Impact of domain knowledge on blinded predictions of binding energies by alchemical free energy calculations

Abstract

Keywords D3R · computer-aided drug design · protein-ligand interactions · alchemical free energy calculations The drug design data resource (D3R) consortium organises blinded challenges to address the latest advances in computational methods for ligand pose prediction, affinity ranking, and free energy calculations. Within the context of the second D3R Grand Challenge several blinded binding free energies predictions were made for two congeneric series of FXR inhibitors with a semi-automated alchemical free energy calculations workflow featuring the FESetup and SOMD tools. Reasonable performance was observed in retrospective analyses of literature datasets. Nevertheless blinded predictions on the full D3R datasets were poor due to difficulties encountered with the ranking of compounds that vary in their net-charge. Performance increased for predictions that were restricted to subsets of compounds carrying the same net-charge. Disclosure of X-ray crystallography derived binding modes maintained or improved the correlation with experiment in a subsequent rounds of predictions. The best performing protocols on D3R set1 and set2 were comparable or superior to predictions made on the basis of analysis of literature

Effect of set up protocols on the accuracy of alchemical free energy calculation over a set of ACK1 inhibitors

“…For each l value MD trajectories were computed in the NPT ensemble with a pressure of 1 atm and temperature of 300 K using the software SOMD 2016.1.0 [20,38,39]. SOMD has been used in several recent studies to model the binding energetics of enzyme inhibitors [21], carbohydrate ligands [19], and hostguest systems [13]. Each l window was sampled for 4 ns.…”

“…Relative free energies for complete datasets were evaluated using version 0.3.5 of the freenrgworkflows python module [https://github.com/michellab/freenrgworkflows]. For more details, see Mey et al [46]. All analysis scripts are available online at https://github.com/michellab/ACK1_Data.…”

“…Some AFE protocols have even achieved predictions of binding energies with root mean square deviations (RMSD) under 1.5 kcal mol −1 , and Pearson Correlation coefficients (R) of around 0.7 or better [16][17][18]. Nevertheless, the performance varies significantly between different AFE protocols and targets [19][20][21] and it is important to explore further the robustness of these methodologies.…”

Automated Assessment of Binding Affinity via Alchemical Free Energy Calculations

“…12,32,33 The SOMD package available within the Sire framework has been successfully applied to relative and absolute binding free energy studies of a range of fragments, drug-like small molecules, carbohydrates, host-guest systems. [5][6][7]29,[34][35][36][37] However, a systematic comparison on standard datasets has not been published to date. Here we report a large-scale validation of Flare's FEP implementation on the Wang et al 1 dataset, as well as on a smaller dataset of more challenging scaffold hopping modifications.…”

Application of ESMACS Binding Free Energy Protocols to Diverse Datasets: Bromodomain-Containing Protein 4

“…The discovery and design of novel drugs is immensely expensive, with one study putting the cost of each new therapeutic molecule that reaches the clinic at US$1.8 billion 1 . A diversity of computational approaches, specifically binding free energy calculations which rely on physics-based molecular dynamics simulations (MD) have been developed 2 and blind tests show that many have considerable predictive potential 3,4 . In this context, recent developments in algorithms and hardware that have reduced the cost of these computational approaches have seen an increase in their appeal to the pharmaceutical industry [5][6][7][8][9] .…”