A method for tracing the origin of dried ginger based on characteristic metabolites

Abstract

The present invention discloses a method for tracing the origin of dried ginger based on characteristic metabolites. The present invention is based on non-targeted metabolomics analysis technology, and uses ultra-high performance liquid chromatography tandem mass spectrometry to detect metabolites in dried ginger from different origins, obtains overall metabolite data of dried ginger from different origins, and further applies chemometrics and machine learning algorithms to construct a classification model for tracing the origin of dried ginger, and screens characteristic metabolites that affect the distinction of dried ginger from different origins, so as to achieve accurate tracing of the origin of dried ginger. The method for tracing the origin of dried ginger based on characteristic metabolites of the present invention has the advantages of accurate and reliable experimental results, stable tracing models, etc., and can provide a new idea for identifying the origin of dried ginger.

Description

Technical Field

[0001] The present invention relates to the field of origin identification of traditional Chinese medicines, and in particular to a method for tracing the origin of dried ginger based on characteristic metabolites. Background Art

[0002] Dried ginger is the dried rhizome of ZingiberoficinaleRoscoe, a plant of the ginger family. It has the effects of warming the middle and dispelling cold, restoring yang and unblocking meridians, warming the lungs and transforming fluid. It is often used to treat cold pain in the abdomen, vomiting and diarrhea, cold limbs and weak pulse, asthma and cough caused by cold fluid, and is an important component of many classic clinical prescriptions. Modern research shows that dried ginger mainly includes gingerols, volatile oils, and diphenylheptanes, which have a series of biological activities such as antioxidant, antiemetic, anti-tumor, anti-inflammatory, and antipyretic.

[0003] Dried ginger is widely distributed in my country, with the main production areas being Sichuan, Yunnan, Shandong and other places. Traditionally, it is believed that the authentic production areas of dried ginger are Muchuan, Qianwei and surrounding areas in Sichuan, with the best quality, also known as Sichuan dried ginger. In recent years, with the expansion of ginger cultivation areas across the country, emerging production areas such as Henan, Jiangxi, Hunan, and Hubei have also begun to plant large areas. Fresh ginger from different origins is easily mixed after being processed into dried ginger, resulting in the phenomenon of other origins pretending to be Sichuan dried ginger in the market. Therefore, it is particularly important to conduct research on the origin traceability of dried ginger.

[0004] At present, existing scientific literature points out that the use of near-infrared spectroscopy technology can preliminarily distinguish dried ginger from different origins, but the shortcoming is that near-infrared spectroscopy can only reflect the vibration information of chemical bonds of compounds in specific bands, and cannot fully obtain the overall composition information of compounds in dried ginger. As a systematic and comprehensive analytical method, non-targeted metabolomics technology can simultaneously perform qualitative and semi-quantitative analysis of hundreds or even thousands of known and unknown metabolites, and can achieve rapid and accurate detection of the composition of compounds in samples. At present, there are no reports on the screening of characteristic metabolites based on non-targeted metabolomics technology for the traceability of the origin of dried ginger. Therefore, in order to overcome the shortcomings of existing methods, it is very necessary to establish a method for identifying the origin of dried ginger based on characteristic metabolites. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to provide a method for tracing the origin of dried ginger based on characteristic metabolites. The method is based on non-targeted metabolomics analysis technology and utilizes ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry to collect metabolite mass spectrometry data of dried ginger from different origins. By applying chemometrics and machine learning algorithms, a classification model for tracing the origin of dried ginger is constructed, and characteristic metabolites that affect the distinction of dried ginger from different origins are screened to achieve accurate tracing of the origin of dried ginger.

[0006] Technical solution: In order to achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for tracing the origin of dried ginger based on characteristic metabolites, comprising the following steps:

[0008] (1) Ultra-high performance liquid chromatography-tandem mass spectrometry was used to detect dried ginger samples from different origins to obtain metabolite mass spectrometry data of different samples, and all metabolite data were processed using mass spectrometry data processing software.

