Mass spectrometric detection method for bisphenol A in milk through rotary evaporation pretreatment and multi-charge peak analysis

Abstract

The invention discloses a mass spectrometric detection method for bisphenol A in milk through rotary evaporation pretreatment and multi-charge peak analysis. The mass spectrometric detection method comprises the following steps: step 1, carrying out rotary evaporation pretreatment on fresh milk; step 2, carrying out rotary evaporation treatment on the milk sample; 3, omics analysis of lipid, metabolites and protein is completed, and UHPLC-Q-Orbitrap HRMS strategy processing based on vDIA is achieved; and 4, analyzing the group database by adopting a multi-charge overlapping peak extraction algorithm and a graph neural network strategy to realize substance identification. The invention belongs to the technical field of mass spectrometry, and solves the problems that the detection error is large when microgram / L-level trace detection of bisphenol A in milk is carried out in the prior art, and false positive is easy to occur due to a wide mass window in the traditional extraction ion chromatography.

Description

Technical Field

[0001] The invention belongs to the technical field of mass spectrometry analysis and relates to a mass spectrometry detection method for bisphenol A in milk by rotary evaporation pre-treatment and multi-charge peak analysis. Background Art

[0002] Currently, the mainstream processing methods for commercially available concentrated milk include thermal evaporation, membrane separation, and freeze concentration. However, all three processes can result in residual contaminants. Thermal evaporation, the most commonly used method in industrial production, rapidly removes water through high-temperature, reduced-pressure evaporation at 60-80°C. However, high temperatures can denature heat-sensitive components in milk (such as vitamins and proteins) and may increase the risk of contaminants like bisphenol A migrating from packaging materials. Membrane separation technologies (such as reverse osmosis and nanofiltration) can selectively separate water at room temperature through pressure, effectively preserving milk flavor and nutrients. However, in practice, they face challenges with membrane fouling, flux attenuation, and high equipment maintenance costs. Freeze concentration, which involves freezing milk at low temperatures and then removing ice crystals, maximizes the preservation of the milk's original quality. However, it is energy-intensive and inefficient, and the final solids concentration typically only reaches 30%-40%, making it difficult to meet the demands of high-concentration multiples.

[0003] None of the above traditional processes are optimized for the pretreatment requirements of trace pollutant detection. The adsorption of pollutants (such as bisphenol A) that may be introduced will affect the quality of milk and affect the μg / L-level trace detection of bisphenol A (BPA) in milk. When using traditional methods for extraction ion chromatography (EIC), the wide mass window can easily lead to false positives. Summary of the Invention

[0004] The purpose of the present invention is to provide a mass spectrometry detection method for bisphenol A in milk with rotary evaporation pre-treatment and multi-charge peak analysis, which solves the problems of large detection errors in the prior art when detecting trace amounts of bisphenol A in milk at the μg / L level, and the proneness to false positives due to the wide mass window of traditional extracted ion chromatography.

[0005] The technical solution adopted by the present invention is a mass spectrometry detection method for bisphenol A in milk with rotary evaporation pre-treatment and multi-charge peak analysis, which is implemented according to the following steps: Step 1, performing rotary evaporation pre-treatment on fresh milk; Step 2, performing rotary evaporation on the milk sample; Step 3: Complete the omics analysis of lipids, metabolites, and proteins, and implement vDIA-based UHPLC-Q-Orbitrap HRMS strategy processing; Step 4: Use the multiple charge overlap peak extraction algorithm and graph neural network strategy to parse the group database and realize substance identification.

[0006] The beneficial effects of the present invention include the following aspects: 1) The optimized low-temperature rotary evaporation concentration technology (40-45°C, 0.08-0.1MPa) is used to significantly reduce the complexity of the milk matrix and improve the detection limit (LOD) of subsequent mass spectrometry analysis to the μg / L level. 2) Strong anti-interference ability: A multiple charge overlapping peak extraction algorithm is used to accurately resolve the multiple charged overlapping peaks of bisphenol A and co-eluting matrix components (such as milk fat and protein degradation products) through deconvolution technology, effectively eliminating background noise interference and improving the signal-to-noise ratio (S / N≥10). 3) Low false positive rate: Compared with the traditional EIC method that only relies on a single mass window to extract ions, the method of the present invention significantly reduces false positive results caused by insufficient mass accuracy through charge state distribution modeling and mass-to-charge ratio offset correction, thereby ensuring the reliability of the detection results.

[0007] In summary, the present invention establishes a high-sensitivity and high-specificity solution for the detection of bisphenol A in milk through the collaborative innovation of rotary evaporation pretreatment and algorithm deconvolution, providing reliable technical support for dairy product safety monitoring. It is not only applicable to liquid milk, but can also be expanded to the detection of bisphenol A in high-fat / high-protein dairy products such as milk powder and whey protein, and has broad practical application value. DETAILED DESCRIPTION

[0008] To address these issues, the present invention proposes low-temperature rotary evaporation as a pretreatment method. By precisely controlling the evaporation pressure and temperature, this method efficiently removes water while significantly reducing the degradation risk of heat-sensitive contaminants such as bisphenol A. It also reduces matrix complexity, providing a high-purity, high-recovery sample foundation for mass spectrometry analysis. In the detection phase, an innovative multiple charge overlap peak extraction algorithm is proposed. Using deconvolution techniques, this algorithm accurately analyzes the multiple charge overlap peak signals generated by bisphenol A and co-eluting interfering substances (such as milk fat and peptides) in the mass spectrometer.

[0009] The mass spectrometric detection method for bisphenol A in milk with rotary evaporation pre-treatment and multi-charge peak analysis of the present invention is implemented according to the following steps: Step 1: Pre-treat fresh milk by rotary evaporation. Fresh milk was collected and 0.1%-0.5% white vinegar or citric acid was added to the fresh milk according to the mass percentage to prevent protein aggregation or the binding of target substances (such as bisphenol A) with milk proteins to prepare a milk sample.

[0010] Step 2: Perform rotary evaporation on the milk sample. The specific process is: 2.1) Configure the rotary evaporator and set the parameters: The rotary evaporator should be equipped with an explosion-proof boiling ball and a low-temperature condensing circulating water bath (4°C). The capacity of the evaporating flask used should be no less than 3 times the volume of the milk sample to avoid violent boiling. The water bath temperature is 40-45°C, and the pressure is reduced to 0.08-0.1 MPa (absolute pressure) to lower the boiling point of the solvent and prevent degradation of bisphenol A caused by high temperature. The rotation speed is 80-100 rpm to form a uniform liquid film on the milk, increase the evaporation surface area, and prevent local overheating or foaming and overflow. 2.2) Concentration: Slowly reduce the pressure to the target pressure and observe the formation of foam. If necessary, pause the pressure reduction and manually shake the evaporating flask to release bubbles. When the evaporation volume reaches 10%-20% of the original volume, stop heating immediately. At this point, most of the water in the milk sample has been removed, leaving a viscous, milky concentrate. 2.3) Post-processing and transfer: Rinse the inner wall of the evaporating flask with a small amount of methanol or acetonitrile (containing 0.1% formic acid), transfer the concentrate to a centrifuge tube, vortex mix, and then solubilize by ultrasound (5-10 min, 20-25°C). Filter through a 0.22 μm organic filter membrane or further perform solid-phase extraction (e.g., HLB column) to remove residual proteins and lipids to obtain a clarified extract for subsequent mass spectrometry analysis.

