JP2009014700A - Green tea quality prediction method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for easily and accurately predicting the quality of green tea, particularly the conventionally difficult overall quality of product green tea. <P>SOLUTION: This green tea quality prediction method includes processes for pretreating green tea to obtain an assay sample, presenting the assay sample to an instrumental analysis to obtain an analysis result, converting the analysis result into numeric data to make a multivariate analysis, and predicting quality from the obtained analysis result. It is preferable to make a PLS regression analysis as the multivariate analysis on the instrumental analysis result of a plurality of green tea of known quality to prepare a quality prediction model, and the analysis result on the green tea of unknown quality is collated with the quality prediction model. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a method for predicting the quality of green tea. More specifically, the present invention relates to a quality prediction method by instrumental analysis of metabolites or degradation products of green tea.

  Tea is the most popular beverage and is made from the leaves of the plant Camellia sinensis. Based on the length of the oxidation reaction and the leaf processing method, it is classified into three types: black tea, oolong tea, and green tea. Among these, green tea is not oxidized and not fermented because oxidase is inactivated by heating. Therefore, the taste of green tea is determined by the type of tea tree, plucking time, cultivation method, and the like.

  Currently, the final quality of green tea is determined by a sensory test based on leaf appearance, aroma, color, and taste by a skilled worker. It takes years of experience to acquire these skills. In addition, there is a drawback that objectivity and reproducibility are low. Therefore, a simple mechanized quality prediction method is necessary for manufacturing process management and product quality management. Until now, methods using the near-infrared spectroscopy (Patent Documents 1 and 2) and methods based on the properties of tea leaves (Patent Documents 3 to 5) have been disclosed as green tea quality prediction methods.

  Patent Document 1 describes a method for measuring the quality of tea leaves, particularly raw tea leaves before processing, using visible light and near infrared rays. It is described that the color of tea leaves is measured from visible light, and the components of tea leaves are measured from near-infrared light, but what is specifically measured from near-infrared light and how to predict quality is described. Not. Patent Document 2 describes a method for evaluating the quality by determining the content of free amino acids and the content of neutral detergent fibers in tea leaves using near infrared spectroscopy. However, these methods based on near-infrared spectroscopy are effective for quality prediction of fresh leaves, but have little track record in quality prediction of product green tea.

Patent Documents 3 to 5 describe that quality is evaluated by measuring properties such as water content, fiber content, nitrogen content, and appearance of fresh tea leaves. However, a method based on properties such as water content is simple, but lacks accuracy.
JP 2000-346797 A JP-A-8-114543 JP 2000-228950 A JP 11-313608 A Japanese Patent Laid-Open No. 9-224573

  Chemical analysis appears to be the most reliable method for assessing the quality of green tea. The combination of instrumental analysis methods and powerful computer-driven pattern recognition technology offers new possibilities for quality control and characterization of complex materials. Therefore, an object of the present invention is to provide a method for easily predicting the quality of green tea, in particular, the overall quality of product green tea, which has been difficult in the past.

  The taste of green tea is attributed to a unique theanine occupying about 60-70% amino acids in amino acids, especially tea leaves. Its astringent taste is due to catechin (tannin) and its bitterness is due to caffeine. In addition, the volatile compound mainly contributes to the difference in tea fragrance. Due to cultivar differences, environmental effects, processing methods, etc., the apparent steady state amount of intermediate pathway metabolites and / or terminal accumulation of final metabolites also varies. Moreover, the decomposition product by heat processing also affects the taste of green tea. Therefore, monitoring the metabolites and degradation products in tea leaves both spatially and temporally is important for quality assessment.

  In chromatography, fingerprinting based on pattern recognition methods is the process monitoring and control (raw material grading, routine online quality check, or product manufacturing process) in various analytical areas such as food and nutrition. ), And geographic origin (measurement of the source by the chemical composition of the ingredients, tracking of the origin of the final product by flavor and aroma components).

  Thus, the present invention provides a limited number of individual groups of specific groups that are considered important for specific functions, such as antioxidant compounds for medical purposes, and catechins and other phenolic compounds for taste and aroma. It was completed based on the ability to analyze green tea comprehensively and quickly by focusing on metabolites and degradation products instead of characterization of these compounds and combining them with chemical fingerprints and multivariate analysis. .

