A high-throughput method for comprehensive toxicity of water quality

Abstract

The application discloses a kind of water quality comprehensive toxicity high flux determination method, the determination method includes using sewage sample to expose the test organism after being contaminated, obtains the phenotypic characteristic data of test organism;Toxicity matrix is constructed;Establish machine learning model, determine the water quality comprehensive toxicity of sewage sample in combination with toxicity matrix.The water quality comprehensive toxicity determination method provided by the application does not need sample concentration enrichment, guarantees the comprehensiveness and accuracy of detection result, and can be applied to the comprehensive toxicity high flux determination of sewage treatment plant whole-process sewage.

Description

Technical Field

[0001] This invention belongs to the field of water quality risk control, and specifically relates to a high-throughput method for determining the comprehensive toxicity of water. Background Technology

[0002] Wastewater generated during human life and production processes is complex in composition, highly variable in water quality, and contains a large number of pollutants with diverse properties, making it one of the key factors causing water ecological security problems. With the gradual advancement of industrialization and urbanization, pollutants generated by human life and production activities are increasing day by day. Large amounts of toxic and harmful pollutants with extremely low concentrations but high risks (such as antibiotics, microplastics and their derivatives, perfluorinated and polyfluoroalkyl substances, and new pollutants such as brominated flame retardants) enter wastewater, leading to a significant increase in the potential risks of wastewater and causing widespread concern about wastewater toxicity worldwide.

[0003] Because wastewater contains a wide variety of pollutants, which not only have direct toxic effects on organisms but may also form complex toxic effects with other pollutants, using the concentration level or toxicity value of a single / specific pollutant is insufficient to accurately reflect the overall toxicity of wastewater and may lead to significant biases. Commonly used methods for evaluating the overall toxicity of wastewater both domestically and internationally include the comprehensive wastewater toxicity assessment method, toxicity identification assessment method, and direct toxicity assessment method. The comprehensive wastewater toxicity assessment method involves exposing standard model organisms to serially diluted samples and measuring the acute toxicity effect of the samples on the test organisms at a fixed exposure time. However, these methods require a significant amount of time to determine the acute toxicity effect values ​​through exposure experiments with multiple serially diluted samples, making it difficult to rapidly detect the toxicity effects of a large number of wastewater samples in a short period. Furthermore, the sensitivity and tolerance of test organisms to different pollutants / water qualities vary, and the differences in wastewater from different treatment process stages are substantial. Therefore, there is an urgent need for a highly adaptable and sensitive high-throughput comprehensive toxicity determination method applicable to the entire wastewater treatment process. Summary of the Invention

[0004] Purpose of the invention: The purpose of this invention is to provide a high-throughput method for determining the overall toxicity of water, thereby improving the existing methods for evaluating the overall toxicity effects of wastewater, which suffer from low detection throughput, variations in the sensitivity and tolerance of test organisms to different pollutants / water qualities, and significant differences in wastewater from different treatment stages.

[0005] Technical solution: The high-throughput method for determining the comprehensive toxicity of water quality according to the present invention includes the following steps: exposing test organisms to toxicity using wastewater samples and obtaining phenotypic characteristic data of the test organisms; constructing a toxicity matrix; establishing a machine learning model and combining it with the toxicity matrix to determine the comprehensive toxicity of the wastewater samples.

[0006] Preferably, the exposure time is 24 hours.

[0007] Preferably, the wastewater samples are pretreated before exposing the test organisms to the toxic substances.

[0008] Preferably, the wastewater sample pretreatment process involves filtering the wastewater sample through a 0.22 µm aqueous filter membrane.

[0009] Preferably, the test organisms are algal cells and fish gill cells.

[0010] More preferably, the test organisms are *Cercopithecus argyrophylla* cells and rainbow trout gill cells.

[0011] Preferably, after exposing the test organism to the toxin using wastewater samples, the phenotypic characteristics of the test organism are obtained by using multiplex fluorescence staining, high-content automated imaging, and cell morphology feature extraction.