[0011] Step 3: Complete the omics analysis of lipids, metabolites, and proteins, and implement the vDIA-based UHPLC-Q-Orbitrap HRMS strategy. The specific process is: 3.1) Lipidomics analysis: 500 μL of the clarified extract and 960 μL of the extractant (density 1 mg / kg) were mixed at room temperature and then sonicated in an ice bath for 10-20 minutes. The mixture was then centrifuged at 10,000-12,000 × g (unit, rpm) for 20 ± 2 minutes at 0-4°C, and the supernatant was dried with nitrogen. Finally, the mixture was placed in a methanol and isopropanol (1:1, in / out ) was redissolved in a mixture (1.0 mL) to obtain a mixed solution sample for LC-MS analysis; after obtaining the mixed solution sample, the primary mass spectrum and secondary mass spectrum of the lipid were obtained by injection into the machine; 3.2) Metabolomics analysis: 100 μL of the clarified extract was mixed with 400 μL of metabolite extractant (the metabolite extractant was composed of methanol and acetonitrile in a 1:1 ratio). in / out The mixture was mixed and then ultrasonicated at 0-4°C for 10-20 minutes. The mixture was then incubated at -20°C for 1±0.1 hours to precipitate proteins. The mixture was then centrifuged at 12,000 × g for 15±1 minutes at 0-4°C to obtain supernatant A. After obtaining supernatant A, the sample was injected into the mass spectrometer to obtain the primary mass spectra and secondary mass spectra of the metabolites. 3.3) Proteomic Analysis: Mix 5 mL of the clarified extract with 4% sodium dodecyl sulfate, 1 mM dithiothreitol, and 150 mM Tris-HCl (pH 8). Heat the mixture at 90°C-100°C for 15 ± 1 min and sonicate at 0°C-4°C to obtain a suspension. Subsequently, centrifuge the suspension at 16,000 × g for 30-40 min at 20-25°C to obtain supernatant B. After obtaining supernatant B, inject the sample into the mass spectrometer to obtain mass spectra of the proteins. 3.4) vDIA-based ion acquisition strategy: Separation and elution of metabolites and lipids were performed using a Dionex UltiMate 3000 HPLC system equipped with a C18 column. The polar mobile phase for lipids was aqueous solution (0.1% formic acid), and the nonpolar mobile phase was isopropanol solution (ACN / i-PrOH 10:90). in / out , 0.1% formic acid); the polar mobile phase for metabolites was acetonitrile solution (C2H3N / H2O 3:2, in / out , 0.1% formic acid), the non-polar mobile phase was methanol solution (0.1% formic acid); all "0.1% formic acid" are volume percentage concentrations; A Q-Orbitrap HRMS (Thermo Fisher Scientific, San Jose, USA) was used for mass information capture, with a spray voltage of ±3.5 kV. Due to the presence of trace amounts of proteins and peptides, a stable front-end chromatography was performed to support trace detection. Peptides were separated on a NanoLC1000 (Thermo Scientific) equipped with a QExactive HF (Thermo Fisher Scientific, San Jose, USA). Acetonitrile (0.1% formic acid) and aqueous solution (0.1% formic acid) were used as elution buffers, and peptides were loaded onto a Nano-LC column (100 × 2.1 mM, 5 μm, Thermo Fisher Scientific, Bremen, Germany) at a flow rate of 30 μL / min. The mass spectrometer was operated in positive ion mode (+3.5 kV) and a vDIA data acquisition strategy was used. The vDIA data acquisition strategy included two acquisition modes: MS spectra were acquired at a resolution of 70,000 full-width half-maximum (FWHM); and MS / MS spectra were acquired at a resolution of 35,000 FWHM with step collision energies set to 17.5, 35.0, and 52.5 eV.

[0012] Perform peak clustering-aware extraction from the mass spectra obtained in steps 3.1) to 3.3) to obtain the corresponding primary and secondary fragment mass spectra, and record metabolite information including retention time (RT), isotope distribution characteristics, mass-to-charge ratio, intensity information, and fragmentation characteristics.

[0013] Step 4: Use the multiple charge overlap peak extraction algorithm and graph neural network strategy to parse the group database and realize substance identification. The specific process is: Based on the obtained mass spectrometry data, the fragment tree is used as supplementary data for setting up the protocol, and the structure of the substance is predicted through the primary and secondary fragment mass spectra. Specifically: i) Labeling of substance nodes based on the distribution characteristics of MS1 abundance, retention time (RT) and MS2 spectra; ii) identifying substance nodes and secondary fragments by matching mzCloud, MassBank, Lipidmaps, and KEGG with a mass tolerance of 10 ppm and RT with RSD < 10%, and subsequently distinguishing known substance nodes from unknown substance nodes from mixed labels; iii) Construct edges between the substance node and MS2 based on the relative abundance of the fragments in the MS2 spectrum. Calculate the weight of the edge between the molecule and the secondary fragment. If the calculated result is less than 0, the weight is defined as 0.

[0014] iv) The molecule nodes identified in step ii) will be used as input data for the graph convolution layer, and the final score is obtained by multiplying the results of each secondary fragment in the molecule.

[0015] It can be seen that in this step, the following objectives need to be achieved in essence. Since the multi-charged peaks (MultichargedPeaks) overlap due to the charge state distribution (+2H⁺, +3H⁺), deconvolution is used to separate the overlapping peaks and identify the true molecular weight (MW) and charge state (z) of each component, and deconvolution is performed by Fourier transform; then, the principal component analysis (PCA) algorithm based on singular value decomposition is used to visualize the differences between samples of different categories, and Hotelling's T² diagram is used to perceive and assist in eliminating abnormal variables, where 99% of the critical constraints are set as the threshold; secondly, the Pearson correlation coefficient is used to screen factors related to metabolic differences (|r| ≥ 0.75, p<0.05), chemometric analysis was performed using GraphPad Prism (version 6.5, San Diego, CA, USA), and quantitative results were expressed as mean ± standard error. Again, principal component analysis was used to assess the reproducibility of parallel samples under the same intervention conditions to eliminate outliers. Supervised partial least squares discriminant analysis was performed to classify the classification problem Y according to the predictor variables (differentially fluctuating compounds), and the variable projection importance VIP value was calculated by mean centering and Pareto scaling to screen out substances with significant differences (VIP>1, p <0.05); then, MetaboAnalyst 6.0 (https: / / www.metaboanalyst.ca / ) was used for topological analysis, pathway enrichment analysis of differential metabolites and lipids and proteins was performed to analyze biological characteristics and construct a database of key candidates for graph neural network virtual screening; finally, the data were sorted according to the calculation results of VIP to mine spectral signals that help distinguish samples belonging to different groups, and samples with VIP>1 and p Variables with a value <0.05 were marked as possible markers.

