Allicin is a reactive sulfur-based antioxidant and Fulvic acid is a stable chelator. This paper studies how a Allicin– Fulvic Acid (FA–Al) conjugate improves metalchelating properties using molecular analysis and machine learning. The research predicts that the combination of allicin thiol–sulfoxide reactivity with fulvic acid polycarboxylic oxygen donors will produce a more stable compound that binds to metals better. The research team used DFT (ωB97X-D/def2-TZVP level under PCM-water solvation) to obtain binding energy (ΔE) values and HOMO–LUMO gaps and electrostatic potential (ESP) maps for 16 metal ions which included Cr², Fe³, Hg², Cu², Pb², Zn², Cd², Co², Fe², Mn², Ca², Mg², Al³, Ni², UO² and As³ ions (all cations). The binding energies of the FA-Al conjugate reached between −4.6 eV and −5.1 eV which exceeded the binding energies of fulvic acid (−3.1 to −3.7 eV) and allicin (−2.1 to −2.6 eV) thus proving its stronger thermodynamic binding properties. The conjugate formed dual-site chelation bonds through oxygen-dentate interactions with hard acids Al³, Fe³ and UO² ions and sulfur-assisted soft-core coordination with Pb², Cd² and Hg² cations. The electrostatic potential mapping showed that the carboxylate and sulfinyl sites of the compound possess electron-rich areas which enable multi-dentate coordination. The Random Forest regression model based on Integrating DFT and Machine Learning to Evaluate Detoxification Efficiency of an Allicin– Fulvic Acid Sieun Jeong 16 physicochemical descriptors achieved R² = 0.92 and 0.08 eV mean absolute error when predicting binding energy and affinity through cross-validation. The model's feature importance analysis showed that ionic charge and Pauling electronegativity and ΔGap change values determine the strength of metal binding. The Allicin–Fulvic conjugate shows improved chelation properties and wide-range selectivity and strong thermodynamic stability which makes it suitable for biological multi-metal detoxification applications. The DFT–AI predictive system enables researchers to create new antioxidant and chelating agents with adjustable selectivity and bioavailability through scalable design methods.

References

1. Aguilera, G., Varela, P., & Marcos, M. (2020). Computational insights into heavy metal chelation by humic and fulvic acids. Journal of Molecular Modeling, 26(11), 321. https://doi.org/10.1007/s00894-020-04525-9
2. Benedí, J., Arroyo, C., & Almajano, M. P. (2017). Role of singlet oxygen and sulfur groups in photodynamic and detoxification mechanisms. Photodiagnosis and Photodynamic Therapy, 17, 115–123. https://doi.org/10.1016/j.pdpdt.2016.11.008
3. Brezová, V., Gaburjaková, J., & Gaburjaková, M. (2015). Fullerene and sulfur-containing antioxidants as potential reactive oxygen scavengers. Journal of Photochemistry and Photobiology C, 22, 1–34. https://doi.org/10.1016/j.jphotochemrev.2014.12.001
4. Chen, C. Y., Lee, C. H., & Wang, H. C. (2019). Density functional theory study on chelation mechanisms of oxygen–sulfur hybrid ligands with heavy metals. Chemical Physics Letters, 731, 136619. https://doi.org/10.1016/j.cplett.2019.136619
5. deRosa, M. C., & Crutchley, R. J. (2002). Photosensitized singlet oxygen and its biomedical applications. Coordination Chemistry Reviews, 233–234, 351–371. https://doi.org/10.1016/S0010-8545(02)00034-6
6. Gupta, A., Rawat, P., & Singh, R. (2022). Machine learning approaches to predict metal–ligand stability constants for environmental chelators. Computational and Theoretical Chemistry, 1214, 113850. https://doi.org/10.1016/j.comptc.2022.113850
7. Hanwell, M. D., et al. (2012). Avogadro: An advanced semantic chemical editor, visualization, and analysis platform. Journal of Cheminformatics, 4(1), 17. https://doi.org/10.1186/1758-2946-4-17
8. Jin, S., Zhang, L., & Zhou, X. (2021). Computational modeling of bioactive sulfur compounds in detoxification pathways. ACS Omega, 6(29), 19043–19052. https://doi.org/10.1021/acsomega.1c02133
9. Kwiatkowski, S., et al. (2018). Mechanistic and computational studies of reactive oxygen and sulfur species interactions. Biomedicine & Pharmacotherapy, 106, 1098–1107. https://doi.org/10.1016/j.biopha.2018.07.049
10. Marković, Z. M., Trajković, V. S., & Todorović-Marković, B. (2012). Photodynamic and redox behavior of sulfur-containing organic antioxidants. Biomaterials, 33(29), 7084–7092. https://doi.org/10.1016/j.biomaterials.2012.06.060
11. Sharma, D., & Basu, S. (2023). Integrating DFT and deep learning to predict metal–ligand binding energies in complex biological systems. Computational Materials Science, 225, 112161. https://doi.org/10.1016/j.commatsci.2023.112161
12. Yamago, S., et al. (1995). In vivo biological behavior of fullerene–based antioxidants. Chemical & Pharmaceutical Bulletin, 43(8), 1234–1240. https://doi.org/10.1248/cpb.43.1234