A dataset for machine learning-based QSAR models establishment to screen beta-lactamase inhibitors using the FARM -BIOMOL chemical library.

Abstract

Objectives: Beta-lactamase is a bacterial enzyme that deactivates beta-lactam antibiotics, and it is one of the leading causes of antibiotic resistance problems globally. In current drug discovery research, molecular simulation, like molecular docking, has been routinely integrated to virtually screen an enzyme inhibitory effect. However, a commonly known limitation of molecular docking is a low percent success rate. Previously, we reported a proof-of-concept of combining machine learning with a quantitative structure-activity relationship (QSAR) model that overcame this limitation ( https://doi.org/10.1186/s13065-024-01324-x ). Here, we presented and navigated the dataset used in our previous report, including sixty trained models (thirty for random forest and another thirty for logistic regression).

Data description: This data note has three essential parts. The first part is an in vitro beta-lactamase inhibitory screening of eighty-nine bioactive molecules. The second part consisted of three molecular docking approaches (AutoDock Vina, DOCK6, and consensus docking). The last part is machine learning integrated with QSAR models. Therefore, this data note is vital for further model development to increase performance.