Curated Database and Preliminary AutoML QSAR Model for 5-HT1A Receptor

Abstract

Introduction of a new drug to the market is a challenging and resource-consuming process. Predictive models developed with the use of artificial intelligence could be the solution to the growing need for an efficient tool which brings practical and knowledge benefits, but requires a large amount of high-quality data. The aim of our project was to develop quantitative structure–activity relationship (QSAR) model predicting serotonergic activity toward the 5-HT1A receptor on the basis of a created database. The dataset was obtained using ZINC and ChEMBL databases. It contained 9440 unique compounds, yielding the largest available database of 5-HT1A ligands with specified pKi value to date. Furthermore, the predictive model was developed using automated machine learning (AutoML) methods. According to the 10-fold cross-validation (10-CV) testing procedure, the root-mean-squared error (RMSE) was 0.5437, and the coefficient of determination (R2) was 0.74. Moreover, the Shapley Additive Explanations method (SHAP) was applied to assess a more in-depth understanding of the influence of variables on the model’s predictions. According to to the problem definition, the developed model can efficiently predict the affinity value for new molecules toward the 5-HT1A receptor on the basis of their structure encoded in the form of molecular descriptors. Usage of this model in screening processes can significantly improve the process of discovery of new drugs in the field of mental diseases and anticancer therapy.

Do AutoML-Based QSAR Models Fulfill OECD Principles for Regulatory Assessment? A 5-HT1A Receptor Case

“…The guidance presents algorithms commonly used in the QSAR modeling (Univariate regression, Multiple Linear Regression, Principal Component Analysis, Principal Component Regression, Partial Least Squares, Artificial Neural Nets, Fuzzy Clustering and Regression, K-nearest Neighbour Clustering, Genetic Algorithms) [ 21 ]. In our previous work, the obtained results were presented with the use of an automated machine learning technique in comparison with a simpler computational technique-linear regression model (LM) [ 22 ]. The LM structure and the principle of operation are easy to read and understand by a human.…”

“…The curated database containing 9440 unique ligands of 5-HT 1A receptor, collected from ZINC and ChEMBL databases, was used in this study [ 28 , 29 ]. Curation of the database was described in previous research [ 22 ]. The affinity of compounds was presented by the negative logarithm of the constant inhibition, the pKi value.…”

“…In our previous work, the number of inputs was reduced from over 1200 to 216 descriptors, which were then used to develop a preliminary model [ 22 ]. In this work, we use these 216 descriptors as the primary dataset.…”

“…Compared to our preliminary AutoML model, which contained 342 models [22], the final QSAR model has fewer models (17) for better transparency. The authors of the OECD principles are particularly careful when, in the case of models based on neural networks, users must rely on a validation process to determine whether an ambiguous algorithm can produce reliable results in regulatory applications.…”

“…The aim of this work was to create a QSAR model of ligand affinity for serotonin 5-HT 1A receptor according to the above principles. The presented work is a continuation of our research on the published preliminary model and the curated database [ 22 ]. To comply with all the OECD principles, we focused on reducing the number of descriptors while maintaining high-quality affinity predictions.…”