[0012] Preferably, the multiplex fluorescent staining agent for algal cell staining consists of Hoechst 33342 dye, concanavalin A / Alexa Fluor 488 dye, SYTO 14 dye, phalloidin / Alexa Fluor 568 and wheat-germ agglutinin / Alexa Fluor 555 dye; the multiplex fluorescent staining agent for fish gill cell staining consists of Hoechst 33342 dye, concanavalin A / Alexa Fluor 488 dye, SYTO 14 dye, phalloidin / Alexa Fluor 568 and wheat-germ agglutinin / Alexa Fluor 555 dye, and MitoTracker Deep Red dye.

[0013] Preferably, the high-content automated imaging is performed using a high-content cell imaging and analysis system to automatically acquire subcellular structure images of algal cells and fish gill cells inoculated in 4 to 8 parallel experiments in a well plate at high throughput.

[0014] Preferably, the image acquisition conditions of the high-content cell imaging and analysis system are as follows: 9 (3×3) imaging field points are set in each well of the plate, using 2×2 pixel merging. Each point automatically captures 5-color fluorescence channel images and 3 bright-field channel images originating from different z-axis focal points. A 63x immersion objective is used to acquire subcellular structure images of algal cells, and a 20x immersion objective is used to acquire subcellular structure images of fish gill cells. The excitation / emission wavelengths of the 5-color fluorescence channels used for automatic imaging of algal cells are: DNA 376~398 nm / 417~477 nm, ER 442~502 nm / 503~538 nm, RNA 491~571 nm / 573~613 nm, AGP 502~622 nm / 622~662 nm, Cy5 588~668 nm / 652~732 nm. The excitation / emission wavelengths of the 5-color fluorescence channels used for automatic imaging of fish gill cells are: DNA... 376~398 nm / 417~477nm, ER 442~502 nm / 503~538 nm, RNA 491~571 nm / 573~613 nm, AGP 502~622 nm / 622~662 nm, Mito 588~668 nm / 672~712 nm.

[0015] Preferably, the cell morphology feature extraction step is as follows: The cells, nuclei, and cytoplasm of each image are located, and the image quality must meet the following conditions: average image intensity of 10-240, image focus score >0.5, standard deviation of image edge precession <0.2, cell area of ​​50-500, number of cell debris holes <5, cell density >50, and cell nucleus staining clarity >1.5; The gray-level co-occurrence matrix algorithm for texture feature analysis is used to calculate the morphology, intensity, texture, brightness, average gray level of each cell, and the minimum distance, adjacency value, and aggregation degree between cells, obtaining the cell morphology feature items of each cell and the arithmetic mean of the cell morphology feature values ​​corresponding to each cell morphology feature item.

[0016] Preferably, 5797 morphological characteristics of each cell were obtained.

[0017] Preferably, the toxicity matrix is ​​constructed by clustering algal cell and fish gill cell phenotypic feature data after feature item filtering and feature value standardization; the feature item filtering excludes collinear and intersecting feature items and retains feature items with feature values ​​not equal to 0; the feature value standardization method is Z-Score method and maximum-minimum method; the clustering method is to classify and integrate feature items according to the subcellular structure corresponding to the feature items, and the arrangement is classified into ultra-high-dimensional feature items composed of algal cell DNA, algal cell endoplasmic reticulum, algal cell nucleosomes and cytoplasmic RNA, algal cell actin and Golgi apparatus and plasma membrane, algal cell chloroplasts, algal cell bright field, fish gill cell DNA, fish gill cell endoplasmic reticulum, fish gill cell nucleosomes and cytoplasmic RNA, fish gill cell actin and Golgi apparatus and plasma membrane, fish gill cell mitochondria, and fish gill cell bright field.

[0018] Preferably, the machine learning model is established based on the acute toxicity effect values ​​and phenotypic feature data of algal cells and fish gill cells, using one of the following models: random forest, XGBoost, Lasso regression, content-based recommendation, and support vector machine.

