Free amino acid multi-index combination for identifying wild silver carp, distinguishing method and application

Abstract

The invention discloses a free amino acid multi-index combination for identifying wild silver carps, a distinguishing method and application. The free amino acid multi-index combination comprises at least four of phosphoserine, taurine, threonine, sarcosine, beta-aminoisobutyric acid and hydroxyproline. The distinguishing method comprises the following steps: measuring the content of amino acids in a specific combination in a to-be-detected sample, inputting the content into a preset logistic regression model to calculate a wild probability value, and comparing the wild probability value with a distinguishing threshold value optimized and determined through an ROC curve so as to judge a source. According to the technical scheme, the specific biomarker for discriminating the source of the silver carp is defined for the first time, and the problem that a traditional experience discriminating method is high in subjectivity is solved. Through construction of the quantitative model, rapid, objective and high-precision discrimination is realized, and the discrimination accuracy of the optimal combination model can reach 97.8%. The method provides reliable technical support for identifying illegal fished catches in fishery administration law enforcement in the Yangtze River basin, and has important significance for protecting wild silver carp resources.

Description

Technical Field

[0001] This application relates to the field of silver carp identification or confirmation technology, specifically to the combination of multiple free amino acid indicators, identification methods and applications for identifying wild silver carp. Background Technology

[0002] Silver carp is an important aquaculture species in my country, mainly distributed in the Yangtze River basin. Due to water pollution and overfishing, the biodiversity of fish in the Yangtze River basin is declining. Therefore, since 2020, my country has fully implemented the "Ten-Year Fishing Ban on the Yangtze River," severely cracking down on illegal fishing and the sale of Yangtze River catches. However, due to the current difficulty in distinguishing between wild and farmed aquatic products in the Yangtze River basin, it is difficult to meet the requirements of the "Yangtze River Fishing Ban" policy. Therefore, developing or confirming technologies for wild and farmed silver carp, identifying biomarkers for wild and farmed silver carp, and promoting their application are reliable ways to support the "Yangtze River Fishing Ban" policy. Summary of the Invention

[0003] Free amino acids exist as single molecules and can directly participate in physiological and biochemical reactions. Their content is relatively low in some samples, but their variety is richer than that of hydrolyzed amino acids. As active components directly involved in metabolism within organisms, changes in the specific types and contents of free amino acids may better reflect the immediate physiological state and metabolic differences of silver carp in wild and farmed environments. Therefore, this application addresses the problem of similar biological traits and lack of differentiation technology between wild and farmed silver carp. For the first time, muscle samples from different wild and farmed silver carp populations were collected, and free amino acid profiles were analyzed using an automated amino acid analyzer. The commonalities, differences, and changes in free amino acids between wild and farmed silver carp were explored, and biomarkers of free amino acids were screened: phosphoserine (CAS: 407-41-0), taurine (CAS: 107-35-7), threonine (CAS: 72-19-5), sarcosine (CAS: 107-97-1), β-aminoisobutyric acid (CAS: 10569-72-9), and hydroxyproline (CAS: 51-35-4). Statistical and logistic regression analyses were performed using the results from an automated amino acid analyzer to obtain a logistic regression model. This model accurately distinguishes between wild and farmed samples, demonstrating good sensitivity and specificity. Furthermore, the free amino acid biomarkers and their applications provided in this application can offer scientific evidence and technical support to relevant law enforcement departments, strengthen the supervision and crackdown on illegal fishing in the Yangtze River Basin, reduce overfishing of wild resources, and thus provide a technical foundation for better management and protection of silver carp resources.

[0004] Therefore, this application provides the following technical solution:

[0005] In a first aspect, this application provides a combination of free amino acid biomarkers for distinguishing between wild and farmed silver carp, said biomarker combination comprising any one of the following (a)-(d):

[0006] (a) Phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, hydroxyproline;

[0007] (b) Phosphoseserine, taurine, threonine, β-aminoisobutyric acid, hydroxyproline;

[0008] (c) Taurine, threonine, β-aminoisobutyric acid, hydroxyproline;

[0009] (d) Phosphoseserine, taurine, threonine, β-aminoisobutyric acid;

[0010] A discriminant model can be constructed using the content information of any of the above combinations of biomarkers.

[0011] Secondly, this application provides a method for distinguishing between wild and farmed silver carp, comprising the following steps:

[0012] S1. Obtain muscle samples from the silver carp to be tested;

[0013] S2. Determine the content of all amino acids in the combination of biomarkers as described in the first aspect in the sample;

[0014] S3. Input the measured content into the preset logistic regression model to calculate the probability that the sample is a wild silver carp;

[0015] S4. Compare the wild probability value with a preset discrimination threshold. If it is greater than the discrimination threshold, it is determined to be a wild silver carp; if it is less than the discrimination threshold, it is determined to be a farmed silver carp.

[0016] In some embodiments, the preset discrimination threshold is determined by the following method: obtaining muscle samples from wild and farmed silver carp of known origin as a training set; measuring the content of the biomarker combination in the training set samples; performing binary logistic regression analysis with the content as the independent variable and the sample origin as the dependent variable to obtain a logistic regression model; calculating the wild probability value of each sample in the training set based on the logistic regression model; plotting ROC curves, calculating the Youden index corresponding to each wild probability value, and determining the wild probability value corresponding to the highest Youden index as the discrimination threshold.

[0017] In some preferred embodiments, when the biomarker combination is combination (a) in the first aspect, the logistic regression model is the first logistic regression model, calculated as follows: X = -1.802×A - 0.160×B - 0.362×C + 0.258×D + 1.823×E + 0.531×F + 129.612; the wild probability is the first wild probability, with a value of 1 / (1+e －X ); where A, B, C, D, E, F represent the contents of phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline in the sample, respectively; the discrimination threshold is 0.415.

