Marker combination for identifying year of ham and application of marker combination

Abstract

The invention relates to a marker combination for identifying the year of ham and application of the marker combination, and belongs to the field of food identification, firstly, the content of flavor compounds is detected through a gas chromatography-triple quadrupole mass spectrometer, and the combination capable of being used as an identification marker is obtained through two rounds of screening, so that the year of Jinhua ham is identified; moreover, key discriminant compound data can be screened and simplified, and a set of equipment and algorithm for realizing the model are provided.

Description

Technical Field

[0001] The invention belongs to the field of food identification, and in particular relates to a marker combination for identifying the year of ham and an application thereof. Background Art

[0002] Jinhua ham is a geographical indication product in China. Originated from Jinhua City, Zhejiang Province, Jinhua ham has a long history and unique production process. It enjoys a high reputation in China and internationally for its unique flavor and quality. According to China's geographical indication product protection system, Jinhua ham obtained the geographical indication protection product certification in 2004, which means that the production of Jinhua ham must be carried out in the Jinhua area according to specific standards and processes.

[0003] The identification of Jinhua ham is indeed difficult, mainly reflected in the following aspects:

[0004] Complexity of process: The production process of Jinhua ham is very unique, including salting, air-drying, fermentation and other steps, and these processes may vary from year to year and from producer to producer. The color, texture, aroma and taste of each ham are closely related to production process, climatic conditions and other factors, so it is difficult to control standardization and consistency.

[0005] Protection of geographical indications: To determine whether Jinhua ham meets the standards of geographical indication products, it is necessary to determine its origin, production process and quality characteristics. Although the geographical indication certification of Jinhua ham has been recognized, there are still some counterfeit or substandard Jinhua hams on the market, so accurate identification requires the use of professional institutions and advanced technologies, such as DNA identification and chemical analysis.

[0006] Although Jinhua ham has obtained geographical indication certification, there are still some ham products in the market that are not of origin or do not meet traditional standards. Therefore, consumers need to be more cautious when purchasing and choose reputable brands and channels.

[0007] In recent years, with the improvement of food safety and quality testing technology, it has become more feasible to scientifically identify Jinhua ham using modern technologies (such as DNA fingerprinting, isotope analysis, chemical composition analysis, etc.). These technologies can help distinguish authentic Jinhua ham from other counterfeit and inferior products, but they still require certain professional equipment and technical support.

[0008] Regarding the method for identifying the grade of Jinhua ham, the ham and cured meat samples are cut into pieces and left to stand, and the sampling needle of the electronic nose is used to adsorb the headspace gas in the container. The sensor collects data and then performs principal component analysis, linear discriminant analysis and partial least squares analysis on the data. A mathematical model is established based on the results to achieve the identification and prediction of the grade of Jinhua ham.

[0009] Regarding the identification method of Jinhua ham quality, solid phase microextraction (SPME) technology was used to extract the flavor components in Jinhua ham, and then the characteristic fingerprint peaks of Jinhua ham odor were collected through gas chromatography-mass spectrometry (GC-MS), and the standard flavor fingerprint of Jinhua ham was established. The quality of ham was judged by fingerprint technology and principal component analysis.

[0010] Method for identifying the origin of ham: Gas phase-ion mobility spectrometry (GC-IMS) is used to detect the volatile components of hams from different origins, and statistical analysis methods are used to compare and analyze the response intensity data of ham standards, and a volatile organic compound database for hams from different origins is established; the volatile compounds of the ham to be tested are detected by gas phase-ion mobility spectrometry with the same parameters, and compared with the established database to determine the origin of the ham to be tested.

[0011] In addition, there is also prior art that discloses a published patent on a method for identifying strawberry varieties and maturity. It is based on the collection of large sample data of strawberry fruit flavor substances of different varieties and maturity, and the classification of the measured flavor substance data sets, and the construction and optimization of four algorithm models adapted to them, namely neural network (NN), partial least squares (PLS), decision tree, and support vector machine (SVM), to achieve prediction of different strawberry varieties and maturity. The prediction effect of the neural network model is significantly better than the synchronously constructed partial least squares, decision tree and support vector machine algorithm models.

[0012] Currently, there are no detection markers and computer models with good specificity. Therefore, the present invention will develop and apply markers and algorithm models for detecting and identifying Jinhua ham based on the existing technology. Summary of the invention

[0013] In view of the current lack of a simple and objective method for identifying the marker, method, system and algorithm for the fermentation year of ham, the present invention provides a marker combination, a machine learning algorithm and a device for identifying the year of ham. The year of Jinhua ham is identified by this simple and objective identification method, which can avoid the subjectivity, variability and instability caused by manual identification and reduce labor costs.