[0019] Preferably, the method for determining the comprehensive water toxicity of wastewater samples is as follows: the dimensionality reduction of the constructed toxicity matrix is ​​performed using least partial squares discriminant analysis to obtain the comprehensive water toxicity characteristic variables, which are then substituted into a machine learning model to obtain the comprehensive water toxicity of the wastewater samples; the comprehensive water toxicity of the wastewater samples is the numerical value of the acute toxicity effect caused by the wastewater samples, defined as the concentration (EC50) that causes the 10% maximum effect. 10 )express.

[0020] Preferably, the comprehensive toxicity characteristic variables of the water quality are 12 items.

[0021] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0022] (1) This invention provides a method for detecting the comprehensive toxicity of wastewater with high throughput and rapid detection, which solves the problems of cumbersome detection steps, long time consumption and low detection throughput in existing methods.

[0023] (2) Based on high-content cell imaging technology and machine learning model, this invention automatically captures the toxic effects of the comprehensive water quality toxicity of sewage samples at the cellular and subcellular structural levels, and accurately quantifies the acute toxicity effect of sewage samples, thereby greatly simplifying the measurement steps and ensuring the comprehensiveness and accuracy of the test results.

[0024] (3) This invention utilizes the different response characteristics of algal cells and fish gill cells to the toxic effects of sewage samples to construct a toxicity matrix, which reduces the selectivity of existing methods for specific sewage samples, enhances the universality of the high-throughput method for comprehensive water toxicity determination for various types of sewage samples, and improves the practical applicability of the method. Attached Figure Description

[0025] Figure 1 This is a flowchart of the high-throughput method for determining the comprehensive toxicity of water in this invention;

[0026] Figure 2 Images of the subcellular structures of algal cells and fish gill cells in Example 1 of this invention;

[0027] Figure 3 This is the toxicity matrix constructed based on the phenotypic characteristics of algal cells and fish gill cells in Example 1 of the present invention;

[0028] Figure 4 The comprehensive toxicity of the wastewater samples from the entire process of Plant B in Example 2 of this invention;

[0029] Figure 5 The comprehensive toxicity of the effluent from plants C, D, and E in Example 3 of this invention is given. Detailed Implementation

[0030] The technical solution of the present invention will be further described below with reference to the embodiments and accompanying drawings.

[0031] Example 1

[0032] This embodiment applies to influent samples from a municipal wastewater treatment plant in Jiangsu Province. Plant A has a daily treatment capacity of 80,000 cubic meters per day, and the influent contains COD of 254.0 mg / L, total nitrogen of 29.27 mg / L, and total phosphorus of 2.07 mg / L. A high-throughput method for determining the comprehensive toxicity of water is established, and its specific steps are as follows:

[0033] Step 1: Filter the wastewater sample from Plant A through a 0.22 µm water-based filter membrane.