[0018] In some preferred embodiments, when the biomarker combination is combination (b) in the first aspect, the logistic regression model is a second logistic regression model, calculated as follows: X = -0.318×A - 0.029×B - 0.059×C + 0.691×D + 0.117×E + 24.246; the wild probability is the second wild probability, with a value of 1 / (1+e) －X ); where A, B, C, D, and E are respectively phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline; the discrimination threshold is 0.529.

[0019] In some preferred embodiments, when the biomarker combination is combination (c) in the first aspect, the logistic regression model is a third logistic regression model, calculated as follows: X = -0.017×A - 0.045×B + 0.475×C + 0.089×D + 11.437; the wild probability is the third wild probability, with a value of 1 / (1+e －X ); where A, B, C, and D are taurine, threonine, β-aminoisobutyric acid, and hydroxyproline, respectively; the discrimination threshold is 0.441.

[0020] In some preferred embodiments, when the biomarker combination is combination (d) in the first aspect, the logistic regression model is the fourth logistic regression model, calculated as follows: X = -0.307×A - 0.019×B - 0.027×C + 0.404×D + 19.318; the wild probability is the fourth wild probability, with a value of 1 / (1+e －X ); where A, B, C, and D are phosphoserine, taurine, threonine, and β-aminoisobutyric acid, respectively; the discrimination threshold is 0.645.

[0021] In some preferred embodiments, the determination in step S2 is performed using an automated amino acid analyzer, wherein the chromatographic separation uses a Li-type cation exchange column with lithium salt buffer as the mobile phase; the separation temperature is kept constant at 57 °C, and a derivatization unit is connected after the column, with ninhydrin solution as the derivatizing agent, and the temperature of the reaction zone is maintained at 135 °C.

[0022] Thirdly, this application provides the use of the combination of biomarkers described in the first aspect and / or reagents for detecting them in distinguishing between wild and farmed silver carp.

[0023] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is used to implement steps S3 and S4 of the method described in the second aspect.

[0024] Compared with the prior art, this application has at least the following advantages:

[0025] 1. This application is the first to screen and validate a combination of free amino acids, with phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline as the core, as a discriminant marker. Based on the metabolic differences between wild and farmed silver carp, this combination provides a direct chemical basis for discrimination, overcoming the inherent subjectivity of traditional morphological identification.

[0026] 2. The model provided in this application has high accuracy and reliable discrimination results. By combining the content of the above-mentioned biomarkers with a logistic regression model and optimizing the threshold using ROC curves, a quantifiable and high-precision discrimination method is established. The AUC value of the discrimination model can reach above 0.990, with an overall accuracy exceeding 97.8%, exhibiting excellent sensitivity and specificity.

[0027] 3. The method for distinguishing between wild and farmed silver carp provided in this application can achieve rapid and accurate traceability of silver carp samples. It can be directly applied to the front line of market supervision and fishery law enforcement, providing a much-needed technical tool for implementing the "Ten-Year Fishing Ban in the Yangtze River" policy, combating illegal fishing, and protecting wild resources. Attached Figure Description

[0028] Figure 1 The confusion matrix test results of the free amino acid biomarker composition (phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, hydroxyproline) provided in the examples for distinguishing between wild and farmed silver carp.

[0029] Figure 2 The confusion matrix test results of the free amino acid biomarker composition (phosphoserine, taurine, threonine, β-aminoisobutyric acid, hydroxyproline) provided in the examples for distinguishing between wild and farmed silver carp.

[0030] Figure 3The confusion matrix test results of the free amino acid biomarker composition (taurine, threonine, β-aminoisobutyric acid, hydroxyproline) provided in the examples for distinguishing between wild and farmed silver carp.

[0031] Figure 4 The confusion matrix test results of the free amino acid biomarker composition (phosphoserine, taurine, threonine, β-aminoisobutyric acid) provided in the examples for distinguishing between wild and farmed silver carp. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. Reagents not specifically described in detail herein are all conventional reagents and are commercially available; methods not specifically described in detail are all conventional experimental methods and can be learned from the prior art.

[0033] I. Discovery of free amino acid biomarkers in wild and farmed silver carp

[0034] 1. Collection of muscle samples from silver carp

[0035] The wild silver carp populations involved in this embodiment and the following embodiments were respectively from the Minjiang River, Jialing River, Three Gorges Reservoir area, Dongting Lake, and Poyang Lake in the Yangtze River. The farmed silver carp populations involved in the embodiments were respectively from farms in Hubei Province, Hunan Province, and Jiangsu Province. All samples were taken from the upper back muscle of silver carp, and were immediately frozen in liquid nitrogen after sampling and then transferred to a -80 ℃ freezer for storage, pending subsequent extraction of free amino acids.

[0036] 2. Determination and analysis of free amino acid content in samples

[0037] Accurately weigh 1.00 g of dried sample into a 50 mL centrifuge tube, add 25 mL of 10% sulfosalicylic acid solution, mix thoroughly, and place in a 4 ℃ refrigerator for precipitation for 2 h. Remove and centrifuge for 10 min at 8000 r / min. Take 100 µL of the supernatant and add 900 µL of 0.02 mol / L hydrochloric acid. Mix thoroughly, then add 1 mL of n-hexane, mix again, and centrifuge at 12000 r / min for 3 min. Take 1 mL of the lower layer solution after centrifugation, filter through a membrane, and analyze using an L-8900 fully automated amino acid analyzer.

[0038] Chromatographic conditions: A Li-type cation exchange column (4.6 mm × 60 mm, particle size 3 µm, stationary phase: sulfonic acid-based cation exchange resin) was used for chromatographic separation. Lithium salt buffer was used as the mobile phase, and a dual-wavelength detection mode (420 nm and 570 nm) was employed. The flow rate was set to 0.45 mL / min.-1 and 0.35 mL·min -1 The separation temperature was maintained at 57 °C, and a derivatization unit was connected after the column, using ninhydrin solution as the derivatizing agent. The reaction zone temperature was maintained at 135 °C. The mixed standard stock solution was diluted to 100 nmol·mL⁻¹. -1 Subsequently, both the sample and the test sample were injected at a volume of 20 µL, and the single collection time was set to 120 min.

