A system and method for predicting the toxicity of pollutants based on a cross-species AHR-qAOP prediction model
By constructing a cross-species AHR-qAOP prediction model, the problems of incomplete species coverage and low prediction accuracy of traditional models are solved, enabling rapid and quantitative pollutant risk assessment and efficient cross-species risk assessment, thus overcoming the limitations of traditional animal experiments.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING NORMAL UNIV AT ZHUHAI
- Filing Date
- 2026-06-17
- Publication Date
- 2026-07-31
AI Technical Summary
The existing AHR-AOP framework is difficult to achieve accurate pollutant risk assessment, especially with low prediction accuracy for different species. Traditional animal experiments are time-consuming and costly, and the model does not cover all species.
A cross-species AHR-qAOP prediction model was constructed. Toxicity data were collected through a data acquisition module, and a cross-species AHR-qAOP prediction model was constructed using linear regression analysis, including a logarithmic transformation unit and a regression analysis unit. Combined with data from *Mulletus edulis* and *Goby*, a quantitative relationship between molecular initiation events and harmful outcomes was established.
It enables rapid screening and quantitative prediction of pollutant toxicity, improves the efficiency of cross-species risk assessment, has a wider model coverage, and enhances predictive ability and applicability, overcoming the limitations of traditional animal experiments.
Smart Images
Figure CN122490480A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of pollutant identification and prediction technology, and in particular relates to a pollutant toxicity prediction system and method based on a cross-species AHR-qAOP prediction model. Background Technology
[0002] Dioxins and dioxin-like substances (DLCs) are persistent organic pollutants widely present in aquatic environments, characterized by high lipid solubility, poor degradation, and bioaccumulation, posing a serious threat to aquatic organisms. Traditional toxicity assessments rely on animal experiments, which are not only time-consuming, costly, and low-throughput, but also fail to meet the urgent need for rapid screening and risk assessment of the increasing number of emerging pollutants. The aromatic hydrocarbon receptor (AHR) pathway is a key molecular initiation event mediating the embryonic developmental toxicity of dioxins and DLCs. Studies have shown that after dioxins and DLCs bind to AHR, they regulate the expression of downstream target genes, ultimately leading to adverse outcomes (AO) such as embryonic malformation and death. Based on this clear toxic mechanism, researchers constructed the AHR-AOP framework, linking the activation of AHR by DLCs (molecular initiation event), the formation of AHR / ARNT heterodimers, and the induction of related phase I and phase II metabolic enzyme expression (key events) to individual-level toxic effects (adverse outcomes).
[0003] However, existing AHR-AOP frameworks are merely qualitative descriptions of knowledge cascades, making accurate risk assessment difficult. Constructing quantitative adverse outcome pathways (qAOPs) and corresponding molecular initiation event detection methods is crucial for quantitative prediction of the toxic effects of environmental pollutants. The core of this approach lies in clarifying the quantitative relationship between molecular initiation events (AHR activity) and adverse outcomes (embryo death). The luciferase reporter gene assay, an in vitro cell assay based on the AHR activation mechanism, offers advantages such as high sensitivity and short detection cycles, and has been used for large-scale screening of DLCs in environmental and biological samples. This method provides a reliable technical means to obtain AHR activity data from different species and lays an important foundation for constructing qAOP models. In particular, there is an urgent need to construct AHR-qAOP prediction models capable of cross-species prediction (such as for fish and birds) to improve the efficiency of DLC toxicity prediction and ecological risk assessment. However, existing models are mainly based on toxicity data from freshwater fish. Since different species differ in AHR sensitivity, metabolic capacity, and toxicity response mechanisms, relying on data from only a few species for prediction introduces significant uncertainty, affecting the accuracy of the assessment results. Therefore, it is urgent to incorporate toxicity data from more key ecological species to optimize and expand the existing AHR-qAOP model, and improve its predictive ability and applicability for different species. Summary of the Invention
[0004] The purpose of this invention is to provide a pollutant toxicity prediction method based on cross-species AHR-qAOP, in order to solve the problems of incomplete species coverage and low accuracy in predicting novel DLCs in existing technologies.
