The farnesoid X receptor (FXR) emerges as a promising drug target involved in regulating various metabolic pathways, yet some xenobiotic compounds binding to FXR would be an important determinant to induce the receptor dysfunctions that lead to undesirable side effects. Thus, it is critical to identify potential xenobiotics that disrupt normal FXR functions. In this work, five machine learning methods coupled with eight molecular fingerprints and 20 molecular descriptors were used to develop classification models for prediction of FXR binders. The built models were evaluated using the test set and two external validation sets. The best model was obtained using a combination of molecular descriptors and fingerprints, which exhibited the AUC values of 0.83 and 0.92 for the test set and the first external validation set, respectively. The overall prediction accuracy for the second external validation set with the best model was over 85%. Furthermore, several representative privileged substructures that are essential for FXR binders, such as benzimidazole, indole, and stilbene moiety, were detected using information gain and substructure frequency analysis. The applicability domain analysis via the Euclidean distance-based approach demonstrated a marked impact on the improvement of prediction accuracy. Overall, our built models could be helpful to rapidly identify potential chemicals binding to FXR.

Prediction of Farnesoid X Receptor Disruptors with Machine Learning Methods.

Pin1 (peptidyl-prolyl cis-trans isomerase NIMA-interacting 1) is directly involved in cancer cell-cycle regulation because it catalyses the cis-trans isomerization of prolyl amide bonds in proteins. In this sense, a modeling evaluation of the inhibition of Pin1 using quinazoline, benzophenone, and pyrimidine derivatives was performed by using multilinear, random forest, SMOreg, and IBK regression algorithms on a dataset of 51 molecules, which was divided randomly in 78% for the training and 22% for the test set. Topological descriptors were used as independent variables and the biological activity (pIC50) as a dependent variable. The most robust individual model contained 9 features, and its predictive capability was statistically validated by the correlation coefficient for adjusting, 10-fold cross validation, test set, and bootstrapping with values of 0.910, 0.819, 0.841, and 0.803, respectively. In order to improve the prediction of the pIC50 values, the aggregation of the individual models was performed through the construction of an ensemble, and the most robust one was constructed by two individual models (LR3 and RF1) by applying the IBK algorithm, and a substantial improvement in predictive performance is reflected in the values of R2ADJ = 0.982, Q2CV = 0.962, and Q2EXT = 0.918. Mean square errors <0.165 and good fitting between calculated and experimental pIC50 values suggest a robustness on the prediction of pIC50. Regarding the docking simulation, a binding affinity between the molecules and the active site for the Pin1 inhibition into the protein (3jyj) was estimated through the calculation of the binding free energy (BE), with values in the range of −5.55 to −8.00 kcal/mol, implying a stabilizing interaction molecule receptor. The ligand interaction diagrams between the drugs and amino acid in the binding site for the three most active compounds denoted a good wrapper of these organic compounds into the protein mainly by polar amino acids.

/pdf/computational-molecular-modeling-of-pin1-inhibition-activity-56r1cv6okp.pdf

Computational Molecular Modeling of Pin1 Inhibition Activity of Quinazoline, Benzophenone, and Pyrimidine Derivatives

The virtual screening (VS) is an important tool used in the modern drug discovery process to identify new leads and drug-like molecules for therapeutic interventions. With the rapid advancements in the computational hardware supported by advanced algorithmic progress and comprehensive data, the VS has gained fast momentum in drug discovery paradigm. The VS protocol is performed predominantly by using the ligand-based (LB) and structure-based (SB) VS methods with each one having its own merits and demerits. The LBVS method works on the similarity approach based on the physicochemical parameters, chemical functionality, and shape similarity of the ligands. The SBVS method works on the complementarity of the ligand with the target binding site and is more preferred than LBVS due to the consideration of both ligand and target information. Further in the SBVS, the available conformational information of the active chemical space helps in making reasonable decisions for the ligand selection. However, the limitations in scoring functions and incorrect pose prediction may result in the inaccurate SB models of limited performance in VS experiments. Various factors interplay in the development of successful VS models and the fine tuning of these factors leads to the efficient VS models with high sensitivity. In this chapter, some of the recent ligand- and structure-based approaches for virtual screening in drug discovery are described.