[0009] (2) Chemometrics and machine learning classification algorithms were used to analyze the metabolite data of dried ginger samples from different origins, a traceability model for the origin of dried ginger was established, and the main differential metabolites of dried ginger from different origins were screened out as characteristic metabolites for origin classification.

[0010] As a preferred embodiment, in the above-mentioned method for tracing the origin of dried ginger, the method for preparing the dried ginger sample to be tested in step (1) is: accurately weigh the sample powder, accurately add methanol with a volume concentration of 50% to 80%, seal it, weigh it, perform ultrasonic extraction for 30 to 50 minutes, weigh it again, make up for the lost weight, shake it well, and filter it through a 0.22μm microporous filter membrane to obtain the solution to be tested. The QC sample is prepared by mixing equal amounts of extracts from all samples.

[0011] As a preferred embodiment, in the above-mentioned method for tracing the origin of dried ginger, the conditions for ultra-high performance liquid chromatography determination in step (1) are: chromatographic column: reverse phase C18 chromatographic column, mobile phase: a mixed solution consisting of an aqueous phase and an organic phase, flow rate: 0.4-0.6 mL / min; mass spectrometry conditions are: using an electrospray ionization ion source (ESI), the scanning mode is a positive ion mode, the capillary voltage: 3.0-5.0 kV, the cone voltage: 30-50 V, the ion source temperature: 140-160°C, the desolvation gas flow and temperature: 400-600 L / h and 400-500°C, and the mass scanning range: 50-1500 m / z.

[0012] As a more preferred embodiment, in the above-mentioned method for tracing the origin of dried ginger, the chromatographic column model is: ACQUITYTM PREMIER HSS T3 chromatographic column, the aqueous phase (A) is 0.1% formic acid aqueous solution, the organic phase (B) is acetonitrile solution, gradient elution: 0-5 min, 95% A→70% A; 5-11 min, 70% A→30% A; 11-17 min, 30% A→0% A; 17 min-18 min, 0% A; 18-20 min, 0% A→95% A; 20-21 min, 95% A, flow rate: 0.5 mL / min.

[0013] As a preferred embodiment, in the above-mentioned method for tracing the origin of dried ginger, the characteristic metabolites used for origin differentiation in step (2) are: 6-dehydro-7-(4-hydroxy-3-methoxyphenyl)-1-phenyl-4-heptene-3-one, 5-dehydro-1,7-bis(4-hydroxyphenyl)heptane-4-ethylenediamine-3-one, vanillin, 5-(6-amino-9H-purine-9-yl)-1-(4-hydroxy-3-methoxyphenyl)decane-3-one, ethyl p-coumarate, 1,7-bis(3,4-dihydroxyphenyl)heptane-3,5-diacetate, 5-dehydro-1,7-bis(4-dehydro-3-methoxyphenol)heptane-4-ethylenediamine-3-one, gingerol B, 6-gingerol, cinnamaldehyde, 5-dehydro-7-(4-dehydro- 3-methoxyphenol)-1-phenyl-3-one, methyl-6-gingerol, 7-shogaol, dehydro-6-gingerol, safrole, 8-gingerol, 8-shogaol, 4-((2E,6E)-3,7-dimethyl-2,6-diene-1-yl)-2-methoxyphenol, dehydro-8-gingerol, dehydro-6-gingerone, 10-shogaol, 10-gingerol, methyl-8-gingerol, 4-(2-hexyl-6-methyl-4H-pyran-4-yl)-2-methoxyphenol, dehydro-8-gingerone, isokaurenoic acid, dehydro-14-gingerone, dehydro-12-gingerol, 8-gingerone phenol, dehydro-10-gingerone, 1-dehydro-10-gingerone, α-curcumene, dehydro-12-gingerone, dehydro-10-gingerol, p-cymene.

[0014] As a preferred embodiment, in the above-mentioned method for tracing the origin of dried ginger, the chemometric method in step (2) is partial least squares discriminant analysis (PLS-DA).

[0015] As a more preferred embodiment, in the above-mentioned ginger origin tracing method, the characteristic metabolites differentiated by the origin are screened by the variable weight value (VIP) of the PLS-DA model fitting, and the best fitting parameter of the discriminant model is R 2 X=0.696,R 2 Y=0.927,Q 2 (cum)=0.836, the screening conditions for metabolites were VIP>1 and significance p<0.05.