[0034] Step 2: Using *Creepingia spp.* cells and rainbow trout gill cells as test organisms, the *Creepingia spp.* cells and rainbow trout gill cells were exposed to the toxin for 24 h using filtered wastewater samples. Multiplex fluorescent staining agents were prepared using Hoechst 33342 dye, concanavalin A / Alexa Fluor 488 dye, SYTO 14 dye, phalloidin / Alexa Fluor 568, and wheat-germ agglutinin / Alexa Fluor 555 dye. These agents were used to perform multiplex fluorescent staining on the exposed algal cells. MitoTracker Deep... Multiplex fluorescent staining agents were prepared using red dye to perform multiplex fluorescent staining on fish gill cells exposed to the drug. A high-throughput, automated high-content cell imaging and analysis system was used to acquire subcellular structure images of algal and fish gill cells inoculated in six parallel experiments in well plates. Subcellular structure images of algal cells were acquired using a 63x immersion objective, and subcellular structure images of fish gill cells were acquired using a 20x immersion objective. The excitation / emission wavelengths of the five fluorescent channels used for automated algal cell imaging were: DNA 376~398 nm / 417~477 nm, ER 442~502 nm / 503~538 nm, RNA 491~571 nm / 573~613 nm, AGP 502~622 nm / 622~662 nm, Cy5 588~668 nm / 652~732 nm. The excitation / emission wavelengths of the five fluorescent channels used for automated fish gill cell imaging were: DNA 376~398 nm / 417~477 nm, ER 588~668 nm / 652~732 nm. The wavelength ranges for each well in the plate are 442~502 nm / 503~538 nm, RNA 491~571 nm / 573~613 nm, AGP 502~622 nm / 622~662 nm, and Mito 588~668 nm / 672~712 nm. Nine (3×3) imaging field points are set in each well, using a 2×2 pixel merging method. Each field point automatically captures five fluorescence channel images and three bright-field channel images originating from different z-axis focal points. Cell morphology features are extracted using the automatically captured images, and the location of cells, nuclei, and cytoplasm in each image is determined. Image quality must meet the following conditions: average image intensity of 10~240, image focus score >0.5, and standard deviation of image edge pre-exposure <0.2. Cell area ranges from 50 to 500 g / m², cell debris pore count is <5, cell density is >50 cells / m², and nuclear staining clarity is >1.5. A gray-level co-occurrence matrix algorithm based on texture feature analysis was used to calculate the morphology, intensity, texture, brightness, average gray level, minimum distance between cells, adjacency value, and aggregation degree of each cell. This yielded 5797 morphological features for each cell and the arithmetic mean of the corresponding morphological feature values.

[0035] Step 3: Eliminate collinear and intersecting feature terms within the phenotypic feature data of algal cells and fish gill cells, retaining feature terms with non-zero eigenvalues; standardize the phenotypic feature data of algal cells and fish gill cells after feature filtering using the Z-Score method and the maximum-minimum method; classify and integrate feature terms according to the subcellular structures corresponding to the feature terms, and arrange and classify the phenotypic feature data of algal cells and fish gill cells after feature filtering and feature value standardization into ultra-high-dimensional feature terms composed of algal cell DNA, algal cell endoplasmic reticulum, algal cell nucleosomes and cytoplasmic RNA, algal cell actin and Golgi apparatus and plasma membrane, algal cell chloroplasts, algal cell bright field, fish gill cell DNA, fish gill cell endoplasmic reticulum, fish gill cell nucleosomes and cytoplasmic RNA, fish gill cell actin and Golgi apparatus and plasma membrane, fish gill cell mitochondria, and fish gill cell bright field, and construct a toxicity matrix.

[0036] Step 4: Based on the acute toxicity effect values ​​and phenotypic characteristics of algal cells and fish gill cells, a machine learning model is established using a random forest model. The dimensionality of the constructed toxicity matrix is ​​reduced using the least partial squares discriminant analysis method to obtain 12 comprehensive water quality toxicity characteristic variables. These variables are then substituted into the machine learning model to obtain the comprehensive water quality toxicity of the influent to Plant A, expressed as the concentration that causes the maximum effect of 10% (EC10).

[0037] like Figure 2 The image shown is an image of the subcellular structure of algal cells and fish gill cells in the influent sample of Plant A, obtained using this method. Figure 3 As shown in Table 1, cell morphology features were extracted to obtain cell phenotypic data of algal cells and fish gill cells, and a toxicity matrix was constructed. The toxicity matrix was then subjected to dimensionality reduction using least partial squares discriminant analysis, resulting in 12 comprehensive water quality toxicity feature variables. These variables were then incorporated into a machine learning model, yielding a comprehensive water quality toxicity of 55.2% for the influent of Plant A.