[0039] The content of each amino acid in the sample solution is denoted as c. i The calculation formula is as follows:

[0040] c i (nmol / mL) = c s (nmol / mL) / As × A i

[0041] Where c i A represents the concentration of free amino acid i in the sample solution. i A represents the peak area of ​​free amino acid i in the sample solution. s c represents the peak area of ​​free amino acid S in the free amino acid standard working solution. s This represents the concentration of free amino acids s in the free amino acid standard working solution.

[0042] The content of each free amino acid in the sample is represented by Xi, and the calculation formula is as follows:

[0043] X i (µg / g) =c i (nmol / mL)×F×V (mL)×M (g / mol) / m (g)×10 -3

[0044] In the formula: Xi represents the content of free amino acid i in the sample. ci represents the concentration of free amino acid i in the sample test solution, F is the dilution factor, V is the volume transferred from the sample hydrolysate for volume adjustment, M is the molar mass of amino acid i, and m is the sample weight. Conversion factor 10 -3 Convert the sample content from nanograms (ng) to micrograms (µg).

[0045] 3. Analytical Methods

[0046] The t-test was used to calculate the significant differences in free amino acid data between different wild and farmed silver carp populations. The content of each free amino acid was entered into SPSS for the construction and analysis of a binary logistic regression model. The content of each free amino acid is presented as mean plus or minus standard deviation (SD), where an asterisk (*) indicates a statistically significant difference between the experimental and control groups. Specifically, *P < 0.05 indicates a significant difference, and **P < 0.01 indicates a highly significant difference.

[0047] 4. Results

[0048] Analysis revealed six differentially expressed free amino acids: phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline. The content (µg / g) and p-values ​​of these free amino acids are shown in Table 1.

[0049] Table 1. Free amino acid content in wild and farmed silver carp

[0050]

[0051] II. Free amino acid biomarker compositions for distinguishing between wild and farmed silver carp and their applications

[0052] Current data discrimination models primarily rely on multi-indicator model construction for comprehensive judgment. Among these, logistic regression, as a simple and intuitive model, offers numerous advantages. First, it is easy to understand and interpret, requiring no complex mathematical derivation or calculation, while also being computationally efficient. Second, logistic regression provides highly interpretable output results; classification can be achieved by setting thresholds, and the model's coefficients and intercepts can indicate the degree and direction of contribution of variables to the target variable.

[0053] 1. Selection of free amino acid biomarker compositions

[0054] This embodiment is based on the detection of free amino acid content in muscle samples of wild and farmed silver carp, and the binary logistic regression model is trained according to the detection results. Free amino acids with high statistical significance, small measurement error, and high data integrity are selected as biomarkers to distinguish between wild and farmed silver carp. The free amino acid biomarkers provided in this embodiment to distinguish between wild and farmed silver carp include phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline.

[0055] 2. Methods for distinguishing between wild and farmed silver carp

[0056] The embodiments also disclose a method for distinguishing between wild and farmed silver carp. The method includes: obtaining a first logistic regression model and a first threshold; obtaining the content of multiple biomarkers in a test sample; inputting the multiple contents into the first logistic regression model to obtain a first wild probability; and determining whether the test sample is a wild or farmed silver carp based on the magnitude of the first wild probability and the first threshold. The multiple biomarkers include phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline.

[0057] In some embodiments, the step of obtaining the first logistic regression model includes:

[0058] 1) Obtain training samples

[0059] In this example, muscle samples from 30 wild-caught and 30 farmed silver carp were analyzed for free amino acid content using the same methods described above. The peak areas of phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline in each sample were obtained. The target free amino acid content was calculated using the respective formulas. The contents of phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline in these muscle samples were used as training samples.

[0060] 2) Training

[0061] Using the contents of phosphoseserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline as independent variables and the probability of a sample being wild-type as the dependent variable, a binary logistic regression model was trained. In some embodiments, the training process of the binary logistic regression model includes: performing compositional analysis on the dependent and independent variables to determine whether they meet the preconditions for logistic regression; conducting significance tests on the independent variables, including performing degrees of freedom, significance, and scoring; performing the Hosmer-Lemersho test to verify the applicability and rationality of the model and ensure its accuracy; and training the binary logistic regression model to obtain a first logistic regression model based on the degree of influence of the independent variables on the dependent variable.

[0062] In some steps, the examples performed single-degree-of-freedom tests on the independent variables phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline in muscle samples from wild and farmed silver carp, including providing degrees of freedom, significance, and scoring. The results are shown in Table 2. The overall statistical scores for these six biomarkers as independent variables were high, with a significance level less than 0.01, indicating that they have a significant impact on the overall independent variables.

[0063] Table 2. Significance tests and scores of independent variables in the logistic regression models for biomarkers of wild and farmed silver carp.

[0064]

[0065] In some steps, the embodiments performed the Hosmer-Lemersho test on the logistic regression models of biomarkers for wild and farmed silver carp. Under multi-degree-of-freedom conditions, the significance of the binary logistic regression model was P=0.995>0.05, i.e., the null hypothesis was accepted. The results are shown in Table 3. The model fits the real data well and can reliably reflect the reliable relationship between the original variables.

[0066] Table 3. Hossmer-Lempers test for logistic regression models of biomarkers in wild and farmed silver carp.

[0067]

[0068] 3) Obtain the first logistic regression model

[0069] In some steps, the examples performed binary logistic regression analysis on logistic regression models of biomarkers for wild and farmed silver carp. Table 4 shows the effect size Exp(B) of the regression models and the significance of each biomarker.

[0070] Table 4. Analysis of binary logistic regression models in biomarker models of wild and farmed silver carp.

[0071]

[0072] Based on the muscle content of 30 wild silver carp samples and 30 farmed silver carp samples, the first logistic regression model obtained in the example is as follows: X = -1.802×A - 0.160×B - 0.362×C + 0.258×D + 1.823×E + 0.531×F + 129.612; First wild probability = 1 / (1+e －X ); where A, B, C, D, E, and F are the contents of phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline, respectively, and the first wild probability characterizes the probability that the test sample is a wild silver carp.

