An antibacterial lead identification of novel 1,3,4-oxadiazole derivatives based on molecular computer aided design approaches

Merging Density Functional Theory (DFT) with the Quantitative Structure-Activity Relationship (2D/3D-QSAR) modeling represents a promising avenue for exploring antibacterial activity and discovering potential drugs effective against both gram-positive and gram-negative microorganisms. In this study, we employed this integrated approach to investigate a newly synthesized and promising class of 1,3,4-oxadiazole derivatives renowned for their high performance as antibacterial agents. To ensure an accurate characterization and thorough description of the targeted biological activity, we systematically evaluated various DFT functionals to precisely predict the geometrical and electronic descriptors of the investigated compounds. These descriptors were crucial for developing and validating the proposed 2D/3D-QSAR models. Our results reveal that the introduction of a donor group enhances the antibacterial activity of the derivatives. The analysis of molecular descriptors underscores the positive impact of this modification on the compounds' efficacy against bacteria. Additionally, our experimental compounds exhibit favorable characteristics concerning oral bioavailability, a crucial aspect in drug development. The identification of robust correlations between antibacterial activity and specific descriptors was achieved by conducting an extensive analysis that encompassed multiple linear regression (MLR), Random Forest (RF), and Artificial Neural Networks (ANN). Subsequently, a partial least squares (PLS) model was employed to construct 3D-QSAR models based on the Comparative Molecular Field Analysis (CoMFA) and Comparative Similarity Indices Analysis (CoMSIA) descriptors. Validation of the output models was performed using leave-one-out and bootstrapping strategies. The resultant 2D/3D-QSAR models demonstrated a high correlation between experimental and predicted activity values. Leveraging these models, the developed MLR model, expressed as pMIC = -12.704 + 0.260 logP – 6.104 10-03 SAG – 51.385 qN33, serves as a valuable tool for predicting antibacterial activity. we indicate that the machine learning methods, Artificial Neural Networks (ANN) and Random Forest (RF), outperform traditional models in accurately predicting antibacterial activity. We designed ten novel molecules, subjecting them to molecular docking and molecular dynamics simulations to predict their optimal postures when docked in the target and gain insights into the formed interactions. Additionally, Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) analysis was applied to discern the potential behavior of these novel compounds in the human body. As a result, the newly suggested chemical, X4, X10, exhibited robust inhibitory potential against gram-negative microbes.