[0038] Table 1. 12 water quality comprehensive toxicity characteristic variables obtained in Example 1

[0039] Comprehensive water quality toxicity characteristics variables numerical values DNA_1 -20.173203 DNA_2 3.73463273 RNA_1 -15.385017 RNA_2 4.7597349 ER_1 -21.962306 ER_2 2.68094709 AGP_1 -18.088792 AGP_2 2.38816954 Chl -18.7958 Mito 4.47955748 BR_1 -17.150802 BR_2 7.18030231

[0040] Example 2

[0041] Unlike Example 1, this example applies to wastewater samples from the entire process of a municipal wastewater treatment plant in Southwest China, including samples from the inlet, aerated grit chamber, anoxic tank, aerobic tank, secondary sedimentation tank, sand filter, disinfection tank, and effluent. Plant B has a daily treatment capacity of 450,000 cubic meters / day. The influent contains COD of 241.1 mg / L, total nitrogen of 27.02 mg / L, and total phosphorus of 2.94 mg / L; the effluent contains COD of 55.40 mg / L, total nitrogen of 10.37 mg / L, and total phosphorus of 0.38 mg / L. A high-throughput method for determining comprehensive water toxicity is established, with the following specific steps:

[0042] Step 1: Filter 8 wastewater samples from the entire process of Plant B through a 0.22 µm water-based filter membrane.

[0043] Step 2: Using *Creepingia spp.* cells and rainbow trout gill cells as test organisms, the *Creepingia spp.* cells and rainbow trout gill cells were exposed to the toxin for 24 h using filtered wastewater samples. Multiplex fluorescent staining agents were prepared using Hoechst 33342 dye, concanavalin A / Alexa Fluor 488 dye, SYTO 14 dye, phalloidin / Alexa Fluor 568, and wheat-germ agglutinin / Alexa Fluor 555 dye. These agents were used to perform multiplex fluorescent staining on the exposed algal cells. MitoTracker Deep... Multiplex fluorescent staining agents were prepared using red dye to perform multiplex fluorescent staining on fish gill cells exposed to the drug. A high-throughput automated imaging and analysis system was used to acquire subcellular structure images of algal and fish gill cells inoculated in four parallel experiments in well plates: a 63x immersion objective was used to acquire subcellular structure images of algal cells, and a 20x immersion objective was used to acquire subcellular structure images of fish gill cells. The excitation / emission wavelengths of the five fluorescent channels used for automated algal cell imaging were: DNA 376~398 nm / 417~477 nm, ER 442~502 nm / 503~538 nm, RNA 491~571 nm / 573~613 nm, AGP 502~622 nm / 622~662 nm, Cy5 588~668 nm / 652~732 nm. The excitation / emission wavelengths of the five fluorescent channels used for automated fish gill cell imaging were: DNA 376~398 nm / 417~477 nm, ER 588~668 nm / 652~732 nm. The wavelength ranges for each well in the plate are 442~502 nm / 503~538 nm, RNA 491~571 nm / 573~613 nm, AGP 502~622 nm / 622~662 nm, and Mito 588~668 nm / 672~712 nm. Nine (3×3) imaging field points are set in each well, using a 2×2 pixel merging method. Each field point automatically captures five fluorescence channel images and three bright-field channel images originating from different z-axis focal points. Cell morphology features are extracted using the automatically captured images, and the location of cells, nuclei, and cytoplasm in each image is determined. Image quality must meet the following conditions: average image intensity of 10~240, image focus score >0.5, and standard deviation of image edge pre-exposure <0.2. Cell area ranges from 50 to 500 g / m², cell debris pore count is <5, cell density is >50 cells / m², and nuclear staining clarity is >1.5. A gray-level co-occurrence matrix algorithm based on texture feature analysis was used to calculate the morphology, intensity, texture, brightness, average gray level, minimum distance between cells, adjacency value, and aggregation degree of each cell. This yielded 5797 morphological features for each cell and the arithmetic mean of the corresponding morphological feature values.