[0005] This invention provides the following technical solution: A pollutant toxicity prediction system based on a cross-species AHR-qAOP prediction model includes a data acquisition module, a model building module, and a risk prediction module. The data acquisition module is used to collect toxicity data; The model building module is used to perform linear regression analysis on the toxicity data and the corresponding AHR effect data to build a cross-species AHR-qAOP prediction model. The risk prediction module is used to predict the median lethal dose based on the cross-species AHR-qAOP prediction model.
[0006] Preferably, the toxicity data collected by the data acquisition module includes LD50 toxicity data of conventional DLCs, AHR effect data of conventional DLCs, and toxicity data of *Mulletus edulis*.
[0007] Preferably, the model building module includes a logarithmic transformation unit and a regression analysis unit; The logarithmic transformation unit is used to perform logarithmic transformation on the toxicity data; The regression analysis unit is used to construct the cross-species AHR-qAOP prediction model by using AHR effect data as independent variable X and embryo lethal toxicity data as dependent variable Y, and performing univariate linear regression analysis using the least squares method.
[0008] Preferably, the standard equation of the cross-species AHR-qAOP prediction model is: ; In the formula, a is the intercept of the regression equation, and b is the slope of the regression equation. This represents the logarithmic value of the species-specific AHR activation threshold determined by the in vitro reporter gene assay.
[0009] This invention also provides a pollutant toxicity prediction method based on a cross-species AHR-qAOP prediction model. The method, applied to the aforementioned system, includes the following steps: Collect toxicity data; The toxicity data and the corresponding AHR effect data were subjected to linear regression analysis to construct a cross-species AHR-qAOP prediction model. Based on the cross-species AHR-qAOP prediction model, the median lethal dose is predicted.
[0010] Preferably, the toxicity data includes conventional DLCs LD50 toxicity data, conventional DLCs AHR effect data, and toxicity data of *Mulletus edulis*.
[0011] Preferred methods for constructing cross-species AHR-qAOP prediction models include: Perform a logarithmic transformation on the toxicity data; Then, using AHR effect data as independent variable X and embryo lethal toxicity data as dependent variable Y, a univariate linear regression analysis was performed using the least squares method to construct the cross-species AHR-qAOP prediction model.
[0012] Preferably, the standard equation of the cross-species AHR-qAOP prediction model is: ; In the formula, a is the intercept of the regression equation, and b is the slope of the regression equation. This represents the logarithmic value of the species-specific AHR activation threshold determined by the in vitro reporter gene assay.
[0013] The beneficial effects of this invention are as follows: This invention provides a pollutant toxicity prediction system and method based on a cross-species AHR-qAOP prediction model, which has the following advantages compared with traditional methods: (1) It breaks through the limitations of traditional animal experiments. Compared to traditional toxicity assessments that rely on animal experiments, the AHR-qAOP prediction model developed in this protocol effectively overcomes the problems of long testing times, high costs, low throughput, and ethical controversies associated with animal experiments. This enables rapid screening and quantitative prediction of pollutant toxicity, which is of great significance for addressing the increasing demand for risk assessment of emerging pollutants.
[0014] (2) It has enabled efficient risk assessment across species. This model clarifies the quantitative relationship between molecular initiation events (AHR activity) and harmful outcomes (embryo death), enabling quantitative extrapolation of toxic effects from the molecular to the individual level. Furthermore, a cross-species prediction model covering different species, such as fish and birds, has been constructed, significantly improving the efficiency and applicability of ecological risk assessment.
[0015] (3) It fills the data gaps of key ecological species. Models built solely based on freshwater fish data suffer from incomplete species coverage and high predictive uncertainty. This proposed solution introduces the subtropical estuarine fish *Mulletus zhurensis*, optimizing the model's species representativeness, broadening its species coverage, and enhancing its predictive ability and applicability across different species. Attached Figure Description
[0016] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly described below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 This is a schematic diagram of the pollutant toxicity prediction system structure of the cross-species AHR-qAOP prediction model according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the pollutant toxicity prediction method of the cross-species AHR-qAOP prediction model according to an embodiment of the present invention. Detailed Implementation
[0017] *Mulletus zhurensis* is a subtropical estuarine benthic fish characterized by its small size, transparent embryos, high reproductive rate, short reproductive cycle, ease of laboratory husbandry and management, and sensitivity to environmental pollutants. Its well-defined genome makes it an ideal species for monitoring emerging pollution in estuaries and the ocean, playing an irreplaceable role in freshwater experimental fish research. The AHR (Anaerobic Luciferase Reporter Gene Therapy) for *Mulletus zhurensis* was previously established, providing crucial technical support for this study. Based on this, a cross-species AHR-qAOP prediction model was constructed using species-specific toxicity data from *Mulletus zhurensis*. By comparing the prediction results for novel DLCs with and without data from this species, the study evaluated the role of incorporating estuarine fish toxicity data in improving the model's predictive accuracy.