[0073] The steps provided in the embodiments are used to calculate the first threshold using ROC curves. Some embodiments include the following steps: using the contents of phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline from 30 muscle samples from wild silver carp and 30 muscle samples from farmed silver carp as a training set; training a binary logistic regression model with this training set to obtain multiple first wild probability values ​​as described above; calculating the sensitivity (true positive rate) and 1-specificity (false positive rate) for each first wild probability value, plotting an ROC curve with 1-specificity as the x-axis and sensitivity as the y-axis; calculating the Youden index (Youden index = sensitivity + specificity - 1) corresponding to each predicted wild probability, and obtaining the maximum Youden index, where the first predicted wild probability corresponding to the maximum Youden index is the first threshold.

[0074] In some embodiments, if the first wild probability is greater than a first threshold, the test sample is identified as a wild silver carp. If the first wild probability is less than the first threshold, the test sample is identified as a farmed silver carp.

[0075] In some embodiments, the first threshold obtained by calculating the ROC curve using the above test set is 0.415. That is, when the first wild probability is greater than 0.415, the test sample is determined to be a wild silver carp; when the first wild probability is less than 0.415, the test sample is determined to be a farmed silver carp.

[0076] 3. Identify applications and conduct tests.

[0077] In some test cases, phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline were selected from 45 muscle samples (18 wild and 27 farmed) as biomarkers. Their content was input into the first logistic regression model mentioned above to obtain the first wild probability. Based on the magnitude of the first wild probability and the first threshold (0.415), the test sample was determined to be either wild or farmed silver carp.

[0078] Binary logistic regression models were constructed individually for the contents of phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline in the training set. A first logistic regression model was constructed by combining the contents of these four substances. These models were then used to analyze 45 muscle samples, and the ROC curves were tested. The results are shown in Table 5. Table 5 shows that the first logistic regression model constructed using the combination of phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline contents exhibits higher AUC values, sensitivity, and specificity.

[0079] Table 5. ROC test results of various biomarkers and their combinations in distinguishing wild and farmed silver carp.

[0080]

[0081] The discrimination results were input into SPSS 20.0 analysis software for confusion matrix testing. The confusion matrix, also known as the error matrix, is a standard format for representing accuracy evaluation, expressed as an n x n matrix. Specific evaluation indicators include overall accuracy, mapping accuracy, and user accuracy, which reflect the accuracy of image classification from different perspectives. Each column of the confusion matrix represents the predicted class, and the total number in each column represents the number of data points predicted as belonging to that class; each row represents the true class of the data, and the total number of data points in each row represents the number of data instances belonging to that class. The results are as follows: Figure 1As shown, only one of the 45 silver carp samples was incorrectly judged, and the overall accuracy of the cross-validation was as high as 97.8%.

[0082] This demonstrates that using phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline as biomarkers can be used to distinguish between wild and farmed silver carp in the Yangtze River basin. The excellent ROC curve results, good diagnostic accuracy, and good complementarity of the biomarkers all indicate that these six free amino acids have the potential to differentiate between wild and farmed silver carp.

[0083] III. Free amino acid biomarker compositions for distinguishing between wild and farmed silver carp and their applications

[0084] 1. Selection of free amino acid biomarker compositions

[0085] This example is based on the detection of free amino acid content in muscle samples from wild and farmed silver carp, and the training of a binary logistic regression model based on the detection results. Free amino acids with high statistical significance, small measurement error, and high data integrity were selected as biomarkers to distinguish between wild and farmed silver carp. The free amino acid biomarkers provided in this example for distinguishing between wild and farmed silver carp include phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline.

[0086] The embodiments also provide kits for detecting these biomarkers, which include reagents for detecting these biomarkers, such as reagents required for automated free amino acid analyzer testing.

[0087] 2. Methods for distinguishing between wild and farmed silver carp

[0088] The embodiments also disclose a method for distinguishing between wild and farmed silver carp. The method includes: obtaining a second logistic regression model and a second threshold; obtaining the content of multiple biomarkers in the test sample; inputting the multiple contents into the second logistic regression model to obtain a second wild probability; and determining whether the test sample is a wild or farmed silver carp based on the magnitude of the second wild probability and the second threshold. The multiple biomarkers include phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline.

[0089] In some embodiments, the step of obtaining the second logistic regression model includes:

[0090] 1) Obtain training samples

[0091] In this example, muscle samples from 30 wild silver carp and 30 farmed silver carp were analyzed using the same automated free amino acid analyzer described above. The peak areas of phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline in each sample were obtained. The content of the target free amino acids was calculated using the respective formulas. The contents of phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline in these muscle samples were used as training samples.

[0092] 2) Training

[0093] Using the contents of phosphoseserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline as independent variables and the probability of a sample being wild-type as the dependent variable, a binary logistic regression model was trained. In some embodiments, the training process of the binary logistic regression model includes: performing compositional analysis on the dependent and independent variables to determine whether they meet the preconditions for logistic regression; conducting significance tests on the independent variables, including performing degrees of freedom, significance, and scoring; performing the Hosmer-Lemersho test to verify the applicability and rationality of the model and ensure its accuracy; and training the binary logistic regression model to obtain a second logistic regression model based on the degree of influence of the independent variables on the dependent variable.

[0094] In some steps, the examples performed single-degree-of-freedom tests on the muscle sample contents of the independent variables phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline, including providing the degrees of freedom, significance, and scoring. The results are shown in Table 6. The overall statistical scores for the contents of these five biomarkers as independent variables were high, with a significance level less than 0.001, indicating that the independent variables had a highly significant impact on the corresponding variables overall.

[0095] Table 6. Significance tests and scores of independent variables in the logistic regression models of biomarkers for wild and farmed silver carp.