[0044] Step 3: Eliminate collinear and intersecting feature terms within the phenotypic feature data of algal cells and fish gill cells, retaining feature terms with non-zero eigenvalues; standardize the phenotypic feature data of algal cells and fish gill cells after feature filtering using the Z-Score method and the maximum-minimum method; classify and integrate feature terms according to the subcellular structures corresponding to the feature terms, and arrange and classify the phenotypic feature data of algal cells and fish gill cells after feature filtering and feature value standardization into ultra-high-dimensional feature terms composed of algal cell DNA, algal cell endoplasmic reticulum, algal cell nucleosomes and cytoplasmic RNA, algal cell actin and Golgi apparatus and plasma membrane, algal cell chloroplasts, algal cell bright field, fish gill cell DNA, fish gill cell endoplasmic reticulum, fish gill cell nucleosomes and cytoplasmic RNA, fish gill cell actin and Golgi apparatus and plasma membrane, fish gill cell mitochondria, and fish gill cell bright field, and construct a toxicity matrix.

[0045] Step 4: Based on the acute toxicity effect values ​​and phenotypic characteristics of algal cells and fish gill cells, a machine learning model is established using a random forest model. The dimensionality of the constructed toxicity matrix is ​​reduced using the least partial squares discriminant analysis method to obtain 12 comprehensive water quality toxicity characteristic variables. These variables are then substituted into the machine learning model to obtain the comprehensive water quality toxicity of the wastewater samples from the entire process of Plant B, expressed as the concentration that causes the maximum effect of 10% (EC10).

[0046] This method yielded subcellular structure images of algal cells and fish gill cells from the entire process samples of Plant B. Cell morphological features were extracted to obtain corresponding cell phenotypic feature data, and a toxicity matrix was constructed. Least partial squares discriminant analysis was used to reduce the dimensionality of the toxicity matrix of the entire process wastewater samples from Plant B, resulting in 12 comprehensive water quality toxicity feature variables for each sample. These variables were then fed into a machine learning model to obtain the comprehensive water quality toxicity of wastewater samples from the inlet, aerated grit chamber, anoxic tank, aerobic tank, secondary sedimentation tank, sand filter, disinfection tank, and effluent outlet of Plant B. Figure 4 As shown, the percentages are 36.0%, 42.3%, 67.8%, 56.3%, 58.3%, 64.4%, 56.2%, and 60.6%, respectively.

[0047] Example 3

[0048] Unlike Example 1, this example applies to effluent samples from three municipal wastewater treatment plants in the Beijing-Tianjin-Hebei region. Plants C, D, and E have a daily treatment capacity of 1.2 to 2.8 million cubic meters per day, with effluent containing 42.00 to 58.89 mg / L COD, 6.26 to 10.09 mg / L total nitrogen, and 0.09 to 0.35 mg / L total phosphorus. A high-throughput method for determining the comprehensive toxicity of water is established, and its specific steps are as follows:

[0049] Step 1: The wastewater samples from plants C, D, and E were filtered through a 0.22 µm water-based filter membrane.