[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0019] Example 1 Embodiment 1 of this invention provides a pollutant toxicity prediction system based on a cross-species AHR-qAOP prediction model, mainly including a data acquisition module, a model building module, and a risk prediction module, such as... Figure 1 As shown, specifically: The data acquisition module is used to collect toxicity data, including LD50 toxicity data of conventional DLCs, AHR effect data of conventional DLCs, and toxicity data of *Mulletus edulis*. Specifically, the LD50 toxicity data of conventional DLCs is collected by gathering embryo lethal toxicity data of fish and bird species after exposure to conventional DLCs, including the median lethal dose (LD50). The AHR effect data of conventional DLCs is collected by gathering AHR activation effect data of the aforementioned DLCs in the corresponding species, including the effect threshold concentration (…). In fish species, embryonic lethality was associated with the AHR2 subtype, so AHR2 activation concentrations were collected; in bird species, embryonic lethality was associated with the AHR1 subtype, so AHR1 activation concentrations were collected. The toxicity data for *Mulletodon zhuyi* were obtained as follows: embryonic lethality data were determined after exposure to typical DLCs (TCDD and PCB126) using embryonic exposure experiments; simultaneously, the corresponding AHR activation data were determined using a luciferase reporter gene assay based on *Mulletodon zhuyi* AHR2a.
[0020] The model building module is used to perform linear regression analysis on toxicity data and corresponding AHR effect data to construct a cross-species AHR-qAOP prediction model. In this embodiment, the model building module is further divided into a logarithmic transformation unit and a regression analysis unit. The logarithmic transformation unit is used to perform logarithmic transformation on the toxicity data. The regression analysis unit is used to perform univariate linear regression analysis using the least squares method with AHR effect data as the independent variable X and embryo lethal toxicity data as the dependent variable Y, thereby constructing a cross-species AHR-qAOP prediction model. The standard equation of the cross-species AHR-qAOP prediction model constructed in this embodiment is: .
[0021] In the formula, a is the intercept of the regression equation, and b is the slope of the regression equation. : The logarithmic value of the species-specific AHR activation threshold determined by in vitro reporter gene assay.
[0022] The risk prediction module is used to predict the median lethal dose based on the cross-species AHR-qAOP prediction model.
[0023] Example 2 Embodiment 2 of the present invention provides a method for predicting pollutant toxicity based on a cross-species AHR-qAOP prediction model, such as... Figure 2 As shown, it mainly includes the following three steps: S1. Collect toxicity data.
[0024] In this embodiment, the toxicity data includes LD50 toxicity data of conventional DLCs, AHR effect data of conventional DLCs, and toxicity data of *Mulletodon spp.* The specific data collection method is as follows: (1) Collection of conventional DLCs LD50 toxicity data: collect embryo lethal toxicity data of fish and bird species after exposure to conventional DLCs, including median lethal dose (LD50).
[0025] (2) Collection of AHR effect data of traditional DLCs: Collect AHR activation effect data of the above DLCs in the corresponding species, including effect threshold concentration ( Among them, the embryo lethal effect of fish species is associated with the AHR2 subtype, so the concentration of AHR2 activation effect was collected; the embryo lethal effect of bird species is associated with the AHR1 subtype, so the concentration of AHR1 activation effect was collected.
[0026] (3) Acquisition of toxicity data of *Mulletus zhunii*: Embryo lethal toxicity data of *Mulletus zhunii* after exposure to typical DLCs (TCDD and PCB126) were determined by embryo exposure experiment; at the same time, the corresponding AHR activation effect data were determined by luciferase reporter gene method based on AHR2a of *Mulletus zhunii*.
[0027] S2. Perform linear regression analysis on the embryo lethal toxicity data of all species collected and generated above with the corresponding AHR effect data to construct a cross-species AHR-qAOP prediction model.