[0096]

[0097] In some steps, the embodiments performed the Hosmer-Lemersho test on the logistic regression models of biomarkers for wild and farmed silver carp. Under multi-degree-of-freedom conditions, the significance of the binary logistic regression model was P=0.832>0.05, i.e., the null hypothesis was accepted. The results are shown in Table 7. The established binary logistic regression model fits the real data well and can reliably reflect the reliable relationship between the original variables.

[0098] Table 7. Hossmer-Lempers test for logistic regression models of wild and farmed silver carp.

[0099]

[0100] 3) Obtain the second logistic regression model

[0101] In some steps, the examples performed binary logistic regression analysis on biomarkers for distinguishing between wild and farmed silver carp. Table 8 shows the effect size Exp(B) of the regression model and the significance of each biomarker.

[0102] Table 8. Binary logistic regression analysis of wild and farmed silver carp.

[0103]

[0104] Based on the muscle content of the 30 wild silver carp muscle samples and the 30 farmed silver carp muscle samples mentioned above, the second logistic regression model is obtained as follows: X = -0.318×A - 0.029×B - 0.059×C + 0.691×D + 0.117×E + 24.246; Second wild probability = 1 / (1+e －X ); where A, B, C, D, and E represent the contents of phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline, respectively, and the second wild probability represents the probability that the test sample is a wild silver carp.

[0105] The steps provided in the embodiments are used to calculate the second threshold using ROC curves. Some embodiments include the following steps: using the contents of phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline from 30 muscle samples of wild silver carp and 30 muscle samples of farmed silver carp as a training set; training a binary logistic regression model with this training set to obtain multiple second wild probability values ​​as described above; calculating the sensitivity (true positive rate) and 1-specificity (false positive rate) for each second wild probability value, plotting an ROC curve with 1-specificity as the x-axis and sensitivity as the y-axis; calculating the Youden index (Youden index = sensitivity + specificity - 1) corresponding to each predicted wild probability, and obtaining the maximum Youden index, where the first predicted wild probability corresponding to the maximum Youden index is the second threshold.

[0106] In some embodiments, if the second wild probability is greater than a second threshold, the test sample is identified as a wild silver carp. If the second wild probability is less than the second threshold, the test sample is identified as a farmed silver carp.

[0107] In some embodiments, the second threshold obtained by calculating the ROC curve using the above test set is 0.529. That is, when the second wild probability is greater than 0.529, the test sample is determined to be a wild silver carp; when the second wild probability is less than 0.529, the test sample is determined to be a farmed silver carp.

[0108] 3. Identify applications and conduct tests.

[0109] In some test cases, phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline were selected from 45 muscle samples (18 wild and 27 farmed) as biomarkers. Their content was input into the second logistic regression model mentioned above to obtain the second wild probability. Based on the magnitude of the second wild probability and the second threshold (0.529), the test sample was determined to be either wild or farmed silver carp.

[0110] Binary logistic regression models were constructed individually for the contents of phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline in the training set. A second logistic regression model was constructed by combining the contents of these four amino acids. These models were then used to analyze 45 muscle samples, and the ROC curves were used to test the results. The results are shown in Table 9. Table 9 shows that the second logistic regression model constructed by combining the contents of phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline exhibits higher AUC values, sensitivity, and specificity in the analysis of the test samples.

[0111] Table 9. ROC test results of various biomarkers and their combinations in distinguishing wild and farmed silver carp.

[0112]

[0113] The discrimination results were input into SPSS 20.0 analysis software for confusion matrix testing, and the results are as follows. Figure 2 As shown, only 4 out of 45 silver carp samples were incorrectly judged, and the overall accuracy of the cross-validation was as high as 91.1%.

[0114] This demonstrates that using phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline as biomarkers can be used to distinguish between wild and farmed silver carp in the Yangtze River basin. The excellent ROC curve results, good diagnostic accuracy, and good complementarity of the biomarkers all indicate that these five free amino acids have the potential to differentiate between wild and farmed silver carp.

[0115] IV. Free amino acid biomarker compositions for distinguishing between wild and farmed silver carp and their applications

[0116] 1. Selection of free amino acid biomarker compositions

[0117] This example is based on the detection of free amino acid content in muscle samples from wild and farmed silver carp, and the training of a binary logistic regression model based on the detection results. Free amino acids with high statistical significance, small measurement error, and high data integrity were selected as biomarkers to distinguish between wild and farmed silver carp. The free amino acid biomarkers provided in this example for distinguishing between wild and farmed silver carp include taurine, threonine, β-aminoisobutyric acid, and hydroxyproline.

[0118] The embodiments also provide kits for detecting these biomarkers, which include reagents for detecting these biomarkers, such as reagents required for automated free amino acid analyzer testing.

[0119] 2. Methods for distinguishing between wild and farmed silver carp

[0120] The embodiments also disclose a method for distinguishing between wild and farmed silver carp. The method includes: obtaining a third logistic regression model and a third threshold; obtaining the content of multiple biomarkers in the test sample; inputting the multiple contents into the third logistic regression model to obtain a third wild probability; and determining whether the test sample is a wild or farmed silver carp based on the magnitude of the third wild probability and the third threshold. The multiple biomarkers include taurine, threonine, β-aminoisobutyric acid, and hydroxyproline.

[0121] In some embodiments, the step of obtaining the third logistic regression model includes:

[0122] 1) Obtain training samples

[0123] In this example, muscle samples from 30 wild silver carp and 30 farmed silver carp were analyzed using the same automated free amino acid analyzer described above. The peak areas of taurine, threonine, β-aminoisobutyric acid (β-aminobutyric acid), and hydroxyproline in each sample were obtained. The content of the target free amino acids was calculated using the respective formulas. The taurine, threonine, β-aminoisobutyric acid, and hydroxyproline content of these muscle samples were used as training samples.

