மருந்து வளர்சிதை மாற்றம் மற்றும் நச்சுயியல் இதழ்
திறந்த அணுகல்

சுருக்கம்

Accurate, fast and easy-to-understand Density Functional Tight Binding and Machine Learning QSAR for the DPPH and ABTS antioxidant activity of phenolic compounds based on No-Code Freeware

Andrés Halabi, Millaray Hernández and Hans Lenes2

In this study, we developed various Quantitative Structure-Activity Relationship (QSAR) models for the 1,1-diphenyl-2-picrylhydrazyl (DPPH) and 2,2-azino-bis-3-ethylbenzthiazoline-6-sulphonic acid (ABTS) experimental values of 55 phenolic antioxidants based on Conceptual DFT descriptors calculated with the Density Functional Tight Binding (DFTB) GFN1-xľB. Machine learning algorithms were used for feature selection and regression analysis, and Leave-One-Out Cross-Validation was used for both multiple linear regression (MLR) and sequential minimal optimization regression (SMOreg). For ABTS activity, two models were obtained with a correlation coefficient of 0.94 (MLR) and 0.92 (SMOreg). For DPPH activity, two models were obtained with a Correlation Coefficient of 0.93 (MLR) and 0.91 (SMOreg). The number of phenolic groups in the molecule, Bond Dissociation Enthalpy and radical Fukui of the most active phenolic oxygen were enough to properly predict the Radical Scavenging Activity (RSA) of phenols. Both developed QSAR models were carried out following the Organisation for Economic Co-operation and Development (OECD) recommendations on QSAR models. Considering the importance of antioxidant activities in medicine, pharma, and food industries, this study proposes a highly valuable and cheap method. It is also extremely easy to understand as we need only three descriptors that are directly related to the known chemistry of the substances.