[0050] Step 2: Using *Creepingia spp.* cells and rainbow trout gill cells as test organisms, the *Creepingia spp.* cells and rainbow trout gill cells were exposed to the toxin for 24 h using filtered wastewater samples. Multiplex fluorescent staining agents were prepared using Hoechst 33342 dye, concanavalin A / Alexa Fluor 488 dye, SYTO 14 dye, phalloidin / Alexa Fluor 568, and wheat-germ agglutinin / Alexa Fluor 555 dye. These agents were used to perform multiplex fluorescent staining on the exposed algal cells. MitoTracker Deep... Multiplex fluorescent staining agents were prepared using red dye to perform multiplex fluorescent staining on fish gill cells exposed to the drug. High-throughput automated acquisition of subcellular structure images of algal and fish gill cells inoculated in eight parallel experiments using a high-content cell imaging and analysis system was employed: a 63x immersion objective was used to acquire subcellular structure images of algal cells, and a 20x immersion objective was used to acquire subcellular structure images of fish gill cells. The excitation / emission wavelengths of the five fluorescent channels used for automated algal cell imaging were: DNA 376~398 nm / 417~477 nm, ER 442~502 nm / 503~538 nm, RNA 491~571 nm / 573~613 nm, AGP 502~622 nm / 622~662 nm, Cy5 588~668 nm / 652~732 nm. The excitation / emission wavelengths of the five fluorescent channels used for automated fish gill cell imaging were: DNA 376~398 nm / 417~477 nm, ER 588~668 nm / 652~732 nm. The wavelength ranges for each well in the plate are 442~502 nm / 503~538 nm, RNA 491~571 nm / 573~613 nm, AGP 502~622 nm / 622~662 nm, and Mito 588~668 nm / 672~712 nm. Nine (3×3) imaging field points are set in each well, using a 2×2 pixel merging method. Each field point automatically captures five fluorescence channel images and three bright-field channel images originating from different z-axis focal points. Cell morphology features are extracted using the automatically captured images, and the location of cells, nuclei, and cytoplasm in each image is determined. Image quality must meet the following conditions: average image intensity of 10~240, image focus score >0.5, and standard deviation of image edge pre-exposure <0.2. Cell area ranges from 50 to 500 g / m², cell debris pore count is <5, cell density is >50 cells / m², and nuclear staining clarity is >1.5. A gray-level co-occurrence matrix algorithm based on texture feature analysis was used to calculate the morphology, intensity, texture, brightness, average gray level, minimum distance between cells, adjacency value, and aggregation degree of each cell. This yielded 5797 morphological features for each cell and the arithmetic mean of the corresponding morphological feature values.

[0051] Step 3: Eliminate collinear and intersecting feature terms within the phenotypic feature data of algal cells and fish gill cells, retaining feature terms with non-zero eigenvalues; standardize the phenotypic feature data of algal cells and fish gill cells after feature filtering using the Z-Score method and the maximum-minimum method; classify and integrate feature terms according to the subcellular structures corresponding to the feature terms, and arrange and classify the phenotypic feature data of algal cells and fish gill cells after feature filtering and feature value standardization into ultra-high-dimensional feature terms composed of algal cell DNA, algal cell endoplasmic reticulum, algal cell nucleosomes and cytoplasmic RNA, algal cell actin and Golgi apparatus and plasma membrane, algal cell chloroplasts, algal cell bright field, fish gill cell DNA, fish gill cell endoplasmic reticulum, fish gill cell nucleosomes and cytoplasmic RNA, fish gill cell actin and Golgi apparatus and plasma membrane, fish gill cell mitochondria, and fish gill cell bright field, and construct a toxicity matrix.

[0052] Step 4: Based on the acute toxicity effect values ​​and phenotypic characteristics of algal cells and fish gill cells, a machine learning model is established using a random forest model. The dimensionality of the constructed toxicity matrix is ​​reduced using the least partial squares discriminant analysis method to obtain 12 comprehensive water quality toxicity characteristic variables. These variables are then substituted into the machine learning model to obtain the comprehensive water quality toxicity of the effluent from plants C, D, and E, expressed as the concentration that causes the maximum effect of 10% (EC10).

[0053] This image shows subcellular structure images of algal cells and fish gill cells from the effluents of plants C, D, and E, obtained using this method. Cell morphological features were extracted to obtain corresponding cell phenotypic feature data, and a toxicity matrix was constructed. Least partial squares discriminant analysis was used to reduce the dimensionality of the toxicity matrices for the wastewater samples from plants C, D, and E, resulting in 12 comprehensive water quality toxicity feature variables for each sample. These variables were then input into a machine learning model to obtain the comprehensive water quality toxicity of the effluents from plants C, D, and E, as shown below. Figure 5 As shown, the percentages are 47.0%, 56.9%, and 47.9%, respectively.