[0014] SHAP (SHapley Additive exPlanations) is a method for explaining the output of machine learning models. It is based on the theory of Shapley values, which comes from cooperative game theory and is used to fairly distribute the benefits of each member in a cooperative group. Shapley value: represents the average marginal contribution of a feature to the model prediction results in all possible feature combinations. By traversing all possible feature subsets, the change in model output after adding the feature is calculated, and finally the average is taken. Properties of SHAP value: efficiency, the sum of the contribution values ​​of all features is equal to the actual output of the model. Symmetry, if two features contribute the same to all predictions, their SHAP values ​​are also equal. Zero contribution, if a feature has no effect on the prediction under any combination, its SHAP value is zero. It can provide a theoretically fair and consistent feature contribution decomposition. It can explain the output of black box models (such as deep learning, random forests, XGBoost, etc.).

[0015] OAV (Odor Activity Value) is an important indicator to measure the contribution of a compound to the overall odor, and is defined as the ratio of the actual concentration of the compound to its odor threshold. OAV = actual concentration odor threshold / actual concentration.

[0016] GC-QQQ is the abbreviation of Gas Chromatography-Triple Quadrupole Mass Spectrometer, which is an analytical instrument that combines gas chromatography (GC) with triple quadrupole mass spectrometry (QQQ), mainly used for high-sensitivity and high-selectivity chemical analysis.

[0017] XGBoost (Extreme Gradient Boosting) is an efficient and flexible machine learning algorithm based on gradient boosting, which is widely used in classification and regression problems of structured data. It is popular for its powerful performance and fast speed, especially in many data science competitions (such as Kaggle) as the core part of the winner's model.

[0018] To achieve the above object, the specific implementation scheme of the present invention is as follows:

[0019] The first aspect of the present invention provides a marker combination for identifying the fermentation year of ham, wherein the marker combination is composed of at least two of the following compounds, the compounds comprising: (E)-2-octen-1-ol, 1-nonanol, 1-octen-3-ol, 2,4-decadienal, 2-pentanone, 3-octanone, acetophenone, γ-nonalactone, γ-octanolactone, p-cresol, nonanal, octanol, octanoic acid, isovaleraldehyde, lauric acid, n-heptanol, n-decanoic acid and n-hexanol;

[0020] Further, the composition comprises at least two of isovaleraldehyde, 1-octen-3-ol, n-decanoic acid, octanoic acid, 2,4-decadienal, (E)-2-octen-1-ol, 3-octanone, 1-nonanol, and nonanal, and preferably at least three of them;

[0021] Further, the composition comprises at least n-decanoic acid and 1-octen-3-ol;

[0022] Furthermore, the composition is used to identify the fermentation year of Jinhua ham; wherein the content of the compound is measured on Jinhua ham samples of different years using a gas chromatography-mass spectrometer GC-QQQ to collect raw data;

[0023] Preferably, the chromatographic conditions are as follows: initial temperature 35-42°C, maintained for 3-8 min, then heated to 240-260°C at a rate of 2-3°C / min, maintained for 8-15 min; the carrier gas is helium, the gas flow rate is 0.8-1.2.0 mL / min, the split ratio is 4-6.0, and the injection chamber temperature is 240-260°C;

[0024] Preferably, the voltage of the electron ionization mass spectrometer is 65-75 eV, the quadrupole scanning range is 35-400 m / z, the temperature of the ion source is 180-220° C., the interface temperature is 240-260° C., the injection volume is 0.5-1.2 μL, and the solvent delay time is 2-4 min.

[0025] The second aspect of the present invention is to provide the use of the composition described in the first aspect in identifying the fermentation year of ham; preferably, the ham is Jinhua ham; and the fermentation year is 3 years or more.

[0026] The third aspect of the present invention is to provide a method for screening markers that can identify the fermentation year of ham, wherein the method is:

[0027] 1) Obtain ham samples from different years;

[0028] 2) Using gas chromatography-mass spectrometry (GC-QQQ) to measure the flavor compounds in ham samples of different years and collect raw data;

[0029] 3) Data processing: normalizing the quantitative analysis data of flavor compounds in the raw data;

[0030] 4) Preliminary screening of flavor compounds with potential identification functions;

[0031] 5) Rescreening the compounds obtained in step 4) to obtain key discriminant compounds.