[0016] In the embodiment, after the above steps, the method of the present invention identified 79 metabolites and 870 lipids as significant changes, of which 31 metabolites and 181 lipids were related to bisphenol A invasion and were used for pathway enrichment annotation and functional exploration.

[0017] Example 1 Preliminary steps: 1) Reagent Preparation: 22 deuterated lipid standards were purchased from Biosolve (The Netherlands). HPLC-grade reagents, including methyl tert-butyl ether (MTBE), methanol (MeOH), acetonitrile (CAN), and isopropanol (i-PrOH), were from Sigma-Aldrich (Germany). Ammonium formate (NH4HCOOH, ≥99.9%), formic acid (CH2O2, FA, ≥95%), bisphenol A (BPA, ≥95%), and tris(hydroxymethyl)methylaminomethane (THAM, ≥99%) were purchased from Avanti (USA). Ultrapure water (25°C, 18.2 MΩ cm) was produced using a Milli-Q Plus ultrapure water preparation system (Millipore, USA).

[0018] 2) Instruments used: Dionex Ultimate 3000 ultra-high performance liquid chromatography-quadrupole-orbitrap tandem mass spectrometry (Thermo Fisher Scientific, USA), equipped with a Syringe autosampler system, an electrospray ionization source, and a Hypersil Gold C18 reversed-phase column (100 mM × 2.1 mM × 1.9 μm); KQ5200E ultrasonic cleaner (Jiangsu Kunshan Ultrasonic Instrument Co., Ltd.); Milli-Q Integral water purifier (Millipore, USA); and high-speed countercurrent chromatograph (Shanghai Tongtian Biotechnology Co., Ltd.).

[0019] A total of 177 dairy cows were randomly selected from 45 commercial farms, and fresh milk (100 mL per cow) was collected at 10:30 am to ensure representativeness of the experiment.

[0020] Step 1: Perform rotary evaporation pre-treatment on fresh milk, add 5 g of white vinegar (the mass of white vinegar is 0.1% of the mass of fresh milk) to 500 g of fresh milk to obtain a milk sample.

[0021] Step 2: Rotary evaporation of the milk sample. Rotary evaporator, parameter settings: Use a rotary evaporator equipped with an explosion-proof boiling ball and a low-temperature condensing circulating water bath (4°C). Use a 500mL round-bottom flask corresponding to 100mL of milk sample. Set the water bath temperature to 40°C, reduce the pressure to 0.09MPa (absolute pressure), and rotate at 80rpm to form a uniform liquid film on the milk, increase the evaporation surface area, and prevent local overheating or foaming. Initially, slowly reduce the pressure to the target pressure and observe the foam formation. Stop heating when the volume has evaporated to 10% of the original volume. Rinse the evaporating flask with 5 mL of methanol, transfer the concentrate to a centrifuge tube, vortex mix, and sonicate for 5 minutes at 25°C. Filter through a 0.22 μm organic filter membrane or perform solid-phase extraction (HLB column) to remove residual proteins and lipids to obtain a clear extract.

[0022] Step 3: Complete the omics analysis of lipids, metabolites, and proteins, and implement vDIA-based UHPLC-Q-Orbitrap HRMS strategy processing. Lipidomics analysis: The clarified extract (500 μL), 960 μL of the extractant and 40 μL of the standard (1 mg / kg) were mixed and sonicated in an ice bath for 10 min. The mixture was then incubated at 4°C at 12,000 × g The supernatant was dried by nitrogen gas. Finally, methanol and isopropanol (1:1, in / out) was redissolved in a mixture (1.0 mL) to obtain a sample for LC-MS analysis, and the primary mass spectrum and secondary mass spectrum of lipids were obtained by injection into the machine.

[0023] Metabolomics analysis: 100 μL of the clarified extract was mixed with 400 μL of metabolite extractant (methanol:acetonitrile, 1:1, v / v ) were mixed and sonicated at 0°C for 10 minutes. Subsequently, the mixture was incubated at -20°C for 1 hour to precipitate proteins and then centrifuged at 12,000 × g for 15 minutes at 4°C to obtain supernatant A, which was used for metabolite data acquisition. The primary and secondary mass spectra of the metabolites were obtained by injection into the instrument.

[0024] Proteomic Analysis: 5 mL of the clarified extract was mixed with 4% sodium dodecyl sulfate, 1 mM dithiothreitol, and 150 mM Tris-HCl (pH 8), heated at 100°C for 15 minutes, and sonicated at 0°C. The suspension was then centrifuged at 16,000 × g for 30 minutes at 25°C to obtain supernatant B, which was used for protein digestion using a filter-paper-assisted sample preparation procedure. The sample was injected into the instrument to obtain primary and secondary mass spectra of the proteins.

[0025] vDIA-based ion acquisition strategy: Metabolites and lipids were separated and eluted using a Dionex UltiMate 3000 HPLC system (Thermo Fisher Scientific, CA, USA) equipped with a C18 column (C18, 100mM × 2.1mM inner diameter, 1.7μm). The Dionex UltiMate 3000 HPLC system is a high-performance liquid chromatography (HPLC) system developed by Dionex, a brand of Thermo Fisher Scientific. Acetonitrile solution (C2H3N / H2O 3:2, in / out , 0.1% formic acid) and aqueous solution (0.1% formic acid) were used as polar mobile phases for metabolite and lipid elution, respectively, while isopropanol solution (ACN / i-PrOH 10:90 in / out, 0.1% formic acid) and methanol (0.1% formic acid) served as nonpolar mobile phases for elution of lipids and metabolites. Mass information was captured using a Q-Orbitrap HRMS (quadrupole-orbitrap high-resolution mass spectrometer, Thermo Fisher Scientific, San Jose, USA) with a spray voltage of ±3.5 kV. Peptide separation was performed on a NanoLC1000 (Thermo Fisher Scientific, San Jose, USA) equipped with a QExactive HF (quadrupole-orbitrap mass spectrometer, Thermo Fisher Scientific, San Jose, USA). Peptides were loaded onto a nano-LC column (100 × 2.1 mM, 5 μm, Thermo Fisher Scientific, Bremen, Germany) at a flow rate of 30 μL / min using acetonitrile (0.1% formic acid) and water (0.1% formic acid) as elution buffers. The mass spectrometer was operated in positive ion mode (+3.5 kV) and a vDIA data acquisition strategy was used, where one MS spectrum was acquired at a resolution of 70,000 full width at half maximum (FWHM) and another MS / MS spectrum was acquired at a resolution of 35,000 FWHM with step collision energies set to 17.5, 35.0, and 52.5 eV.