Claims

1. A high-throughput method for determining the comprehensive toxicity of water, characterized in that, Includes the following steps: After exposing the test organisms to the toxins using wastewater samples, phenotypic data of the test organisms were obtained. Construct a toxicity matrix; Establish a machine learning model and combine it with a toxicity matrix to determine the overall water toxicity of wastewater samples; After exposing the test organisms to the toxins using wastewater samples, phenotypic data of the test organisms were obtained by using multiple fluorescence staining, high-content automated imaging, and cell morphology feature extraction. The toxicity matrix is ​​constructed by clustering algal cell and fish gill cell phenotypic feature data after feature item filtering and feature value standardization. The feature item filtering excludes collinear and intersecting feature items and retains feature items with feature values ​​not equal to 0. The feature value standardization method is Z-Score method and maximum-minimum method. The clustering method classifies and integrates feature items according to the subcellular structure corresponding to the feature items, and the arrangement is classified into ultra-high-dimensional feature items composed of algal cell DNA, algal cell endoplasmic reticulum, algal cell nucleosomes and cytoplasmic RNA, algal cell actin and Golgi apparatus and plasma membrane, algal cell chloroplasts, algal cell bright field, fish gill cell DNA, fish gill cell endoplasmic reticulum, fish gill cell nucleosomes and cytoplasmic RNA, fish gill cell actin and Golgi apparatus and plasma membrane, fish gill cell mitochondria, and fish gill cell bright field. The machine learning model is based on the acute toxicity effect values ​​and phenotypic feature data of algal cells and fish gill cells, and is established using one of the following models: random forest, XGBoost, Lasso regression, content-based recommendation, and support vector machine. The method for determining the comprehensive toxicity of wastewater samples is as follows: The dimensionality of the constructed toxicity matrix is ​​reduced using least partial squares discriminant analysis to obtain the comprehensive toxicity characteristic variables, which are then substituted into a machine learning model to obtain the comprehensive toxicity of the wastewater samples. The comprehensive toxicity of the wastewater samples is the numerical value of the acute toxicity effect caused by the wastewater samples, using the concentration EC that induces the 10% maximum effect. 10 express.

2. The high-throughput method for determining comprehensive water toxicity according to claim 1, characterized in that, Wastewater samples are pretreated before being used to expose test organisms to toxins.

3. The high-throughput method for determining the comprehensive toxicity of water according to claim 1, characterized in that, The test organisms were algal cells and fish gill cells.

4. The high-throughput method for determining comprehensive water toxicity according to claim 1, characterized in that, The multiplex fluorescent staining agent for algal cell staining consists of Hoechst 33342 dye, concanavalin A / Alexa Fluor 488 dye, SYTO 14 dye, phalloidin / Alexa Fluor 568 and wheat-germ agglutinin / Alexa Fluor 555 dye; the multiplex fluorescent staining agent for fish gill cell staining consists of Hoechst 33342 dye, concanavalin A / Alexa Fluor 488 dye, SYTO 14 dye, phalloidin / Alexa Fluor 568 and wheat-germ agglutinin / Alexa Fluor 555 dye, and MitoTracker Deep Red dye.

5. The high-throughput method for determining the comprehensive toxicity of water according to claim 1, characterized in that, The high-content automated imaging refers to the use of a high-content cell imaging and analysis system to automatically acquire subcellular structure images of algal cells and fish gill cells inoculated in 4 to 8 parallel experiments in well plates at high throughput.

6. The high-throughput method for determining the comprehensive toxicity of water according to claim 1, characterized in that, The steps for extracting cell morphology features are as follows: The positions of cells, nuclei, and cytoplasm in each image are located. The image quality must meet the following conditions: average image intensity of 10-240, image focus score > 0.5, standard deviation of image edge precession < 0.2, cell area of ​​50-500, number of cell debris holes < 5, cell density > 50, and nucleus staining clarity > 1.

5. The gray-level co-occurrence matrix algorithm for texture feature analysis is used to calculate the morphology, intensity, texture, brightness, average gray level of each cell, and the minimum distance, adjacency value, and aggregation degree between cells, obtaining the cell morphology features of each cell and the arithmetic mean of the cell morphology feature values ​​corresponding to each cell morphology feature.

Images