[0016] As a preferred embodiment, in the above-mentioned method for tracing the origin of dried ginger, the machine learning algorithm in step (2) is K-nearest neighbor (KNN), support vector machine (SVC), decision tree (DT), random forest (RF) and gradient boosting (GB) algorithm.

[0017] As a more preferred option, in the above-mentioned method for tracing the origin of dried ginger, the reliability of the classification algorithm is measured by the area under the curve (AUC) value, and the corresponding AUC values ​​of K-nearest neighbor, support vector machine, decision tree, random forest and gradient boosting algorithm are 0.99, 1.0, 0.95, 1.0 and 0.98 respectively.

[0018] As a preferred embodiment, in the above-mentioned method for tracing the origin of dried ginger, in step (1), dried ginger samples from seven origins, namely Sichuan, Yunnan, Shandong, Henan, Hunan, Hubei and Jiangxi, were collected.

[0019] As a preferred embodiment, in the above-mentioned method for tracing the origin of dried ginger, the ultra-high performance liquid chromatography tandem mass spectrometry in step (1) is an ultra-high performance liquid chromatography tandem quadrupole time-of-flight mass spectrometry, and the mass spectrometry data processing software is Masslynx software.

[0020] Beneficial effects: The origin tracing method of dried ginger based on characteristic metabolites provided by the present invention has the following advantages compared with the prior art:

[0021] The present invention realizes the accurate distinction of dried ginger from different origins based on the potential metabolite information of dried ginger from different origins, combined with chemometrics and machine learning algorithms. The application of partial least squares discriminant analysis and five machine learning algorithms can not only extract effective information from complex metabolic spectra, but also screen characteristic metabolites that affect the distinction between origins, providing indicative indicators for the origin tracing of dried ginger.

[0022] The method described in the present invention uses non-targeted metabolomics technology to analyze the metabolite composition of dried ginger from different origins. Non-targeted metabolomics technology can simultaneously perform qualitative and semi-quantitative analysis on hundreds or even thousands of known and unknown metabolites, thereby achieving rapid and accurate detection of the compound composition in dried ginger samples from different origins, and providing methodological support for the accurate tracing of the origin of dried ginger. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is an overlay of the base peak ion current (BPI) of dried ginger from different origins measured by ultra-performance liquid chromatography-tandem quadrupole time-of-flight mass spectrometry.

[0024] Figure 2 PLS-DA model score graph (a) and permutation test graph (b) for distinguishing dried ginger from different origins.

[0025] Figure 3 This is a chart showing the VIP score results for distinguishing dried ginger from different origins.

[0026] Figure 4 ROC curves (a) and confusion matrix diagrams (b) of five machine learning algorithms. DETAILED DESCRIPTION

[0027] The present invention is further explained below in conjunction with specific examples. It should be understood that these examples are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent modifications of the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0028] Example 1: A method for tracing the origin of dried ginger based on characteristic metabolites

[0029] 1. Collect fresh ginger samples from seven production areas including Sichuan, Yunnan, Shandong, Henan, Hunan, Hubei and Jiangxi, clean them, cut them into thick slices of about 3-4 mm, and dry them in a forced air drying oven at 60°C. All dried medicinal materials are crushed and passed through a No. 3 sieve for later use. Accurately weigh about 0.5 g of dried ginger sample powder from different production areas, place it in a 50 mL stoppered conical flask, accurately add 20 ml of methanol with a volume concentration of 75%, seal it, weigh it, extract it by ultrasound (250 W, 40 kHz) for 40 min, weigh it again, make up for the weight loss, shake it evenly, centrifuge it at 13000 r / min for 10 min, take the supernatant and filter it through a 0.22 μm microporous filter membrane to obtain the filtrate, which is the test solution. The QC sample is prepared by mixing the extracts of all samples in equal amounts.