[0124] 2) Training

[0125] Using the contents of taurine, threonine, β-aminoisobutyric acid, and hydroxyproline as independent variables and the probability of a sample being wild-type as the dependent variable, a binary logistic regression model was trained. In some embodiments, the training process of this binary logistic regression model includes: performing compositional analysis on the dependent and independent variables to determine whether they meet the preconditions for logistic regression; conducting significance tests on the independent variables, including performing degrees of freedom, significance, and scoring; performing the Hosmer-Lemersho test to verify the applicability and rationality of the model and ensure its accuracy; training the binary logistic regression model, and obtaining a third logistic regression model based on the degree of influence of the independent variables on the dependent variable.

[0126] In some steps, the examples used muscle samples containing the independent variables taurine, threonine, β-aminoisobutyric acid, and hydroxyproline for single-degree-of-freedom tests, including providing the degrees of freedom, significance, and scoring. Table 10 shows that the overall statistical scores for the independent variables were high, with a significance level less than 0.001, indicating that the independent variables had a highly significant impact on the corresponding variables overall.

[0127] Table 10. Significance tests and scores of independent variables in the logistic regression models of biomarkers for wild and farmed silver carp.

[0128]

[0129] To further analyze the impact of constants and independent variables on the dependent variable in terms of multiple degrees of freedom, the example also conducted the Hosmer-Lemersho test on the logistic regression model. Under multiple degrees of freedom conditions, the significance of the binary logistic regression model was P=0.703>0.05, meaning the null hypothesis was accepted. The results, shown in Table 11, indicate that the established binary logistic regression model fits the actual data well and reliably reflects the reliable relationship between the original variables.

[0130] Table 11. Hossmer-Lempers test for logistic regression models of wild and farmed silver carp.

[0131]

[0132] 3) Obtain the third logistic regression model

[0133] In some steps, the embodiments performed binary logistic regression analysis on the logistic regression models of biomarkers for wild and farmed silver carp. Table 12 shows the effect size Exp(B) of the regression model and the significance of each biomarker.

[0134] Table 12 Binary Logistic Regression Analysis of Wild and Farmed Silver Carp

[0135]

[0136] Based on the muscle content of 30 wild silver carp samples and 30 farmed silver carp samples, the third logistic regression model obtained in the example is as follows: X = -0.017×A - 0.045×B + 0.475×C + 0.089×D + 11.437; Third wild probability = 1 / (1+e －X) A, B, C, and D represent the contents of taurine, threonine, β-aminoisobutyric acid, and hydroxyproline, respectively, while the third wild probability characterizes the probability that the test sample is a wild silver carp.

[0137] The steps provided in the embodiments are used to calculate the third threshold using ROC curves. Some embodiments include the following steps: using the contents of taurine, threonine, β-aminoisobutyric acid, and hydroxyproline from 30 muscle samples of wild silver carp and 30 muscle samples of farmed silver carp as a training set; training a binary logistic regression model with this training set to obtain multiple third wild probability values ​​as described above; calculating the sensitivity (true positive rate) and 1-specificity (false positive rate) for each third wild probability value, plotting an ROC curve with 1-specificity as the x-axis and sensitivity as the y-axis; calculating the Youden index (Youden index = sensitivity + specificity - 1) corresponding to each predicted wild probability, and obtaining the maximum Youden index, where the third predicted wild probability corresponding to the maximum Youden index is the third threshold.

[0138] In some embodiments, if the third wild probability is greater than a third threshold, the test sample is identified as a wild silver carp. If the third wild probability is less than the third threshold, the test sample is identified as a farmed silver carp.

[0139] In some embodiments, the third threshold obtained by calculating the ROC curve using the above test set is 0.441. That is, when the third wild probability is greater than 0.441, the test sample is determined to be a wild silver carp; when the third wild probability is less than 0.441, the test sample is determined to be a farmed silver carp.

[0140] 3. Identify applications and conduct tests.

[0141] In some test cases, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline were selected from 45 muscle samples (18 wild and 27 farmed) as biomarkers. Their content was input into the third logistic regression model mentioned above to obtain the third wild probability. Based on the magnitude of the third wild probability and the third threshold (0.441), the test sample was determined to be either wild or farmed silver carp.

[0142] The contents of taurine, threonine, β-aminoisobutyric acid, and hydroxyproline in the training set were used to construct binary logistic regression models individually, and a third logistic regression model was constructed by combining the contents of these four substances. These models were then used to analyze 45 muscle samples, and the ROC curves were used to test the results. The results are shown in Table 13. Table 13 shows that the third logistic regression model constructed by combining the contents of taurine, threonine, β-aminoisobutyric acid, and hydroxyproline exhibits higher AUC values, sensitivity, and specificity.

[0143] Table 13 ROC test results of various biomarkers and their combinations in distinguishing wild and farmed silver carp.

[0144]

[0145] The discrimination results were input into SPSS 20.0 analysis software for confusion matrix testing, and the results are as follows. Figure 3 As shown, only 5 out of 45 silver carp samples were incorrectly identified, and the overall accuracy of the cross-validation was as high as 88.9%.

[0146] This demonstrates that a logistic regression model constructed using taurine, threonine, β-aminoisobutyric acid, and hydroxyproline as biomarkers can be used to distinguish between wild and farmed silver carp in the Yangtze River basin. The excellent ROC curve results, good diagnostic accuracy, and strong complementarity of the biomarkers indicate that these four free amino acids have the potential to differentiate between wild and farmed silver carp.

[0147] V. Biomarker Compositions for Differentiating Wild and Farmed Silver Carp and Their Applications

[0148] 1. Selection of free amino acid biomarker compositions

[0149] This example is based on the detection of free amino acid content in muscle samples from wild and farmed silver carp, and the training of a binary logistic regression model based on the detection results. Free amino acids with high statistical significance, small measurement error, and high data integrity are selected as biomarkers to distinguish between wild and farmed silver carp. The free amino acid biomarkers provided in this example to distinguish between wild and farmed silver carp include phosphoserine, taurine, threonine, and β-aminoisobutyric acid.

