Prediction of Protein−compound Binding Energies from Known Activity Data: Docking‐score‐based Method and its Applications

Abstract

We used protein−compound docking simulations to develop a structure‐based quantitative structure−activity relationship (QSAR) model. The prediction model used docking scores as descriptors. The binding free energy was approximated by a weighted average of docking scores for multiple proteins. This approximation was based on a pharmacophore model of receptor pockets and compounds. The weights of the docking scores were restricted to small values to avoid unrealistic weights by a regularization term. Additional outlier elimination improved the results. We applied this method to two groups of targets. The first target was the kinase family. The cross‐validation results of 107 kinase proteins showed that the RMSE of predicted binding free energies was 1.1 kcal/mol. The second target was the matrix metalloproteinase (MMP) family, which has been difficult for docking programs. MMPs require metal‐binding groups in their inhibitor structures in many cases. A quantum effect contributes to the metal−ligand interaction. Despite this difficulty, the present method worked well for the MMPs. This method showed that the RMSE of predicted binding free energies was 1.1 kcal/mol. In comparison, with the original docking method the RMSE was 1.7 kcal/mol. The results suggest that the present QSAR model should be applied to general target proteins.

Analysis of non-structural proteins, NSPs of SARS-CoV-2 as targets for computational drug designing

“…and if there are no grid parameter, default grid parameters based on binding pockets were taken for docking by PyRx. Calculation of binding score or energy or affinity is completely based on the pharmacophores of ligand and receptors and mathematical expressions of the same can be found from the articles [ 47 , 48 ] even though explanation of mathematical derivation is beyond the scope of this paper. BaseByBase is a whole genome pairwise and multiple alignment editor.…”

Prediction of Passive Membrane Permeability by Semi‐Empirical Method Considering Viscous and Inertial Resistances and Different Rates of Conformational Change and Diffusion

“…[37,38] The regression model was the same as that used in our previous study. [37,38] The regression model was the same as that used in our previous study.…”

“…Our physical-property prediction method was a principal component regression (PCR) with an L2 regularization term based on the molecular descriptors. [37,38] The regression model was the same as that used in our previous study. Namely, the principal component (PC) analysis projected each compound into each point in a chemical space of the PC, and a multiple linear regression was applied to the molecular coordinates in the chemical space.…”

<p>Developing a Kinase-Specific Target Selection Method Using a Structure-Based Machine Learning Approach</p>

“…For such cases, Fukunishi et al developed a quantitative structureactivity relationship (QSAR) model based on docking scores to predict the binding energy in protein-ligand complexes, and tested it on kinase family proteins and matrix metalloproteinase (MMP) proteins. 18 Lately, multi-task machine learning approaches attracted the attention of many computational chemists. Recent studies demonstrated high accuracy of such methods in tasks related to activity prediction across multiple targets and pharmacokinetic properties.…”

“…For such cases, Fukunishi et al developed a quantitative structure-activity relationship (QSAR) model based on docking scores to predict the binding energy in protein-ligand complexes, and tested it on kinase family proteins and matrix metalloproteinase (MMP) proteins. 18 …”