[0028] Specifically, by performing a logarithmic transformation on the original data, we obtain: , Using AHR effect data as the independent variable X and embryo lethal toxicity data as the dependent variable Y, a univariate linear regression analysis was performed using the least squares method. The specific calculation steps are as follows: 1) Calculate the sample statistic For observational data of n species , ,..., ,calculate: Mean of X: ; Mean of Y: ; Sum of squared deviations of X: ; Sum of the products of the deviations of X and Y: .
[0029] 2) Calculate the regression coefficients Regression equation The formula for calculating the parameters in the formula is: slope ; intercept ; x is the independent variable in the regression equation.
[0030] 3) Model significance test Calculate the coefficient of determination The model's goodness of fit is evaluated, and the significance of the regression relationship is determined through the F-test. ).
[0031] Substituting the regression coefficients obtained above into the equation, we obtain the standard equation for the cross-species AHR-qAOP prediction model: .
[0032] in, The logarithmic value of the species-specific AHR activation threshold determined by in vitro reporter gene assay is used as the model input variable; The logarithm of the median lethal dose (LD50) for this species, predicted by the model, is used as the model output variable.
[0033] S3. Based on the cross-species AHR-qAOP prediction model, predict the median lethal dose.
[0034] Example 3 This third embodiment provides the validation and evaluation of the cross-species AHR-qAOP prediction model.
[0035] 1. Validation of compound selection and effect determination Two novel DLCs (6-OH-BDE47 and 5-Cl-6-OH-BDE47) were selected as validation compounds. Cross-species predictions were performed using a constructed model to obtain the predicted lethal effects of these two novel DLCs on avian (red junglefowl) embryos. Actual lethal concentrations and AHR data for red junglefowl embryos were obtained from previous studies.
[0036] 2. Methods for evaluating the predictive power of models The predicted values of embryo lethal concentrations in red junglefowl using the constructed cross-species AHR-qAOP model were compared with the actual measured values. Three indicators were used to evaluate the model's predictive ability: mean absolute error (MAE), mean absolute percentage error (MAPE), and fold difference. Smaller MAE and MAPE values indicate better predictive ability, and a fold difference closer to 1 indicates even better predictive ability.
[0037] 3. Impact assessment of target species data on model performance To evaluate the effect of *Mulletus edulis* toxicity data on improving the model's predictive accuracy, a new predictive model was constructed without *Mulletus edulis* data as a control. The MAE, MAPE, and Fold difference were compared between the predicted and observed values of the validation compound for the two models (with and without *Mulletus edulis* data). Smaller MAE and MAPE values indicate better predictive ability, and a Fold difference closer to 1 indicates better predictive ability.
[0038] result 1. Toxicity data collection status Embryo lethal toxicity (LD50) and adverse event-responsiveness (AHR) data of five traditional DLCs (TCDD, PCB126, PCB77, PCB105, and PCB118) against eight fish species (zebrafish, pike, white mullet, Japanese killifish, lake sturgeon, blackhead trout, American red-spotted salmon, and lake trout) and three bird species (red junglefowl, ring-necked pheasant, and quail) were collected from the literature. Toxicity data for *Mulletus edulis* are also listed in Table 1.
[0039] Table 1 2. Model Construction Based on the AHR effect data of all species collected above (EC Threshold Using embryo lethality (LD50) data and linear regression analysis, a cross-species AHR-qAOP prediction model was constructed. The constructed model equation is as follows: This equation quantitatively characterizes the dose-response relationship between molecular initiation events (AHR activation effect) and harmful outcomes (embryoletic death).
[0040] 3. Model prediction and validation of embryo lethal effects of novel DLCs The constructed model was used to predict the embryo lethal effects of two novel DLCs (6-OH-BDE47 and 5-Cl-6-OH-BDE47), and the results were compared with the measured values from red junglefowl embryo exposure experiments. The results are shown in Table 2: for 6-OH-BDE47, the measured and predicted LD50 values for red junglefowl embryos were 1.94 nM and 1.66 nM, respectively; for 5-Cl-6-OH-BDE47, the measured and predicted LD50 values for red junglefowl embryos were 0.57 nM and 0.35 nM, respectively. The calculated mean MAE, mean MAPE, and mean Folddifference for the two compounds were 0.25, 26.51%, and 1.4, respectively. These results indicate that the cross-species AHR-qAOP model constructed in this invention has good predictive performance for novel DLCs.