[0026] Step 4: Use the multiple charge overlap peak extraction algorithm and graph neural network strategy to analyze the group database and realize substance identification. Based on the obtained mass spectrometry data, the fragment tree is used as supplementary data for setting up the protocol, and the structure of the substance is predicted through the primary and secondary fragment mass spectra. Specifically: i) Labeling substance nodes based on the distribution characteristics of MS1 abundance, retention time (RT) and MS2 spectrum; ii) identifying substance nodes and secondary fragments by matching mzCloud, MassBank, Lipidmaps, and KEGG with a mass tolerance of 10 ppm and RT with RSD < 10%, and subsequently distinguishing known substance nodes from unknown substance nodes from mixed labels; iii) Construct the edge connection between the substance node and MS2 according to the relative abundance of the fragments in the MS2 spectrum. Calculate the weight of the edge between the molecule and the secondary fragment. When the calculated result is less than 0, the weight is defined as 0. Take the molecular network construction of benzyl butyrate as an example: the total number of metabolites is 251, involving m / z 121.06479 and m / z The fragment of 123.04406 is the MS2 ion of benzyl butyrate. m / z123.04406 fragment reaction generation m / z The number of molecules involved in the fragmentation reaction of 121.06479 is 1. In the MS2 spectrum, m / z 123.04406 and m / z 121.06479 were 52% and 26% respectively, while m / z 123.04406 and m / z The molecular numbers of 121.06479 fragments are 4 and 6 respectively. After calculation, m / z 121.06479 and m / z The edge weights between 123.04406 and benzyl butyrate are 0.624 and 0.935 respectively, while m / z 121.06479 and m / z The edge weight between 123.04406 is 1.196.

[0027] iv) The molecule nodes identified in step ii) will be used as input data for the graph convolution layer, and the final score is obtained by multiplying the results of each secondary fragment in the molecule.

[0028] After manual verification, a total of 251 metabolites and 3,517 lipids were detected. Methodological validation results demonstrated good linearity for all compounds within the appropriate range, with correlation coefficients (R²) >0.999. The mean intra-day and inter-day recoveries and relative standard deviations for the two analyte levels ranged from 86.9% to 119.1% and 3.4% to 15.1%, respectively, demonstrating the reliability of the data. Matrix effects across all systems ranged from -17.4% to 19.2% for the substances, demonstrating the suitability of the pretreatment method developed in this invention.

[0029] Example 2 The preparatory steps are the same as those in Example 1.

[0030] One hundred cows were randomly selected from 30 commercial farms. Fresh milk (100 mL per cow) was collected at 5:30 PM to ensure representativeness of the experiment. The concentration of bisphenol A in the fresh milk was determined using ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS). Fresh milk with a deviation of ≤10% was pooled.

[0031] Step 1: Perform rotary evaporation pretreatment on fresh milk, add 6 g of white vinegar with a mass concentration of 0.1% to 500 g of fresh milk to obtain a milk sample.

[0032] Step 2: Rotary evaporation of the milk sample. Rotary evaporator and parameter settings: Use a rotary evaporator equipped with an explosion-proof boiling ball and a low-temperature condensing circulating water bath (4°C). Use a 400mL round-bottom flask for every 100mL of milk sample. Set the water bath temperature to 45°C, reduce the pressure to 0.08MPa (absolute pressure), and rotate at 90rpm to form a uniform liquid film on the milk, increasing the evaporation surface area while preventing local overheating or foaming. Initially, slowly reduce the pressure to the target pressure and observe the foam formation. Stop heating when the volume has evaporated to 10% of the original volume. Rinse the inner wall of the evaporating flask with 5 mL of methanol, transfer the concentrate to a centrifuge tube, vortex mix, and sonicate for 5 minutes at 25°C. Filter through a 0.22 μm organic filter membrane to obtain a clear extract.

[0033] Step 3: Complete the omics analysis of lipids, metabolites, and proteins, and implement vDIA-based UHPLC-Q-Orbitrap HRMS strategy processing. Lipidomics analysis: The lipids in the milk of the test subjects were enriched and purified. The clarified extract (500 μL) was mixed with 960 μL of the extractant and 40 μL of the standard (1 mg / kg), and then sonicated in an ice bath for 10 minutes. The mixture was then heated at 4°C at 12,000 × g The supernatant was dried by nitrogen gas. Finally, methanol and isopropanol (1:1, in / out ) was redissolved to obtain a mixed solution sample for LC-MS analysis; after obtaining the mixed solution sample, the primary mass spectrum and secondary mass spectrum of the lipids were obtained by injection into the machine.

[0034] Metabolomics analysis: 100 μL of the clarified extract was mixed with 400 μL of metabolite extractant (methanol:acetonitrile, 1:1, v / v ) were mixed and sonicated at 2°C for 10 minutes. Subsequently, the mixture was incubated at -20°C for 1 hour to precipitate proteins and then centrifuged at 12,000 × g at 4°C for 15 minutes to obtain supernatant A. After obtaining supernatant A, the primary and secondary mass spectra of the metabolites were obtained.

[0035] Proteomic Analysis: 5 mL of the clarified extract was mixed with 4% sodium dodecyl sulfate, 1 mM dithiothreitol, and 150 mM Tris-HCl (pH 8), heated at 110°C for 15 minutes, and sonicated at 4°C. The suspension was then centrifuged at 16,000 × g for 30 minutes at 20°C to obtain supernatant B. This was then injected into a centrifuge to obtain mass spectra of the proteins.