[0030] 2. Ultra-high performance liquid chromatography-tandem quadrupole time-of-flight mass spectrometry was used to collect mass spectrometric information of dried ginger samples from different origins. The conditions used were as follows:

[0031] Chromatographic conditions: chromatographic column: ACQUITYTM PREMIER HSS T3 chromatographic column (100mm×2.1mm, 1.8μm); mobile phase: phase A is 0.1% formic acid aqueous solution, phase B is acetonitrile solution, gradient elution: 0-5min, 95%A→70%A; 5-11min, 70%A→30%A; 11-17min, 30%A→0%A; 17min-18min, 0%A; 18-20min, 0%A→95%A; 20-21min, 95%A; flow rate is 0.4mL / min.

[0032] Mass spectrometry conditions: electrospray ionization (ESI) source, positive ion mode, capillary voltage: 4.0 kV, cone voltage: 30 V, ion source temperature: 150 °C, desolvation gas flow and temperature: 500 L / h and 400 °C, mass scanning range: 100-1000 m / z.

[0033] Data acquisition was performed in real time using Waters' Masslynx software to obtain the BPI overlay of dried ginger metabolites from different origins. Figure 1 As shown, Masslynx software was used to perform peak extraction, peak matching, peak identification and other preprocessing on the raw data obtained by mass spectrometry for subsequent modeling and analysis.

[0034] 3. Partial least squares discriminant analysis (PLS-DA) was used to analyze the metabolite data of dried ginger samples from different origins, and the PLS-DA origin differentiation model was obtained. The VIP score was obtained through the PLS-DA model, and the main differences in distinguishing dried ginger from different origins were screened out as characteristic metabolites of origin.

[0035] Figure 2 The score graph (a), permutation test graph (b) and Figure 3 The VIP score result diagram shows that the dried ginger samples from the seven origins can be clearly distinguished. Among them, the authentic dried ginger produced in Sichuan, Yunnan and Shandong can be independently classified into clusters. The cumulative contribution of the model is R 2 X=0.696,R 2 Y = 0.927, goodness of fit Q 2 (cum)=0.836, and the permutation test results show that Q 2 <0.05, indicating that the established model did not overfit and was accurate and reliable. From the differentiation results, the PLS-DA model is suitable for tracing the origin of dried ginger from different sources, and the main metabolites that affect the differences between origins can be screened according to the VIP score results.

[0036] According to the VIP values ​​of metabolites provided by the PLS-DA model, metabolites with VIP>1.0 and significant difference p<0.05 were selected as characteristic metabolites for distinguishing dried ginger from different origins. By comparing the database with the existing literature, a total of 35 different metabolites were screened out among the origins (see Table 1). The identified differences mainly included 6-dehydro-7-(4-hydroxy-3-methoxyphenyl)-1-phenyl-4-heptene-3-one, 5-dehydro-1,7-di(4-hydroxyphenyl)heptane-4-ethylenediamine-3-one, vanillin, 5-(6-amino-9H-purine-9-yl)-1-(4-hydroxy-3-methoxyphenyl)decane-3-one, ethyl p-coumarate, 1,7-di(3,4-dihydroxyphenyl)heptane-3,5-diacetate, 5-dehydro-1,7-di(4-dehydro-3-methoxyphenol)heptane-4-ethylenediamine-3-one, gingerol, cinnamaldehyde, 5-dehydro-7-(4-hydroxyphenyl)heptane-3,5-diacetate, gingerol B, 6-gingerol, cinnamaldehyde, 5-dehydro-7-(4- dehydrogenated-3-methoxyphenol)-1-phenyl-3-one, methyl-6-gingerol, 7-gingerol, dehydrogenated-6-gingerol, safrole, 8-gingerol, 8-gingerol, 4-((2E,6E)-3,7-dimethyl-2,6-diene-1-yl)-2-methoxyphenol, dehydrogenated-8-gingerol, dehydrogenated-6-gingerone, 10-gingerol, 10-gingerol, methyl The 1-dehydro-10-gingerol, α-curcumene, dehydro-12-gingerol, dehydro-10-gingerol, and p-cymene are particularly characteristic. In particular, 5-(6-amino-9H-purin-9-yl)-1-(4-hydroxy-3-methoxyphenyl)decane-3-one, dehydro-10-gingerol, dehydro-12-gingerol, 8-gingerol, safrole, and methyl-8-gingerol have remarkable characteristics.

