Fig. 3

Optimization of the 7C model’s source and sink parameters. a The heatmap represents the RMSE value of the 7C model using literature-derived values for the body parameters with the exception of F_gill and k_m, which are set according to the values on their respective axes. RMSE values are calculated based on the predicted log₁₀ BCF vs. observed log₁₀ BCF for all 76 chemicals in the adult zebrafish dataset. The blue lines represent the values of F_gill and k_m that parameterize the 7C model to predict the BCFs with RMSE = 1.0. The white dots represent the resulting parameter values from 1000 PSOs optimizing only F_gill and k_m. b The predicted log₁₀ BCF values estimated by the 7C model from the log₁₀K_ow values for each of the chemicals in the training and validation sets vs. the observed log₁₀ BCF values after optimization of the F_gill and k_m values. c The blue dots represent the resulting parameter values from 1000 PSOs optimizing F_gill and k_m so that the 7C model accurately predicts each of the 76 chemicals independently (relative error of the predicted vs. observed log₁₀ BCF < 10^− 3). The black line represents the median value of F_gill for a range of k_m across all 76 chemical-specific optimizations. The red dots are the resulting parameter values from the 1000 PSOs optimizing F_gill and k_m from the 7C model for all chemicals