[0041] Table 2 Note: + indicates results that include data on the toxicity of *Mulletus zhurensis*, and - indicates results that do not include data on the toxicity of *Mulletus zhurensis*.
[0042] 4. Impact assessment of the toxicity data of *Mulletus zhuni* and *Goby* on model performance To evaluate the improvement in model prediction accuracy by incorporating toxicity data from the estuarine fish *Mulletodon zhurenense*, a control model was reconstructed without *Mulletodon zhurenense* data. The equation for the control model is as follows: The control model predicted an LD50 of 0.59 nM for 6-OH-BDE47 and 0.15 nM for 5-Cl-6-OH-BDE47 against *Gnaphalium affine* embryos. The control model's mean MAE was 0.89, mean MAPE was 71.64%, and mean Folddifference was 3.54 for both compounds. Comparing the predictive performance of the two models, the inclusion of *Mulletus edulis* toxicity data improved the model's accuracy: MAE decreased from 0.89 to 0.25, MAPE decreased from 71.64% to 26.51%, and Folddifference decreased from 3.54 to 1.4, closer to 1 (Table 2). This result indicates that the inclusion of estuarine fish toxicity data effectively optimized the cross-species AHR-qAOP model and improved the predictive accuracy for novel DLCs.
[0043] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.
Claims
1. A pollutant toxicity prediction system based on a cross-species AHR-qAOP prediction model, characterized in that, It includes a data acquisition module, a model building module, and a risk prediction module; The data acquisition module is used to collect toxicity data; The model building module is used to perform linear regression analysis on the toxicity data and the corresponding AHR effect data to build a cross-species AHR-qAOP prediction model. The risk prediction module is used to predict the median lethal dose based on the cross-species AHR-qAOP prediction model.
2. The pollutant toxicity prediction system based on the cross-species AHR-qAOP prediction model according to claim 1, characterized in that, The toxicity data collected by the data acquisition module includes LD50 toxicity data of conventional DLCs, AHR effect data of conventional DLCs, and toxicity data of *Mulletus edulis*.
3. The pollutant toxicity prediction system based on the cross-species AHR-qAOP prediction model according to claim 1, characterized in that, The model building module includes a logarithmic transformation unit and a regression analysis unit; The logarithmic transformation unit is used to perform logarithmic transformation on the toxicity data; The regression analysis unit is used to construct the cross-species AHR-qAOP prediction model by using AHR effect data as independent variable X and embryo lethal toxicity data as dependent variable Y, and performing univariate linear regression analysis using the least squares method.
4. The pollutant toxicity prediction system based on the cross-species AHR-qAOP prediction model according to claim 3, characterized in that, The standard equation for the cross-species AHR-qAOP prediction model is: ； In the formula, a is the intercept of the regression equation, and b is the slope of the regression equation. This represents the logarithmic value of the species-specific AHR activation threshold determined by the in vitro reporter gene assay.
5. A method for predicting pollutant toxicity based on a cross-species AHR-qAOP prediction model, wherein the method is applied to the system described in any one of claims 1-4, characterized in that, Includes the following steps: Collect toxicity data; The toxicity data and the corresponding AHR effect data were subjected to linear regression analysis to construct a cross-species AHR-qAOP prediction model. Based on the cross-species AHR-qAOP prediction model, the median lethal dose is predicted.
6. The pollutant toxicity prediction method based on the cross-species AHR-qAOP prediction model according to claim 5, characterized in that, The toxicity data includes LD50 toxicity data for conventional DLCs, AHR effect data for conventional DLCs, and toxicity data for *Mulletus edulis*.
7. The pollutant toxicity prediction method based on the cross-species AHR-qAOP prediction model according to claim 5, characterized in that, Methods for constructing cross-species AHR-qAOP prediction models include: Perform a logarithmic transformation on the toxicity data; Then, using AHR effect data as independent variable X and embryo lethal toxicity data as dependent variable Y, a univariate linear regression analysis was performed using the least squares method to construct the cross-species AHR-qAOP prediction model.
8. The pollutant toxicity prediction method based on the cross-species AHR-qAOP prediction model according to claim 7, characterized in that, The standard equation for the cross-species AHR-qAOP prediction model is: ； In the formula, a is the intercept of the regression equation, and b is the slope of the regression equation. This represents the logarithmic value of the species-specific AHR activation threshold determined by the in vitro reporter gene assay.