Development Of QSAR Models On Bioconcentration Factors Of Chemicals In Fish

Bioconcentration factor (BCF) is an important parameter to evaluate the bioconcentration potential of chemicals and plays a critical and irreplaceable role in screening bioaccumulative chemicals which usually bring about high risks to ecology and human being. The quantity of existing commercial chemicals that have to be evaluated is up to140,000. Experimental measurements of BCF are usually costly, laborious, time-consuming and violating the3R principles of animal test. Thus, employing in silico approaches to develop predictive models to obtain BCF were regarded as promising alternatives. According to the principles of development and validation of quantitative structure activity relationship (QSAR) models, two prediction models were developed on BCF of chemicals based on the QSAR approach. The main results are shown as following:(1) Bioconcentration factors data of780organic compounds were collected and divided into training set and validation set randomly as the proportion of4:1to develop a universal QSAR model. First of all, molecular structures of the chemicals were optimized using Gaussian software package based on density functional theory. Using the optimized molecular structures, molecular descriptors were calculated by Dragon software. Finally,3480descriptors were obtained after preliminary screening. A universal QSAR model containing12descriptors was established by multiple linear regression (MLR). The application domain was characterized via Euclidean distance, and the significances of descriptors were interpretated. The validation parameters indicated that the universal QSAR model has satisfied goodness-of-fit, robustness and predictive ability. This model covers a large and diverse training set.(2) A lot of priority contaminants are halogenated compounds, which generally bring adverse impact on human health and ecological environment after enrichment in tissues. Therefore, it is significant to evaluat the capacity of bioaccumulation for halogenated compounds. A QSAR model for halogenated organic compounds of BCF was constructed. The BCF data of374halogenated organic compounds were collected and used for the model development. A local BCF-QSAR model containing9descriptors was established successfully. The validation parameters indicated that the QSAR model has satisfied goodness-of-fit, robustness and predictive ability. This model can be used to predict the bioconcentration factors of halogenated compounds within the application domain.