In silico toxicity assessment of diverse organic chemicals in channel catfish ( Ictalurus punctatus ): Duration-specific (24–96 hr) QSAR modeling, mechanistic interpretation, and data gap filling

Abstract The channel catfish (Ictalurus punctatus) is vital to aquatic ecosystems and the aquaculture industry, regulating food webs, cycling nutrients, and indicating ecosystem health because of its habitat adaptability. The rise of harmful industrial organic chemicals exposes the limits of traditional toxicity tests, such as animal testing, which are often unreliable, costly, time-consuming, and ethically problematic. To overcome these issues, in silico methods, especially Quantitative Structure-Activity Relationships (QSARs), can offer a reliable, animal-free alternative for predicting toxicity. Using toxicity endpoint data from the U. S. Environmental Protection Agency’s (USEPA) ECOTOXicology database (ECOTOX), we developed the first species-specific QSAR models for I. punctatus, based on two-dimensional (2D) molecular descriptors, enabling safer and more ethical environmental assessments. We developed four duration-specific (24, 48, 72, and 96 hr) QSAR models using curated mortality datasets comprising 187, 89, 49, and 246 compounds, respectively. The models demonstrated strong goodness-of-fit, robustness, and predictive power, validated both internally (R2=0.711 − 0.828; Q2LOO=0.646 − 0.769) and externally (Q2F1=0.717 − 0.773; Q2F2=0.715 − 0.761). To ensure our models are practically reproducible, we predicted toxicity for over 1,500 new chemicals, identifying the ten most and least toxic compounds and their specific structural characteristics, which will help design safer chemicals and manage sustainable channel catfish populations to support a healthy aquatic ecosystem.