[0036] vDIA-based ion acquisition strategy: Metabolites and lipids were separated and eluted using a Dionex UltiMate 3000 HPLC system equipped with a C18 column (C18, 100mM × 2.1mM inner diameter, 1.7μm). The Dionex UltiMate 3000 HPLC system is a high-performance liquid chromatography (HPLC) system developed by Dionex, a brand of Thermo Fisher Scientific. Acetonitrile solution (C2H3N / H2O 3:2, in / out , 0.1% formic acid) and aqueous solution (0.1% formic acid) were used as polar mobile phases for metabolite and lipid elution, respectively, while isopropanol solution (ACN / i-PrOH 10:90 in / out Nonpolar mobile phases (0.1% formic acid, 0.1% formic acid) and methanol (0.1% formic acid) were used for elution of lipids and metabolites. Mass information was captured using a Q-Orbitrap HRMS (quadrupole-orbitrap high-resolution mass spectrometer, Thermo Fisher Scientific, San Jose, USA) with a spray voltage of ±3.5 kV. Peptide separation was performed on a NanoLC1000 (Thermo Fisher Scientific, San Jose, USA) equipped with a Q ExactiveHF (quadrupole-orbitrap mass spectrometer, Thermo Fisher Scientific, San Jose, USA). Peptides were loaded onto a nano-LC column (100 × 2.1 mM, 5 μm, Thermo Fisher Scientific, Bremen, Germany) at a flow rate of 30 μL / min using acetonitrile (0.1% formic acid) and water (0.1% formic acid) as elution buffers. The mass spectrometer was operated in positive ion mode (+3.5 kV) and a vDIA data acquisition strategy was used, whereby MS spectra were acquired at a resolution of 70,000 full width at half maximum (FWHM) and MS / MS spectra were acquired at a resolution of 35,000 FWHM with step collision energies set to 17.5, 35.0, and 52.5 eV.

[0037] Step 4: Use the multiple charge overlap peak extraction algorithm and graph neural network strategy to analyze the group database and realize substance identification. Based on the obtained mass spectrometry data, the fragment tree is used as supplementary data for setting up the protocol, and the structure of the substance is predicted through the primary and secondary fragment mass spectra. Specifically: i) Labeling substance nodes based on the distribution characteristics of MS1 abundance, retention time (RT) and MS2 spectra; ii) identifying substance nodes and secondary fragments by matching mzCloud, MassBank, Lipidmaps, and KEGG with a mass tolerance of 10 ppm and RT with RSD < 10%, and subsequently distinguishing known substance nodes from unknown substance nodes from mixed labels; iii) Construct edge connections between substance nodes and MS2 based on the relative abundance of fragments in the MS2 spectrum. Calculate the weight of the edge between the molecule and the secondary fragment. When the calculated result is less than 0, the weight is defined as 0. For 4-isobutylbenzoic acid, m / z 55.05481 [C4H6+H] + , m / z 57.06988 [C4H8+H] + , m / z 163.11174 [C 11 H 14 O+H] + , m / z 121.06479 [C8H8O+H] + , and m / z 177.09015 [C 11 H 12 O2+H] + etc. fragments; For butylbenzoic acid, m / z 161.09690 [C 11 H 12 O+H] + , m / z 91.05423 [C7H6+H] + , m / z 105.03349 [C7H4O+H] + , m / z 123.04406 [C7H6O2+H] + , m / z 133.06479 [C9H8O+H] + , and m / z 151.07536 [C9H 10 O2+H] + etc. fragments.

[0038] iv) The molecule nodes identified in step ii) will be used as input data for the graph convolution layer, and the final score is obtained by multiplying the results of each secondary fragment in the molecule.

[0039] After manual verification, a total of 252 metabolites and 3,530 lipids were detected. Methodological validation results demonstrated good linearity for all compounds within the appropriate range, with correlation coefficients (R²) >0.999. The mean intra-day and inter-day recoveries and relative standard deviations for the two analytes ranged from 88.6% to 112.3% and 3.5% to 13.5%, respectively, demonstrating the reliability of the data. Matrix effects across all systems ranged from -14.8% to 18.5% for the substances, demonstrating the suitability of the pretreatment method developed in this invention.

[0040] Example 3 The preparatory steps are the same as those in Example 1.

[0041] 100 cows were randomly selected from 30 commercial farms. Fresh milk (100 mL per cow) was collected at 12:30 p.m. to ensure representativeness of the experiment. The concentration of bisphenol A in the fresh milk was determined using ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS). Fresh milk with a deviation of ≤10% was pooled.

[0042] Step 1: Perform rotary evaporation pretreatment on fresh milk, add 7g of white vinegar to 500g of fresh milk to obtain a milk sample.

[0043] Step 2: Rotary evaporation of the milk sample. Rotary evaporator and parameter settings: Use a rotary evaporator equipped with an explosion-proof boiling ball and a low-temperature condensing circulating water bath (4°C). Use a 500mL round-bottom flask for 100mL of milk sample. Set the water bath temperature to 45°C, reduce the pressure to 0.08MPa (absolute pressure), and rotate at 90rpm to form a uniform liquid film on the milk, increase the evaporation surface area, and prevent local overheating or foaming. Initially, slowly reduce the pressure to the target pressure and observe the foam formation. Stop heating when the volume has evaporated to 10% of the original volume. Rinse the inner wall of the evaporating flask with 5 mL of acetonitrile (containing 0.1% formic acid), transfer the concentrate to a centrifuge tube, vortex mix, and sonicate for 5 minutes at 25°C. Filter through a 0.22 μm organic filter membrane to obtain a clear extract.

[0044] Step 3: Complete the omics analysis of lipids, metabolites, and proteins, and implement vDIA-based UHPLC-Q-Orbitrap HRMS strategy processing. Lipidomics analysis: The lipids in the milk of the test subjects were enriched and purified. The clarified extract (500 μL) was mixed with 960 μL of the extractant and 40 μL of the standard (1 mg / kg), and then sonicated in an ice bath for 10 minutes. The mixture was then heated at 2°C at 12,000× g The supernatant was dried by nitrogen gas. Finally, methanol and isopropanol (1:1, in / out ) was redissolved to obtain a mixed solution sample for LC-MS analysis; after obtaining the mixed solution sample, the primary mass spectrum and secondary mass spectrum of the lipids were obtained by injection into the machine.

[0045] Metabolomics analysis: 100 μL of the clarified extract was mixed with 400 μL of metabolite extractant (methanol:acetonitrile, 1:1, v / v ) were mixed and sonicated at 2°C for 10 minutes. Subsequently, the mixture was incubated at -20°C for 1 hour to precipitate proteins and then centrifuged at 12,000 × g at 2°C for 15 minutes to obtain supernatant A. After obtaining supernatant A, the primary and secondary mass spectra of the metabolites were obtained.

[0046] Proteomic Analysis: 5 mL of the clarified extract was mixed with 4% sodium dodecyl sulfate, 1 mM dithiothreitol, and 150 mM Tris-HCl (pH 8), heated at 110°C for 15 minutes, and sonicated at 4°C. The suspension was then centrifuged at 16,000 × g for 30 minutes at 20°C to obtain supernatant B. This was then injected into a centrifuge to obtain mass spectra of the proteins.

[0047] The vDIA-based ion acquisition strategy is the same as that in Example 1.