[0032] Furthermore, the normalized formula in step 3) is:

[0033] The normalization formula is as follows:

[0034]

[0035] Among them, x_new represents the normalized data, x represents the data before normalization, x_min represents the maximum value of the data, and x_min represents the minimum value of the data;

[0036] Further, the condition of the initial screening in step 4) is OAV>1;

[0037] Further, the rescreening condition in step 5) is

[0038] a.SHAP feature selection;

[0039] b. Vintage samples and distribution diagram of key flavor compounds they contain.

[0040] SHAP feature selection is a tool for explaining the predictions of machine learning models. The implementation steps are: 1) train the XGBoost model using the dataset; 2) calculate the SHAP value of each feature; 3) display the SHAP value through visualization methods.

[0041] The fourth aspect of the present invention provides the use of the marker combination described in the first aspect or the marker identified by the method described in the third aspect in preparing a model or system for identifying the fermentation year of ham.

[0042] The beneficial effects of the present invention include:

[0043] 1) The discriminant model of the present invention for the year of Jinhua ham has high accuracy and universality in model algorithm. The key discriminant compounds selected by the present invention have high accuracy in the year identification model based on logistic regression, support vector machine, naive Bayes and decision tree algorithm, and the average accuracy of each model is close.

[0044] 2) The present invention not only identifies the year of Jinhua ham, but also screens and simplifies key discriminant compound data, and provides a set of equipment for realizing the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 Schematic diagram of the overall method and steps of the specific implementation plan;

[0046] Figure 2 Two key flavor compound distributions: bar graph, scatter plot, Figure 2 A is the bar chart of the annual content of n-decanoic acid, Figure 2 B is the bar chart of 1-octen-3-ol content by year. Figure 2 C is a scatter plot of the combined years of the two;

[0047] Figure 3 Visualization diagram of SHAP feature selection for key flavor compounds;

[0048] Figure 4 Schematic diagram of the model building steps;

[0049] Figure 5 Equipment workflow;

[0050] Figure 6 The vintage identification process based on the decision tree vintage identification model;

[0051] Figure 7 Flow chart of the vintage identification program based on the decision tree vintage identification model. DETAILED DESCRIPTION

[0052] The following is a further description of the concept of the present invention and the technical effects produced in conjunction with specific embodiments, so as to fully understand the purpose, features and effects of the present invention. The methods are conventional methods unless otherwise specified. The materials can be obtained from public commercial channels unless otherwise specified. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations of the present invention. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict.

[0053] Example 1 Selection and verification of key compounds

[0054] 1. Collect raw data

[0055] 12 samples of Jinhua ham from different years were obtained. The harvesting time of the Jinhua ham collected in this experiment was fermentation for half a year (3 samples), fermentation for one year (3 samples), fermentation for two years (3 samples), and fermentation for three years (3 samples). The Jinhua ham samples were ground into powder after being frozen with liquid nitrogen, and 2g of Jinhua ham samples were taken in a 20ml sample bottle, and 3g of pure water was added to obtain the sample to be tested. The Jinhua ham samples to be tested from different years were measured using a gas chromatograph-mass spectrometer GC-QQQ, and raw data were collected.

[0056] The chromatographic conditions were as follows: initial temperature 40°C, maintained for 5 min, then heated to 250°C at a rate of 3°C / min and maintained for 10 min. The carrier gas was helium, the gas flow rate was 1.0 mL / min, the split ratio was 5.0, and the injection chamber temperature was 250°C. The voltage of the electron ionization mass spectrometer was 70 eV, the quadrupole scanning range was 35-400 m / z, the ion source temperature was 200°C, the interface temperature was 250°C, the injection volume was 1 μL, and the solvent delay time was 3 min.

[0057] According to the above conditions, each sample was repeated three times, and a total of 36 sets of raw data were obtained.

[0058] 2. Data processing

[0059] The data preprocessing step is to perform normalization processing on the flavor compound quantitative analysis data in the raw data after the raw data collection step. The normalization formula is as follows:

[0060]

[0061] In the above formula, x_new represents the normalized data, x represents the data before normalization, x_min represents the maximum value of the data, and x_min represents the minimum value of the data.

[0062] 3. Preliminary screening of key flavor compounds based on theoretical demonstration

[0063] The key flavor compound initial screening step after data normalization processing specifically includes the following steps: 1) screening out 54 flavor compounds under the condition of OAV>1; 2) performing secondary screening on the 54 flavor compounds to screen out compounds with significant differences in compound concentration content between groups and no significant differences within groups as key flavor compounds for distinguishing the years of Jinhua ham, a total of 18 compounds.