[0037] Table 1 Characteristic metabolites of dried ginger produced by different regions

[0038]

[0039]

[0040] 4. Using the pre-processed mass spectrometry data of dried ginger from 7 origins, a machine learning algorithm was used to further establish a prediction model for the origin of dried ginger, and to achieve accurate traceability of the origin of dried ginger. The programming language used in the machine learning algorithm is Python (version 3.8), the programming tool is Jupyter Notebook, and the machine learning programming library is scikit-learn (version 1.0.2). The robust_scale function in the sklearn.Preprocessing module was used to standardize the data, and all dried ginger samples were randomly divided into a training sample set (70%) and a test sample set (30%). The training sample set was trained and constructed using K-nearest neighbor (KNN), support vector machine (SVC), decision tree (DT), random forest (RF) and gradient boosting (GB) algorithms, and the classification and discrimination indicators such as area under the curve (AUC) and accuracy were statistically calculated from the test sample set.

[0041] ROC curves of five classification learning algorithms under training sets ( Figure 4 a), the AUC values ​​of KNN, SVC, DT, RF, and GB were calculated to be 0.99, 1.0, 0.95, 1.0, and 0.98, respectively, indicating that the algorithm model established based on the training samples has excellent performance in origin classification. Further, 30% of the test samples were used for verification analysis, and the overall discrimination accuracy of the five classification algorithms for prediction samples of dried ginger from different origins was 97.8%, 97.8%, 91.4%, 97.8%, and 87.2%, respectively (corresponding to KNN, SVC, DT, RF, and GB).

[0042] Confusion matrix diagrams of different algorithms (see Figure 4 b) shows the validation set classification prediction results of dried ginger from 7 different origins. From the overall trend, KNN, SVC, and RF algorithms only misjudged the origin of one sample, specifically, the Hunan origin was misjudged as the Henan origin. Although the DT and GB algorithms misjudged several samples, there was no phenomenon that Sichuan dried ginger from the authentic origin was misjudged as other origins, or other origins were misclassified as authentic origins.

[0043] K-nearest neighbor, support vector machine, decision tree, random forest and gradient boosting algorithms showed excellent discriminant performance in the construction of the dried ginger origin traceability model, with discrimination accuracy rates greater than 85%. In particular, the accuracy rates of KNN, SVC and RF reached 97.8%, which can be used to fully extract effective information on secondary metabolites of dried ginger from different origins for origin classification, realize the accurate traceability of the origin of dried ginger, and provide an important reference for the authenticity evaluation of authentic dried ginger medicinal materials.

[0044] The above experimental results show that the method for tracing the origin of dried ginger based on characteristic metabolites combined with chemometrics and machine learning algorithms provided by the present invention has the advantages of accurate experimental results and high feasibility. The non-targeted metabolomics technology comprehensively analyzes the secondary metabolites of dried ginger from different origins, integrates chemometrics to accurately distinguish the origin of dried ginger, and screens out characteristic metabolites that affect the distinction between origins; further combined with 5 machine learning classification algorithms, a dried ginger origin tracing model was successfully constructed. The proposal of the present invention has important reference significance for tracing the origin of dried ginger and other bulk Chinese medicinal materials.

