM level of theory, Figure 6 [72] to integrate estimates from redundant transformations in the relative alchemical transformation network.…”

Carvacrol: An In Silico Approach of a Candidate Drug on HER2, PI3Kα, mTOR, hER‐α, PR, and EGFR Receptors in the Breast Cancer

“…Atomic charges were obtained with the Merz–Kollman scheme [ 52 ] by fitting a restricted electrostatic potential (RESP) model [ 53 ] by the Gamess-US program, and the output file was used into the resp. subprogram of the AmberTools program package [ 54 ]. Assignment of GAFF force field parameters was carried out by the Antechamber program [ 55 ] and the required input files for molecular dynamics simulations were prepared using the ACPYPE python interface [ 56 ].…”

A Role of Newly Found Auxiliary Site in Phospholipase A1 from Thai Banded Tiger Wasp (Vespa affinis) in Its Enzymatic Enhancement: In Silico Homology Modeling and Molecular Dynamics Insights

“…The protein in a sodium chloride (NaCl) solution at 37 • C was later prepared so as to mimic the in vivo-like condition so that the molecular conformation could be reasonably illustrated. The constructed Ves a 1 structure was solvated in the transferable intermolecular potential with 3 points (TIP3P) water rectangular box with a distance of 12 Å from the protein surface, yielding 15,846 water molecules, using the Assisted Model Building with Energy Refinement (AMBER16) package [31]. The protein charge was then neutralized using 9 chloride ions (Cl − ), and 43 NaCl pairs were then added into the system to create the equivalent concentration of 0.15 mol/L NaCl solution.…”