A Deep Learning-Based Quantitative Structure–Activity Relationship System Construct Prediction Model of Agonist and Antagonist with High Performance

Abstract

Molecular design and evaluation for drug development and chemical safety assessment have been advanced by quantitative structure–activity relationship (QSAR) using artificial intelligence techniques, such as deep learning (DL). Previously, we have reported the high performance of prediction models molecular initiation events (MIEs) on the adverse toxicological outcome using a DL-based QSAR method, called DeepSnap-DL. This method can extract feature values from images generated on a three-dimensional (3D)-chemical structure as a novel QSAR analytical system. However, there is room for improvement of this system’s time-consumption. Therefore, in this study, we constructed an improved DeepSnap-DL system by combining the processes of generating an image from a 3D-chemical structure, DL using the image as input data, and statistical calculation of prediction-performance. Consequently, we obtained that the three prediction models of agonists or antagonists of MIEs achieved high prediction-performance by optimizing the parameters of DeepSnap, such as the angle used in the depiction of the image of a 3D-chemical structure, data-split, and hyperparameters in DL. The improved DeepSnap-DL system will be a powerful tool for computer-aided molecular design as a novel QSAR system.

Ensemble Learning, Deep Learning-Based and Molecular Descriptor-Based Quantitative Structure–Activity Relationships

“…Using datasets of approximately 9000 chemical structures in the simplified molecular input line entry system (SMILES) format and the corresponding activity scores, which represent the agonist or antagonist levels of nuclear receptors and stress response proteins, from a database composed of high-throughput quantitative screening results, two datasets were prepared by defining "active" or "inactive" agonist or antagonist activities. The aforementioned database was derived from the Toxicology in the 21st Century (Tox21) 10k library composed of chemicals from commercial sources, such as pesticides, industrial chemicals, food additives, and drugs [42][43][44][45][46][47][48][49]. Then, the SMILES format was applied to a 3D conformational import to generate the SDF files of the chemical database.…”

Computational Models That Use a Quantitative Structure–Activity Relationship Approach Based on Deep Learning

“…Numerous parameters must be adjusted to use the Deep Snap method. Previous investigations have studied the effects of the following parameters on the performance of the prediction models: (1) the number of molecules per SDF, (2) the zoom factor percentage, (3) the atom size for the van der Waals percentage, (4) the bond radius, (5) the minimum bond length, and ( 6) the bond tolerance [57,[104][105][106]. The results suggest that optimal thresholds exist for attaining the best performance with these prediction models.…”

QSAR, ADMET In Silico Pharmacokinetics, Molecular Docking and Molecular Dynamics Studies of Novel Bicyclo (Aryl Methyl) Benzamides as Potent GlyT1 Inhibitors for the Treatment of Schizophrenia

“…The Quantitative structure-activity relationships (QSAR) have the potential to reduce the time and effort of molecular screening using mathematical predictive models [22]. One of these models is obtained by the multiple linear regression (MLR) technique, as a statistical tool for estimating the linear relationship between more than two variables which have cause-effect relations [23].…”