[0048] Step 4: Use the multiple charge overlap peak extraction algorithm and graph neural network strategy to analyze the group database and realize substance identification. Based on the obtained mass spectrometry data, the fragment tree is used as supplementary data for setting up the protocol, and the structure of the substance is predicted through the primary and secondary fragment mass spectra. Specifically: i) Labeling substance nodes based on the distribution characteristics of MS1 abundance, retention time (RT) and MS2 spectra; ii) identifying substance nodes and secondary fragments by matching mzCloud, MassBank, Lipidmaps, and KEGG with a mass tolerance of 10 ppm and RT with RSD < 10%, and subsequently distinguishing known substance nodes from unknown substance nodes from mixed labels; iii) Construct edge connections between substance nodes and MS2 based on the relative abundance of fragments in the MS2 spectrum. Calculate the weight of the edge between the molecule and the secondary fragment. When the calculated result is less than 0, the weight is defined as 0. For 4-isobutylbenzoic acid, m / z 55.05481 [C4H6+H] + , m / z57.06988 [C4H8+H] + , m / z 163.11174 [C 11 H 14 O+H] + , m / z 121.06479 [C8H8O+H] + , and m / z 177.09015 [C 11 H 12 O2+H] + etc. fragments; For butylbenzoic acid, m / z 161.09690 [C 11 H 12 O+H] + , m / z 91.05423 [C7H6+H] + , m / z 105.03349 [C7H4O+H] + , m / z 123.04406 [C7H6O2+H] + , m / z 133.06479 [C9H8O+H] + , and m / z 151.07536 [C9H 10 O2+H] + etc. fragments.

[0049] iv) The molecule nodes identified in step ii) will be used as input data for the graph convolution layer, and the final score is obtained by multiplying the results of each secondary fragment in the molecule.

[0050] After manual verification, a total of 247 metabolites and 3,489 lipids were detected. Methodological validation results demonstrated good linearity for all compounds within the appropriate range, with correlation coefficients (R²) >0.999. The mean intra-day and inter-day recoveries and relative standard deviations for the two analyte levels ranged from 86.7% to 118.4% and 4.2% to 11.6%, respectively, demonstrating the reliability of the data. Matrix effects across all systems varied from -16.5% to 17.6% for the substances, demonstrating the suitability of the pretreatment method developed in this invention.

[0051] Example 4 The preparatory steps are the same as those in Example 1.

[0052] A total of 110 dairy cows were randomly selected from 31 commercial farms. Fresh milk (100 mL per cow) was collected at 11:00 a.m. to ensure representativeness of the experiment. The concentration of bisphenol A in the fresh milk was determined using ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS). Fresh milk with a deviation of ≤10% was pooled.

[0053] Step 1: Perform rotary evaporation pretreatment on fresh milk, add 9 g of white vinegar to 500 g of fresh milk to obtain a milk sample.

[0054] Step 2: Rotary evaporation of milk samples. Rotary evaporator and parameter settings: Use a rotary evaporator equipped with an explosion-proof boiling ball and a low-temperature condensing circulating water bath (4°C). Use a 500mL round-bottom flask for 100mL of milk sample. Set the water bath temperature to 45°C, reduce the pressure to 0.1MPa (absolute pressure), and rotate at 100rpm to form a uniform liquid film on the milk, increase the evaporation surface area, and prevent local overheating or foaming. Initially, slowly reduce the pressure to the target pressure and observe the foam formation. Stop heating when the volume has evaporated to 10% of the original volume. Rinse the evaporating flask with 5 mL of methanol, transfer the concentrate to a centrifuge tube, vortex mix, and sonicate for 5 minutes at 25°C. Filter through a 0.22 μm organic filter membrane to obtain a clear extract for mass spectrometry analysis.

[0055] Step 3: Complete the omics analysis of lipids, metabolites, and proteins, and implement vDIA-based UHPLC-Q-Orbitrap HRMS strategy processing. Lipidomics analysis: The lipids in the milk of the test subjects were enriched and purified. The clarified extract (500 μL) was mixed with 960 μL of the extractant and 40 μL of the standard (1 mg / kg), and then sonicated in an ice bath for 10 minutes. The mixture was then heated at 3°C ​​at 10,000 × g The supernatant was dried by nitrogen gas. Finally, methanol and isopropanol (1:1, in / out ) was redissolved to obtain a mixed solution sample for LC-MS analysis; after obtaining the mixed solution sample, the primary mass spectrum and secondary mass spectrum of the lipids were obtained by injection into the machine.

[0056] Metabolomics analysis: 100 μL of the clarified extract was mixed with 400 μL of metabolite extractant (methanol:acetonitrile, 1:1, v / v) were mixed and sonicated at 2°C for 10 minutes. Subsequently, the mixture was incubated at -20°C for 1 hour to precipitate proteins and then centrifuged at 12,000 × g at 2°C for 15 minutes to obtain supernatant A. After obtaining supernatant A, the primary and secondary mass spectra of the metabolites were obtained.

[0057] Proteomic Analysis: 5 mL of the clarified extract was mixed with 4% sodium dodecyl sulfate, 1 mM dithiothreitol, and 150 mM Tris-HCl (pH 8), heated at 110°C for 15 minutes, and sonicated at 4°C. The suspension was then centrifuged at 16,000 × g for 30 minutes at 20°C to obtain supernatant B. This was then injected into a centrifuge to obtain mass spectra of the proteins.

[0058] The vDIA-based ion acquisition strategy is the same as that in Example 1.

[0059] Step 4: Use the multiple charge overlap peak extraction algorithm and graph neural network strategy to analyze the group database and realize substance identification. Based on the obtained mass spectrometry data, the fragment tree is used as supplementary data for setting the protocol, and the substance structure is predicted through the primary fragment and secondary fragment mass spectra, which is the same as Example 1.

[0060] After manual verification, a total of 248 metabolites and 3,568 lipids were detected. Methodological validation results demonstrated good linearity for all compounds within the appropriate range, with correlation coefficients (R²) >0.999. The mean intra-day and inter-day recoveries and relative standard deviations for the two analyte levels ranged from 88.5% to 121.5% and 4.5% to 12.8%, respectively, demonstrating the reliability of the data. Matrix effects across all systems ranged from -16.8% to 18.5% for the substances, demonstrating the suitability of the pretreatment method developed in this invention.

[0061] Example 5 The preparatory steps are the same as those in Example 1.

[0062] A total of 110 dairy cows were randomly selected from 29 commercial farms. Fresh milk (100 mL per cow) was collected at 11:00 a.m. to ensure representativeness of the experiment. The concentration of bisphenol A in the fresh milk was determined using ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS). Fresh milk with a deviation of 10% or less was pooled.

[0063] Step 1: Perform rotary evaporation pretreatment on fresh milk, add 8 g of white vinegar to 500 g of fresh milk to obtain a milk sample.