[0064] There are 18 key flavor compounds selected, including: (E)-2-octen-1-ol, 1-nonanol, 1-octen-3-ol, 2,4-decadienal, 2-pentanone, 3-octanone, acetophenone, γ-nonalactone, γ-octanolactone, p-cresol, nonanal, octanol, octanoic acid, isovaleraldehyde, lauric acid, n-heptanol, n-decanoic acid and n-hexanol. Figure 2 The distribution diagram of the vintage samples and the key flavor compounds they contain includes: 1) a bar chart showing the content of each key flavor compound and the year of the sample in which it is located ( Figure 2 A and Figure 2 B); 2) Combine key flavor compounds in pairs and visualize their content distribution in samples from different years ( Figure 2 C).

[0065] 4. Select key discriminant compounds

[0066] The step of selecting two key discriminant compounds applies 1) SHAP feature selection; 2) vintage samples and distribution maps of key flavor compounds contained therein.

[0067] SHAP feature selection is a tool for explaining the predictions of machine learning models. The implementation steps are: 1) train the XGBoost model using the dataset; 2) calculate the SHAP value of each feature; 3) display the SHAP value through visualization methods. The results are as follows Figure 3 As shown, by combining the SHAP feature selection and the distribution results of the vintage samples and the key flavor compounds they contain, the key discriminant compounds were selected: 1-octen-3-ol and n-decanoic acid.

[0068] Example 2 Model building and evaluation

[0069] The construction and evaluation of the identification model include: 1) construction of a 3-year Jinhua ham identification model; 2) evaluation of the Jinhua ham year identification model; 3) verification of key discriminant compounds based on the model evaluation results.

[0070] The model building, evaluation and verification steps are as follows Figure 4 As shown:

[0071] Step 1: Randomly select a small amount of sample data from the two types of samples in each data set in proportion to form a test set, and use the other samples to form a training set. The training set and test set are used for training and testing the model in step 2, respectively.

[0072] Step 2: Train and test several Jinhua ham vintage identification models based on machine learning algorithms, and calculate and record the specific indicators of each model.

[0073] Step 3: Repeat steps 1 and 2 several times.

[0074] Step 4: Calculate the average value of the model-specific indicators to evaluate the Jinhua ham year identification model and verify the key discriminant compounds.

[0075] The data set includes original data, key compound data and key discriminant compound data. The test set, training set and cycle number settings in step 1 and step 3 are shown in Table 1.

[0076] Table 1 Test set, training set and number of cycles settings

[0077]

[0078] The training and testing process described in step 2 is: 1) using the training set data to train the year identification model; 2) calculating and recording the training evaluation indicators of the model; 3) using the test data set to test the year identification model obtained in 1); 4) calculating and recording the test evaluation indicators of the model.

[0079] The machine learning algorithm in step 2 includes: 1) logistic regression algorithm; 2) support vector machine algorithm; 3) naive Bayes algorithm; 4) decision tree algorithm. The specific evaluation index is model accuracy, the calculation formula is as follows, and the relevant parameters are shown in Table 2.

[0080]

[0081] Table 2 Model accuracy related parameters

[0082]

[0083] The key identification compound verification indicators described in step 4 include: 1) identification feasibility; 2) algorithm universality; 3) effectiveness in simplifying high-dimensional complex raw data.

[0084] The model-specific indicator results described in step 4 are shown in Table 3.

[0085] Regarding the verification index 1 (identification feasibility): the experimental accuracy of the selected key discriminant compounds in the year identification model based on logistic regression, support vector machine, naive Bayes and decision tree algorithm is above 90% for both the training set and the test set, and there is no significant decrease in the evaluation index relative to the original data, which verifies the feasibility of the selected key discriminant compounds for the identification of the year of Jinhua ham. In addition, compared with the original data and the selected key flavor compounds, the average accuracy of the selected key discriminant compounds for training and testing is closer, indicating that it may have better generalization ability, which indirectly proves the feasibility of the selected key discriminant compounds for year identification.

[0086] Regarding the verification indicator 2 (algorithm universality): compared with the original data and the experimental results of the selected key flavor compounds, the experimental accuracy of the selected key discriminant compounds in the vintage identification model based on logistic regression, support vector machine, naive Bayes and decision tree algorithms is not only above 90% in both the training set and the test set, but also the average accuracy of each model training and test is closer. This shows that the selected key discriminant compounds are generally applicable to the realization of the vintage classification model of various machine learning algorithms, and the data has strong algorithm universality.

