Özet:
In the present study, the toxic potencies of xenobiotics such as halogenated aromatic hydrocarbons inducing 2,3,7,8-tetrachlorodibenzo-p-dioxin/2,3,7,8-tetrachlorodibenzofuran (TCDD/TCDF)-like effects were investigated by quantitative structure-toxicity relationships using their aryl hydrocarbon receptor (AhR) binding affinity data. The descriptor pool was created using SPARTAN 10, DRAGON 6.0 and ADMET 8.0 software and the descriptors were selected using QSARINS (v.2.2.1) software. Data sets were divided into training and test sets. The training sets were comprised of 81% of the complete data set for both models. The generated models for AhR of chemicals with TCDD/F-like effects were internally and externally validated in line with the Organization of Economic Co–operation and Development principles. TCDD-based model had six descriptors from DRAGON 6.0 and ADMET 8.0. TCDF-based model had seven descriptors from DRAGON 6.0 These descriptors were from various blocks including Weighted Holistic Invariant Molecular, Moriguchi Descriptors, and 2D and 3D descriptors blocks. The predictive ability of the generated models was tested for about 1000 diverse group of chemicals from polychlorinated/brominated biphenyls, dioxins/furans, ethers, polyaromatic hydrocarbons with fused heterocyclic rings (i.e. phenoxathiins, thianthrenes and dibenzothiophenes), and polyaromatic hydrocarbons (i.e. halogenated napthalenes and phenanthrenes) with no AhR data. For the external set chemicals, the structural coverage of the generated models was 95.55% and 89.37% for TCDD/F-like chemicals, respectively.