[0064] Step 2: Rotary evaporation of the milk sample. Rotary evaporator and parameter settings: Use a rotary evaporator equipped with an explosion-proof boiling bulb and a low-temperature condensing circulating water bath (4°C). The evaporating flask capacity should be no less than three times the volume of the milk sample. Set the water bath temperature to 44°C, reduce the pressure to 0.09 MPa (absolute pressure), and rotate at 95 rpm to form a uniform liquid film on the milk, increase the evaporation surface area, and prevent local overheating or foaming. Initially, slowly reduce the pressure to the target pressure and observe the foam formation. Stop heating when the milk evaporates to 10% of the original volume. Rinse the inner wall of the evaporating flask with 5 mL of acetonitrile (containing 0.1% formic acid), transfer the concentrate to a centrifuge tube, vortex mix, and sonicate for 5 minutes at 25°C. Filter through a 0.22 μm organic filter membrane. Further solid-phase extraction (e.g., HLB column) can be performed to remove residual proteins and lipids to obtain a clear extract.

[0065] Step 3: Complete the omics analysis of lipids, metabolites, and proteins, and implement vDIA-based UHPLC-Q-Orbitrap HRMS strategy processing. Lipidomics analysis: The lipids in the milk of the test subjects were enriched and purified. The clarified extract (500 μL) was mixed with 960 μL of the extractant and 40 μL of the standard (1 mg / kg), and then sonicated in an ice bath for 10 minutes. The mixture was then heated at 2°C at 12,000 × g The supernatant was dried by nitrogen gas. Finally, methanol and isopropanol (1:1, in / out ) was redissolved to obtain a mixed solution sample for LC-MS analysis; after obtaining the mixed solution sample, the primary mass spectrum and secondary mass spectrum of the lipids were obtained by injection into the machine.

[0066] Metabolomics analysis: 100 μL of the clarified extract was mixed with 400 μL of metabolite extractant (methanol:acetonitrile, 1:1, v / v ) were mixed and sonicated at 2°C for 10 minutes. Subsequently, the mixture was incubated at -20°C for 1 hour to precipitate proteins and then centrifuged at 12,000 × g at 2°C for 15 minutes to obtain supernatant A. After obtaining supernatant A, the primary and secondary mass spectra of the metabolites were obtained.

[0067] Proteomic Analysis: 5 mL of the clarified extract was mixed with 4% sodium dodecyl sulfate, 1 mM dithiothreitol, and 150 mM Tris-HCl (pH 8), heated at 110°C for 15 minutes, and sonicated at 4°C. The suspension was then centrifuged at 10,000 × g for 30 minutes at 20°C to obtain supernatant B. This was then injected into a mass spectrometer to obtain mass spectra of the proteins.

[0068] The vDIA-based ion acquisition strategy is the same as that in Example 1.

[0069] Step 4: Use the multiple charge overlap peak extraction algorithm and graph neural network strategy to analyze the group database and realize substance identification. Based on the obtained mass spectrometry data, the fragment tree is used as supplementary data for setting the protocol, and the substance structure is predicted through the primary fragment and secondary fragment mass spectra, which is the same as Example 2.

[0070] After manual verification, a total of 254 metabolites and 3,875 lipids were detected. Methodological validation results demonstrated good linearity for all compounds within the appropriate range, with correlation coefficients (R²) >0.999. The mean intra-day and inter-day recoveries and relative standard deviations for the two analyte levels ranged from 83.5% to 132.4% and 3.9% to 13.5%, respectively, demonstrating the reliability of the data. Matrix effects across all systems ranged from -14.5% to 19.3% for the substances, demonstrating the suitability of the pretreatment method developed in this invention.

[0071] Example 6 The preparatory steps are the same as those in Example 1.

[0072] A total of 116 dairy cows were randomly selected from 34 commercial farms. Fresh milk (100 mL per cow) was collected at 3:00 PM to ensure representativeness of the experiment. The concentration of bisphenol A in the fresh milk was determined using ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS). Fresh milk with a deviation of ≤10% was pooled.

[0073] Step 1: Perform rotary evaporation pretreatment on fresh milk, add 10 g of citric acid to 500 g of fresh milk to obtain a milk sample.

[0074] Step 2: Rotary evaporation of milk samples. Rotary evaporator and parameter settings: Use a rotary evaporator equipped with an explosion-proof boiling ball and a low-temperature condensing circulating water bath (4°C). Use a 400mL round-bottom flask for 100mL of milk sample. Set the water bath temperature to 45°C, reduce the pressure to 0.09MPa (absolute pressure), and rotate at 95rpm to form a uniform liquid film on the milk, increase the evaporation surface area, and prevent local overheating or foaming. Initially, slowly reduce the pressure to the target pressure and observe the foam formation. Stop heating when the volume has evaporated to 10% of the original volume. Rinse the inner wall of the evaporating flask with 5 mL of methanol, transfer the concentrate to a centrifuge tube, vortex mix, and sonicate for 5 minutes at 25°C. Filter through a 0.22 μm organic filter membrane to obtain a clear extract.

[0075] Step 3: Complete the omics analysis of lipids, metabolites, and proteins, and implement vDIA-based UHPLC-Q-Orbitrap HRMS strategy processing. Lipidomics analysis: The lipids in the milk of the test subjects were enriched and purified. The clarified extract (500 μL) was mixed with 960 μL of the extractant and 40 μL of the standard (1 mg / kg), and then sonicated in an ice bath for 10 minutes. The mixture was then heated at 2°C at 10,000 × g The supernatant was dried by nitrogen gas. Finally, methanol and isopropanol (1:1, in / out ) was redissolved to obtain a mixed solution sample for LC-MS analysis; after obtaining the mixed solution sample, the primary mass spectrum and secondary mass spectrum of the lipids were obtained by injection into the machine.

[0076] Metabolomics analysis: 100 μL of the clarified extract was mixed with 400 μL of metabolite extractant (methanol:acetonitrile, 1:1, v / v ) were mixed and sonicated at 4°C for 10 minutes. The mixture was then incubated at -20°C for 1 hour to precipitate proteins and then centrifuged at 10,000 × g at 4°C for 15 minutes to obtain supernatant A. After obtaining supernatant A, the primary and secondary mass spectra of the metabolites were obtained.

[0077] Proteomic Analysis: 5 mL of the clarified extract was mixed with 4% sodium dodecyl sulfate, 1 mM dithiothreitol, and 150 mM Tris-HCl (pH 8), heated at 110°C for 15 minutes, and sonicated at 4°C. The suspension was then centrifuged at 10,000 × g for 30 minutes at 20°C to obtain supernatant B. This was then injected into a mass spectrometer to obtain mass spectra of the proteins.

[0078] The vDIA-based ion acquisition strategy is the same as that in Example 1.