[0087] Regarding the verification indicator 3 (the effectiveness of simplifying high-dimensional complex original data): compared with the overfitting phenomenon obtained from the original data and the selected key flavor compound experiments, the selected key discriminant compounds increased the average test accuracy to more than 90% in the year identification model based on the decision tree algorithm, which narrowed the accuracy indicators obtained from its training and testing, which reflects the effectiveness of the selected key discriminant compounds in simplifying high-dimensional complex original data.

[0088] Based on the evaluation and verification results obtained in step 4, determine the key discriminant compounds, and select the appropriate model based on the actual application scenario for the design of subsequent related equipment.

[0089] Table 3 Model specific indicator results

[0090]

[0091] Embodiment 3 Jinhua ham year identification equipment design

[0092] The working process of the device is as follows Figure 5 As shown: 1) the user input data is the content value of the key discriminant compound selected for the sample to be identified; 2) the year identification model based on the machine learning algorithm includes the year identification model based on the logistic regression algorithm, the year identification model based on the support vector machine algorithm, the year identification model based on the naive Bayes algorithm and the year identification model based on the decision tree algorithm; 3) the output identification result type includes the proportion of samples from the three years in the sample to be tested input by the user and whether the sample has passed the identification.

[0093] The device comprises: a processor, a memory, a display and a computing program running on the processor.

[0094] In the Jinhua ham year identification device calculation program designed according to the analysis results of the embodiment, the year identification process based on the decision tree year identification model is as follows: Figure 6 As shown, the flowchart is as follows Figure 7 shown.

Claims

1. A marker combination for identifying the fermentation year of ham, characterized in that: The marker combination comprises at least two of the following compounds, including (E)-2-octen-1-ol, 1-nonanol, 1-octen-3-ol, 2,4-decadienal, 2-pentanone, 3-octanone, acetophenone, propyl nonanolactone, propyl octanolactone, p-cresol, nonanal, octanol, octanoic acid, isovaleraldehyde, lauric acid, n-heptanol, n-decanoic acid and n-hexanol.

2. The marker combination according to claim 1, characterized in that: The composition comprises at least two of isovaleraldehyde, 1-octen-3-ol, n-decanoic acid, octanoic acid, 2,4-decadienal, (E)-2-octen-1-ol, 3-octanone, 1-nonanol and nonanal.

3. The marker combination according to claim 2, characterized in that: The composition at least comprises n-decanoic acid and 1-octen-3-ol.

4. The marker combination according to claim 3, characterized in that: The composition is used to identify the fermentation year of Jinhua ham; wherein the content of the compound is measured on Jinhua ham samples of different years using a gas chromatography-mass spectrometer GC-QQQ to collect raw data.

5. The marker combination according to claim 4, characterized in that: The chromatographic conditions are as follows: initial temperature 35-42°C, maintained for 3-8 min, then heated to 240-260°C at a rate of 2-3°C / min, maintained for 8-15 min; the carrier gas is helium, the gas flow rate is 0.8-1.2.0 mL / min, the split ratio is 4-6.0, and the injection chamber temperature is 240-260°C; the voltage of the electron ionization mass spectrometer is 65-75 eV, the quadrupole scanning range is 35-400 m / z, the temperature of the ion source is 180-220°C, the interface temperature is 240-260°C, the injection volume is 0.5-1.2 μL, and the solvent delay time is 2-4 min.

6. Use of the composition according to any one of claims 1 to 5 in identifying the fermentation year of ham.

7. A method for screening markers for identifying the fermentation year of ham, characterized in that: The method described is: 1) Obtain ham samples from different years; 2) Using gas chromatography-mass spectrometry (GC-QQQ) to measure the flavor compounds in ham samples of different years and collect raw data; 3) Data processing: normalizing the quantitative analysis data of flavor compounds in the raw data; 4) Preliminary screening of flavor compounds with potential identification functions; 5) Rescreening the compounds obtained in step 4) to obtain key discriminant compounds.

8. The method according to claim 7, characterized in that in, The normalized formula of step 3) is: The normalization formula is as follows: Among them, x_new represents the normalized data, x represents the data before normalization, x_min represents the maximum value of the data, and x_min represents the minimum value of the data.

9. The method according to claim 8, characterized in that The initial screening condition in step 4) is OAV>1; the rescreening condition in step 5) is a.SHAP feature selection; b. Vintage samples and distribution diagram of key flavor compounds they contain; SHAP feature selection is a tool for explaining the predictions of machine learning models. The implementation steps are: 1) train the XGBoost model using the dataset; 2) calculate the SHAP value of each feature; 3) display the SHAP value through visualization methods.

10. Use of the marker combination according to any one of claims 1 to 5 or the marker identified by the method according to claims 7 to 9 in preparing a model or system for identifying the fermentation year of ham.

Images