Quantitative Prediction Model for Affinity of Drug-Target Interactions Based on Molecular Vibration and Overall System of Ligand-Receptor

Abstract

Background: the study of drug-target interactions (DTIs) affinity plays an important role in safety assessment and pharmacology. Currently, quantitative structure-activity relationship (QSAR) and molecular docking (MD) are most common methods in research of DTIs affinity. However, they often built for a specific target or several targets and most QSAR and MD were based either only on structure of drug molecules or on structure of targets with low accuracy and small scope of application. How to construct quantitative prediction models with high accuracy with wide applicability remains a challenge. To this end, this paper screened molecular descriptors based on molecular vibrations and took molecule-target as a whole system to construct prediction model with high accuracy-wide applicability based on Kd and EC50, and to provide reference for quantifying affinity of DTIs.Methods: Through parametric characterization based on molecular vibrations and protein sequences, taking molecule-target as whole system and feature selection of drug molecule-target, we constructed feature datasets of DTIs quantified by Kd and EC50, respectively. Then, prediction models were constructed using above datasets and SVM, RF and ANN. In addition, optimal models were selected for application evaluation and comprehensive comparison.Results: Under ten-fold cross-validation, evaluation parameters based on RF for EC50 dataset are as follows: R2 (RF) of training and test sets are 0.9611, 0.9641; MSE (RF) of training and test sets are 0.0891, 0.0817. Evaluation parameters based on RF for Kd dataset are as follows: R2 (RF) of training and test sets are 0.9425, 0.9485; MSE (RF) of training and test sets are 0.1208, 0.1191. After comprehensive comparison, the results showed that RF model in this paper is optimal model. In application evaluation of RF model, the errors of most prediction results were in range of 1.5-2.0.Conclusion: Through screening molecular descriptors based on molecular vibrations and taking molecule-target as whole system, we obtained optimal model based on RF with more accurate-widely applicable, which indicated that selection of molecular descriptors associated with molecular vibrations and the use of molecular-target as whole system are reliable methods for improving performance of model. It can provide reference for quantifying affinity of DTIs.