Ligand- and Structure-Based Virtual Screening in Drug Discovery

/pdf/artificial-intelligence-based-drug-design-and-discovery-3ueezjbhqp.pdf

Artificial Intelligence-Based Drug Design and Discovery

Cancer is one of the most serious issues in human life. Blocking programmed cell death protein 1 and programmed death ligand-1 (PD-L1) pathway is one of the great innovations in the last few years, a few numbers of inhibitors can be able to block it. (2-Methyl-3-biphenylyl) methanol derivative is one of them. Here, the quantitative structure-activity relationship (QSAR) established twenty (2-methyl-3-biphenylyl) methanol derivatives as the programmed death ligand-1 inhibitors. Density functional theory at the B3LPY/6-31+G(d, p) level was employed to study the chemical structure and properties of the chosen compounds. Highest occupied molecular orbital energy \begin{document}$E_{\rm{HOMO}}$\end{document} , lowest unoccupied molecular orbital energy \begin{document}$E_{\rm{LUMO}}$\end{document} , total energy \begin{document}$E_{\rm{T}}$\end{document} , dipole moment DM, absolute hardness \begin{document}$\eta$\end{document} , absolute electronegativity \begin{document}$\chi$\end{document} , softness \begin{document}$S$\end{document} , electrophilicity \begin{document}$\omega$\end{document} , energy gap \begin{document}$\Delta E$\end{document} , etc., were observed and determined. Principal component analysis (PCA), multiple linear regression (MLR) and multiple non-linear regression (MNLR) analysis were carried out to establish the QSAR. The proposed quantitative models and interpreted outcomes of the compounds were based on statistical analysis. Statistical results of MLR and MNLR exhibited the coefficient \begin{document}$R^2$\end{document} was 0.661 and 0.758, respectively. Leave-one-out cross-validation, \begin{document}$r^2_{\rm{m}}$\end{document} metric, \begin{document}$r^2_{\rm{m}}$\end{document} test, and "Golbraikh & Tropsha's criteria" analyses were applied for the validation of MLR and MNLR, which indicate two models are statistically significant and well stable with data variation in the external validation towards PD-L1. The obtained results showed that the MNLR model predicts the bioactivity more accurately than MLR, and it may be helpful and supporting for evaluation of the biological activity of PD-L1 inhibitors.

Theoretical investigation on QSAR of (2-Methyl-3-biphenylyl) methanol analogs as PD-L1 inhibitor

Quantitative Structure Activity Relationship (QSAR) is one of the widely used ligand based drug design strategies. Although a number of QSAR studies have been reported, debates over the limitations and accuracy of QSAR models are at large. In this review the applicability of various classes of molecular descriptors in QSAR has been explained. Protocol for QSAR model development and validation is presented. Here we discuss a case study on 7-Phenyl-imidazoquinolin-4(5H)-one derivatives as potent mPGES-1 inhibitors to identify crucial physicochemical properties responsible for mPGES-1 inhibition. The case study explains the methodology for QSAR analysis, validation of the developed models and role of diverse classes of molecular descriptors in defining the inhibitory activity of considered inhibitors. Various molecular descriptors derived from 2D/3D structure and quantum mechanics were considered in the study. Initially, QSAR models for the training set compounds were developed individually for each class of molecular descriptors. Further, a combined QSAR model was developed using the best descriptor from all the classes. The models obtained were further validated using an external test set. Combined QSAR model exhibited the best correlation (r = 0.80) between the predicted and experimental biological activities of test set compounds. The results of the QSAR analysis were further backed by docking studies. From the results of the case study it is evident that rather than a single class of molecular descriptors, a combination of molecular descriptors belonging to different classes significantly improves the QSAR predictions. The techniques and protocol discussed in the present work might be of significant importance while developing QSAR models of various drug targets.

Virender Kumar

Papers

Role of Topological, Electronic, Geometrical, Constitutional and Quantum Chemical Based Descriptors in QSAR: mPGES-1 as a Case Study.