[0045] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for tracing the origin of dried ginger based on characteristic metabolites, characterized in that: The following steps are involved: (1) Ultra-high performance liquid chromatography-tandem mass spectrometry was used to detect dried ginger samples from seven production areas, including Sichuan, Yunnan, Shandong, Henan, Hunan, Hubei and Jiangxi, to obtain metabolite mass spectrometry data from the seven production areas. All metabolite data were processed using mass spectrometry data processing software; (2) The metabolite data of dried ginger samples from different origins were analyzed using chemometrics and machine learning classification algorithms, a traceability model for dried ginger origin was established, and the main differential metabolites of dried ginger from different origins were screened out as characteristic metabolites for origin differentiation; Step (1) The chromatographic column model of the ultra-high performance liquid chromatography is: ACQUITYTM PREMIER HSS T3 chromatographic column, the aqueous phase A is a 0.1% formic acid aqueous solution, the organic phase B is an acetonitrile solution, the gradient elution is: 0-5 min, 95% A→70% A; 5-11 min, 70% A→30% A; 11-17 min, 30% A→0% A; 17 min-18 min, 0% A; 18-20 min, 0% A→95% A; 20-21 min, 95% A, flow rate: 0.5 mL / min; The characteristic metabolites used for producing area differentiation in step (2) are: 6-dehydro-7-(4-hydroxy-3-methoxyphenyl)-1-phenyl-4-heptene-3-one, 5-dehydro-1,7-di(4-hydroxyphenyl)heptane-4-ethylenediamine-3-one, vanillin, 5-(6-amino-9H-purine-9-yl)-1-(4-hydroxy-3-methoxyphenyl)decane-3-one, ethyl p-coumarate, 1,7-di(3,4-dihydroxyphenyl)heptane-3,5-diacetate, 5-dehydro-1,7-di(4-dehydro-3-methoxyphenol)heptane-4-ethylenediamine-3-one, gingerol B, 6-gingerol, cinnamaldehyde, 5-dehydro-7-(4-dehydro-3-methoxyphenol)-1- Phenyl-3-one, methyl-6-gingerol, 7-shogaol, dehydro-6-gingerol, safrole, 8-gingerol, 8-shogaol, 4-((2E,6E)-3,7-dimethyl-2,6-diene-1-yl)-2-methoxyphenol, dehydro-8-gingerol, dehydro-6-gingerone, 10-shogaol, 10-gingerol, methyl-8-gingerol, 4-(2-hexyl-6-methyl-4H-pyran-4-yl)-2-methoxyphenol, dehydro-8-gingerone, isokaurenoic acid, dehydro-14-gingerone, dehydro-12-gingerol, 8-gingerone, dehydro-10-gingerone, 1-dehydro-10-gingerone, α-curcumene, dehydro-12-gingerone, dehydro-10-gingerol and p-cymene.

2. The origin tracing method of dried ginger based on characteristic metabolites according to claim 1, characterized in that: The mass spectrometry conditions of step (1) are as follows: using an electrospray ionization ion source ESI, scanning mode is positive ion mode, capillary voltage: 3.0~5.0kV, cone voltage: 30~50V, ion source temperature: 140~160℃, desolvation gas flow rate and temperature: 400~600 L / h and 400~500℃, mass scanning range: 50~1500 m / z.

3. The method for tracing the origin of dried ginger based on characteristic metabolites according to claim 1, characterized in that: The chemometric method in step (2) is partial least squares discriminant analysis.

4. The method for tracing the origin of dried ginger based on characteristic metabolites according to claim 3, characterized in that: The characteristic metabolites differentiated by the production area were screened by the variable weight value VIP of the PLS-DA model fitting, and the best fitting parameter of the discriminant model was R 2 X = 0.696, R 2 Y = 0.927, Q 2 (cum) = 0.836, and the screening criteria for metabolites were VIP > 1 and significance p < 0.

05.

5. The method for tracing the origin of dried ginger based on characteristic metabolites according to claim 1, characterized in that: The machine learning classification algorithms in step (2) are K-nearest neighbor, support vector machine, decision tree, random forest and gradient boosting algorithm.

6. The method for tracing the origin of dried ginger based on characteristic metabolites according to claim 5, characterized in that: The reliability of the machine learning classification algorithm is measured by the area under the curve AUC value. The corresponding AUC values ​​of K-nearest neighbor, support vector machine, decision tree, random forest and gradient boosting algorithm are 0.99, 1.0, 0.95, 1.0 and 0.98 respectively.

7. The method for tracing the origin of dried ginger based on characteristic metabolites according to claim 1, characterized in that: The ultra-high performance liquid chromatography tandem mass spectrometry in step (1) is ultra-high performance liquid chromatography tandem quadrupole time-of-flight mass spectrometry, and the mass spectrometry data processing software is Masslynx software.

Images