[0150] The embodiments also provide kits for detecting these biomarkers, which include reagents for detecting these biomarkers, such as reagents required for automated free amino acid analyzer testing.

[0151] 2. Methods for distinguishing between wild and farmed silver carp

[0152] The embodiments also disclose a method for distinguishing between wild and farmed silver carp. The method includes: obtaining a fourth logistic regression model and a fourth threshold; obtaining the content of multiple biomarkers in the test sample; inputting the multiple contents into the fourth logistic regression model to obtain a fourth wild probability; and determining whether the test sample is a wild or farmed silver carp based on the magnitude of the fourth wild probability and the fourth threshold. The multiple biomarkers include phosphoserine, taurine, threonine, and β-aminoisobutyric acid.

[0153] In some embodiments, the step of obtaining the fourth logistic regression model includes:

[0154] 1) Obtain training samples

[0155] In this example, muscle samples from 30 wild silver carp and 30 farmed silver carp were analyzed using the same automated free amino acid analyzer described above. The peak areas of phosphoserine, taurine, threonine, and β-aminoisobutyric acid (β-aminobutyric acid) in each sample were obtained. The content of the target free amino acids was calculated according to the respective formulas. The contents of phosphoserine, taurine, threonine, and β-aminoisobutyric acid in these muscle samples were used as training samples.

[0156] 2) Training

[0157] Using the contents of phosphoseserine, taurine, threonine, and β-aminoisobutyric acid as independent variables and the probability of a sample being wild-type as the dependent variable, a binary logistic regression model was trained. In some embodiments, the training process of this binary logistic regression model includes: performing compositional analysis on the dependent and independent variables to determine whether they meet the preconditions for logistic regression; conducting significance tests on the independent variables, including performing degrees of freedom, significance, and scoring; performing the Hosmer-Lemersho test to verify the applicability and rationality of the model and ensure its accuracy; training the binary logistic regression model, and obtaining a fourth logistic regression model based on the degree of influence of the independent variables on the dependent variable.

[0158] In some steps, the examples performed single-degree-of-freedom tests on the muscle sample contents of the independent variables phosphoserine, taurine, threonine, and β-aminoisobutyric acid, including providing the degrees of freedom, significance, and scoring. The results are shown in Table 14. The overall statistical scores of these independent variables were high, with a significance level of less than 0.001, indicating that the independent variables have a highly significant impact on the corresponding variables overall.

[0159] Table 14. Significance tests and scores of independent variables in the combined biomarker model of wild and farmed silver carp.

[0160]

[0161] In some steps, the embodiments performed the Hosmer-Lemersho test on the logistic regression models of biomarkers for wild and farmed silver carp. Under multi-degree-of-freedom conditions, the significance of the binary logistic regression model was P=0.748>0.05, i.e., the null hypothesis was accepted. As shown in Table 15, the established binary logistic regression model fits the real data well and can reliably reflect the reliable relationship between the original variables.

[0162] Table 15. Hossmer-Lemersho test for the binary logistic regression model of wild and farmed silver carp.

[0163]

[0164] 3) Obtain the fourth logistic regression model

[0165] In some steps, the examples performed binary logistic regression analysis on biomarkers for distinguishing between wild and farmed silver carp. Table 16 shows the effect size Exp(B) of the regression model and the significance of each biomarker.

[0166] Table 16 Binary Logistic Regression Analysis of Wild and Farmed Silver Carp

[0167]

[0168] Based on the muscle content of the above 30 wild silver carp and 30 farmed silver carp muscle samples, the fourth logistic regression model is obtained as follows: X = -0.307×A - 0.019×B - 0.027×C + 0.404×D + 19.318; Fourth wild probability = 1 / (1+e －X ); where A, B, C, and D are the contents of phosphoserine, taurine, threonine, and β-aminoisobutyric acid, respectively, and the fourth wild probability characterizes the probability that the test sample is a wild silver carp.

[0169] The steps provided in the embodiments are used to calculate the fourth threshold using ROC curves. Some embodiments include the following steps: using the contents of phosphoserine, taurine, threonine, and β-aminoisobutyric acid from 30 muscle samples of wild silver carp and 30 muscle samples of farmed silver carp as a training set; training a binary logistic regression model with this training set to obtain multiple fourth wild probability values ​​as described above; calculating the sensitivity (true positive rate) and 1-specificity (false positive rate) for each fourth wild probability value, plotting an ROC curve with 1-specificity as the x-axis and sensitivity as the y-axis; calculating the Youden index (Youden index = sensitivity + specificity - 1) corresponding to each predicted wild probability, and obtaining the maximum Youden index, where the fourth predicted wild probability corresponding to the maximum Youden index is the fourth threshold.

[0170] In some embodiments, if the fourth wild probability is greater than the fourth threshold, the test sample is identified as a wild silver carp. If the fourth wild probability is less than the fourth threshold, the test sample is identified as a farmed silver carp.

[0171] In some embodiments, the fourth threshold obtained by calculating the ROC curve using the above test set is 0.645. That is, when the fourth wild probability is greater than 0.645, the test sample is determined to be a wild silver carp; when the fourth wild probability is less than 0.645, the test sample is determined to be a farmed silver carp.

[0172] 3. Identify applications and conduct tests.

[0173] In some test cases, phosphoserine, taurine, threonine, and β-aminoisobutyric acid were selected from 45 muscle samples (18 wild and 27 farmed) as biomarkers. Their content was input into the fourth logistic regression model mentioned above to obtain the fourth wild probability. Based on the magnitude of the fourth wild probability and the fourth threshold (0.645), the test sample was determined to be either wild or farmed silver carp.

[0174] Binary logistic regression models were constructed individually for the contents of phosphoserine, taurine, threonine, and β-aminoisobutyric acid in the training set. A fourth logistic regression model was also constructed by combining the contents of these four amino acids. These models were then used to analyze 45 muscle samples, and the ROC curves were used to test the results. The results are shown in Table 17. Table 17 shows that the fourth logistic regression model constructed by combining the contents of phosphoserine, taurine, threonine, and β-aminoisobutyric acid exhibits higher AUC values, sensitivity, and specificity in the analysis of the test samples.

