水生生態系におけるポリ塩化ナフタレンのインシリコ毒性評価：モンテカルロ法に基づくQSAR研究

The bio-accumulative nature of polychlorinated naphthalenes (PCNs) due to their slow biodegradation raise concerns regarding adverse effects on aquatic organisms and potential human health risks through trophic transfer. The persistent and traditional toxicity assessment of PCNs in aquatic species is constrained by experimental complexity, resource demands, and ethical considerations. Consequently, in silico approaches, particularly QSAR modeling in accordance with OECD guidelines, have emerged as robust alternatives to address data gaps and support hazard evaluation. In the present investigation, a Monte Carlo-based QSAR methodology employing CORAL software was utilized to assess the toxicity of PCNs in aquatic environments. An integrated dataset comprising 225 compounds with toxicity endpoints for green algae, Daphnia magna, and fish was assembled. Molecular structures were encoded using the Simplified Molecular Input Line Entry System (SMILES). Statistical evaluation confirmed the predictive robustness of the models, with Split 4 yielding superior external validation performance (r² = 0.9490, Q² = 0.9453, IIC = 0.9157, CII = 0.9633, CCC = 0.9661, r²_m = 0.8549, MAE = 0.192, S = 0.224, F = 1023). These findings substantiate the reliability of the developed QSAR model, underscoring its applicability in environmental toxicology, potential to design safer drug candidate and early-stage hazard assessment.