[0079] Step 4: Use the multiple charge overlap peak extraction algorithm and graph neural network strategy to analyze the group database and realize substance identification. Based on the obtained mass spectrometry data, the fragment tree is used as supplementary data for setting the protocol, and the substance structure is predicted through the primary fragment and secondary fragment mass spectra, which is the same as Example 3.

[0080] After manual verification, a total of 255 metabolites and 3,726 lipids were detected. Methodological validation results demonstrated good linearity for all compounds within the appropriate range, with correlation coefficients (R²) >0.999. The mean intra-day and inter-day recoveries and relative standard deviations for the two analyte levels ranged from 85.2% to 121.9% and 4.1% to 12.9%, respectively, demonstrating the reliability of the data. Matrix effects across all systems ranged from -14.1% to 17.8% for the substances, demonstrating the suitability of the pretreatment method developed in this invention.

Claims

1. A mass spectrometry method for detecting bisphenol A in milk by rotary evaporation pretreatment and multi-charge peak analysis, characterized in that: Follow these steps to implement: Step 1, performing rotary evaporation pre-treatment on fresh milk; Step 2, performing rotary evaporation on the milk sample; Step 3: Complete the omics analysis of lipids, metabolites, and proteins, and implement vDIA-based UHPLC-Q-Orbitrap HRMS strategy processing; Step 4: Use the multiple charge overlap peak extraction algorithm and graph neural network strategy to parse the group database and realize substance identification.

2. The mass spectrometry method for detecting bisphenol A in milk by rotary evaporation pre-treatment and multi-charge peak analysis according to claim 1, characterized in that: In step 1, the specific process is: Fresh milk was collected, and 0.1%-0.5% of white vinegar or citric acid was added to the fresh milk according to the mass percentage to prepare a milk sample.

3. The mass spectrometry method for detecting bisphenol A in milk by rotary evaporation pre-treatment and multi-charge peak analysis according to claim 1, characterized in that: In step 2, the specific process is: 2.1) Configure the rotary evaporator and set the parameters: The rotary evaporator should be equipped with an explosion-proof boiling ball and a low-temperature condensing circulating water bath, and the capacity of the evaporating flask used should be no less than 3 times the volume of the milk sample; 2.2) Concentration: Slowly reduce the pressure to the target pressure. When the evaporation reaches 10%-20% of the original volume, stop heating immediately to obtain the concentrate. 2.3) Post-processing and transfer: Rinse the inner wall of the evaporating flask with a small amount of methanol or acetonitrile, transfer the concentrate to a centrifuge tube, vortex mix, and sonicate to dissolve the extract. Filter through an organic filter membrane to obtain a clarified extract.

4. The method for mass spectrometry detection of bisphenol A in milk by rotary evaporation pre-treatment and multi-charge peak analysis according to claim 3, characterized in that: In step 2.1), the specific process parameters are: water bath temperature of 40°C-45°C, reduced pressure to 0.08-0.1 MPa, and rotation speed of 80-100 rpm.

5. The mass spectrometry method for detecting bisphenol A in milk by rotary evaporation pre-treatment and multi-charge peak analysis according to claim 3, characterized in that: In step 2.3), the specific process is: rinse the inner wall of the evaporating flask with methanol or acetonitrile; the process parameters for ultrasonic-assisted dissolution are: 5-10 min, 20-25°C.

6. The method for mass spectrometry detection of bisphenol A in milk by rotary evaporation pre-treatment and multi-charge peak analysis according to claim 1, characterized in that: In step 3, the omics analysis is as follows: 3.1) Lipidomics analysis: 500 μL of clarified extract and 960 μL of extraction agent were mixed at room temperature and then sonicated in an ice bath. The mixture was then centrifuged at 10,000-12,000 × g at 0-4°C, and the supernatant was blown dry with nitrogen. Finally, the mixture was redissolved in a mixture of methanol and isopropanol to obtain a mixed solution sample, which was injected into the instrument to obtain the primary and secondary mass spectra of the lipids. 3.2) Metabolomics Analysis: Mix 100 μL of clarified extract with 400 μL of metabolite extraction reagent and sonicate at 0-4°C. The mixture was then incubated and centrifuged at 12,000 × g at 0-4°C to obtain supernatant A, which was injected into the instrument to obtain the primary mass spectra and secondary mass spectra of the metabolites. 3.3) Proteomic Analysis: Mix 5 mL of the clarified extract with 4% sodium dodecyl sulfate, 1 mM dithiothreitol, and 150 mM Tris-HCl, heat at 90°C-100°C, and sonicate at 0°C-4°C to obtain a suspension. Subsequently, the suspension was centrifuged at 20-25° C. to obtain supernatant B, which was injected into a mass spectrometer to obtain primary mass spectra and secondary mass spectra of the protein.

7. The mass spectrometry method for detecting bisphenol A in milk by rotary evaporation pre-treatment and multi-charge peak analysis according to claim 1, characterized in that: In step 3, the vDIA-based ion acquisition strategy is as follows: Separation and elution of metabolites and lipids were performed using a Dionex UltiMate 3000 HPLC system. The polar mobile phase for lipids was aqueous solution, and the nonpolar mobile phase was isopropanol solution; the polar mobile phase for metabolites was acetonitrile solution, and the nonpolar mobile phase was methanol solution. Q-Orbitrap HRMS was used to capture mass information. Peptides were separated in a NanoLC1000 equipped with Q Exactive HF, using acetonitrile and aqueous solution as elution buffers to load the peptides onto the Nano-LC column. The mass spectrometer was operated in positive ion mode, and the vDIA data acquisition strategy was adopted. The vDIA data acquisition strategy included two methods: one was to acquire MS spectra at a resolution of 70,000 full-width half-maximum (FWHM); The other is that the MS / MS spectrum was acquired at a resolution of 35,000 FWHM.

8. The mass spectrometry method for detecting bisphenol A in milk by rotary evaporation pre-treatment and multi-charge peak analysis according to claim 1, characterized in that: In step 4, the specific process is: i) Labeling substance nodes based on the distribution characteristics of MS1 abundance, retention time RT and MS2 spectrum; ii) identifying substance nodes and secondary fragments by matching mzCloud, MassBank, Lipidmaps, and KEGG with a mass tolerance of 10 ppm and RT with RSD < 10%, and subsequently distinguishing known substance nodes from unknown substance nodes from mixed labels; iii) construct edge connections between material nodes and MS2 based on the relative abundance of fragments in the MS2 spectrum; Calculate the weight of the edge between the molecule and the secondary fragment. If the calculated result is less than 0, the weight is defined as 0; iv) The molecule nodes identified in step ii) will be used as input data for the graph convolution layer, and the final score is obtained by multiplying the results of each secondary fragment in the molecule.