[0175] Table 17 ROC test results of various biomarkers and their combinations in distinguishing wild and farmed silver carp.

[0176]

[0177] The discrimination results are then input into SPSS 20.0 analysis software for confusion matrix testing, such as... Figure 4 As shown, only 7 out of 45 silver carp samples were incorrectly identified, and the overall accuracy of the cross-validation was as high as 84.4%.

[0178] This demonstrates the good stability of using phosphoserine, taurine, threonine, and β-aminoisobutyric acid as biomarkers and regression models to distinguish between wild and farmed silver carp. The excellent ROC curve results, good correct diagnostic rate, and good complementarity of the biomarkers indicate that these four free amino acids have the potential to differentiate between wild and farmed silver carp.

[0179] It is understood that, based on the logistic regression model and discrimination threshold disclosed in this application, those skilled in the art can easily construct an automated discrimination system. This system may include: a module for inputting or receiving amino acid content data; a processing module for running the discrimination model to calculate the wild-type probability value; and a judgment module for comparing the probability value with a preset threshold and outputting the discrimination result. The computer program implementing the system's functions may be stored on a computer-readable storage medium. For example, the program may be connected to an automated amino acid analyzer to automate the entire process from detection to judgment.

[0180] The present application has been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present application. The descriptions of the embodiments above are only for the purpose of helping to understand the present application and its core ideas. It should be noted that those skilled in the art can make several improvements and modifications to the present application without departing from the principles of the present application, and these improvements and modifications also fall within the protection scope of the claims of the present application.

Claims

1. A combination of free amino acid biomarkers for differentiating wild and farmed silver carp, characterized in that, The biomarker combination includes any one of the following (a)-(d): (a) Phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, hydroxyproline; (b) Phosphoseserine, taurine, threonine, β-aminoisobutyric acid, hydroxyproline; (c) Taurine, threonine, β-aminoisobutyric acid, hydroxyproline; (d) Phosphoseserine, taurine, threonine, β-aminoisobutyric acid; A discriminant model can be constructed using the content information of any of the above combinations of biomarkers.

2. A method for distinguishing between wild and farmed silver carp, characterized in that, Includes the following steps: S1. Obtain muscle samples from the silver carp to be tested; S2. Determine the content of all amino acids in the biomarker combination as described in claim 1 in the sample; S3. Input the measured content into the preset logistic regression model to calculate the probability that the sample is a wild silver carp; S4. Compare the wild probability value with a preset discrimination threshold. If it is greater than the discrimination threshold, it is determined to be a wild silver carp; if it is less than the discrimination threshold, it is determined to be a farmed silver carp.

3. The method according to claim 2, characterized in that, The preset discrimination threshold is determined by the following method: Muscle samples from wild and farmed silver carp of known sources were obtained as the training set; The content of the biomarker combination in the training set samples was determined; Using the content as the independent variable and the sample source as the dependent variable, a binary logistic regression analysis was performed to obtain the logistic regression model; The wild probability value of each sample in the training set is calculated based on the logistic regression model. Plot the ROC curve, calculate the Youden index corresponding to each wild probability value, and determine the wild probability value corresponding to the highest Youden index as the discrimination threshold.

4. The method according to claim 2, characterized in that, When the biomarker combination is the combination (a) in claim 1, the logistic regression model is the first logistic regression model, and its calculation formula is as follows: X = -1.802×A - 0.160×B - 0.362×C + 0.258×D + 1.823×E + 0.531×F + 129.612; the wild probability is the first wild probability, which has a value of 1 / (1+e －X ); Wherein, A, B, C, D, E, F represent the contents of phosphoserine, taurine, threonine, sarcosine, β-aminoisobutyric acid, and hydroxyproline in the sample, respectively; the discrimination threshold is 0.

415.

5. The method according to claim 2, characterized in that, When the biomarker combination is the combination (b) in claim 1, the logistic regression model is the second logistic regression model, and its calculation formula is as follows: X = -0.318×A - 0.029×B - 0.059×C + 0.691×D + 0.117×E + 24.246; the wild probability is the second wild probability, with a value of 1 / (1+e －X ); Wherein A, B, C, D, and E are phosphoserine, taurine, threonine, β-aminoisobutyric acid, and hydroxyproline, respectively; the discrimination threshold is 0.

529.

6. The method according to claim 2, characterized in that, When the biomarker combination is combination (c) in the first aspect, the logistic regression model is the third logistic regression model, and its calculation formula is as follows: X = -0.017×A - 0.045×B + 0.475×C + 0.089×D + 11.437; the wild probability is the third wild probability, with a value of 1 / (1+e －X ); Wherein, A, B, C, and D are taurine, threonine, β-aminoisobutyric acid, and hydroxyproline, respectively; the discrimination threshold is 0.

441.

7. The method according to claim 2, characterized in that, When the biomarker combination is combination (d) in the first aspect, the logistic regression model is the fourth logistic regression model, and its calculation formula is as follows: X = -0.307×A - 0.019×B - 0.027×C + 0.404×D + 19.318; the wild probability is the third wild probability, with a value of 1 / (1+e －X ); Wherein A, B, C, and D are phosphoserine, taurine, threonine, and β-aminoisobutyric acid, respectively; the discrimination threshold is 0.

645.

8. The method according to claim 2, characterized in that, In step S2, an automated amino acid analyzer is used for determination. The chromatographic separation uses a Li-type cation exchange column with lithium salt buffer as the mobile phase. The separation temperature is kept constant at 57°C. A derivatization unit is connected after the column, and the derivatizing agent is ninhydrin solution. The temperature of the reaction zone is maintained at 135°C.

9. The use of the biomarker combination of claim 1 and / or the reagents used to detect it in distinguishing between wild and farmed silver carp.

10. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, is used to implement steps S3 and S4 of the method according to any one of claims 2-8.

Images