The present invention provides a method for predicting the quality of green tea, the method comprising:
Pre-treating green tea to obtain an analytical sample;
Subjecting the analysis sample to instrumental analysis to obtain an analysis result;
A step of converting the analysis result into numerical data and performing multivariate analysis; and a step of predicting quality from the obtained analysis result.

The present invention also pre-processes a plurality of green teas of known quality to obtain individual analytical samples;
Subjecting the individual analysis samples to instrumental analysis to obtain individual analysis results; and converting the relationship between the individual analysis results and the quality into numerical data for multivariate analysis;
To provide a green tea quality prediction model.

  In one embodiment, the multivariate analysis is a PLS regression analysis.

  In a preferred embodiment, the pretreatment includes a step of pyrolyzing the green tea, and the instrumental analysis is a combination of gas chromatography and mass spectrometry. In a more preferred embodiment, the analysis result is quality predicted based on the retention time.

  In another preferred embodiment, the pretreatment includes a step of extracting a hydrophilic compound from the green tea to obtain an extract.

  In a more preferred embodiment, the instrumental analysis is a combination of gas chromatography and mass spectrometry, and the pretreatment further comprises silylating the extract.

  In a more preferred embodiment, among the above analysis results, arabinose, glutamine, citric acid, glutamic acid, aspartic acid, inositol, shikimic acid, phosphoric acid, fructose, theanine, malic acid, caffeine, quinic acid, glucose, mannose , Ribose, and sucrose data are analyzed.

  In another preferred embodiment, the instrumental analysis comprises a combination of liquid chromatography and mass spectrometry (particularly a combination of high performance liquid chromatography and mass spectrometry, or a combination of ultra high performance liquid chromatography and mass spectrometry), Alternatively, nuclear magnetic resonance analysis (particularly, Fourier transform nuclear magnetic resonance analysis).

  In another preferred embodiment, the pretreatment includes a step of pulverizing the green tea to obtain a powder, and a step of pasting the powder with a solvent.

  In a more preferred embodiment, the instrumental analysis is infrared spectroscopy or near infrared spectroscopy (in particular, Fourier transform infrared spectroscopy or Fourier transform near infrared spectroscopy).

In a further preferred embodiment, the instrumental analysis is Fourier transform near-infrared spectroscopic analysis, and in the multivariate analysis step, the result of 5500-5200 cm ^-1 is converted and converted in the analysis result. The data is further second-order differentiated.

  In another preferred embodiment, the analysis result is a combination of gas chromatography and mass spectrometry, a combination of high performance liquid chromatography and mass spectrometry, a combination of ultra high performance liquid chromatography and mass spectrometry, Fourier transform infrared. The analysis results of at least two instrumental analyzes selected from the group consisting of near infrared spectroscopy and Fourier transform nuclear magnetic resonance analysis.

The present invention further provides a method for predicting the quality of green tea, the method comprising pretreating green tea to obtain an analytical sample;
Subjecting the analysis sample to instrumental analysis to obtain an analysis result;
Converting the analysis result into numerical data and performing multivariate analysis; and comparing the obtained analysis result with any of the above-described quality prediction models.

  According to the present invention, it is possible to predict the overall quality of product green tea, which has been difficult in the past, with a more accurate and simple method.

  Green tea is a non-fermented tea in which fresh tea leaves are heated to lose enzyme activity, prevent oxidation of ingredients, and maintain a green color. The green tea that is subject to quality prediction in the present invention may be raw tea leaves before processing, but preferably green tea that has been processed as green tea such as rough tea or finished tea (ie, green tea product). Leaves. In the present invention, more preferably, the product green tea is the target of quality prediction.

  In the present invention, green tea is appropriately pretreated according to the instrumental analysis described below, and then subjected to various instrumental analyzes.

  In the present invention, instrumental analysis refers to analysis / measurement means using an analytical instrument, and includes gas chromatography (GC), liquid chromatography (LC) (for example, high performance liquid chromatography (HPLC), ultrahigh performance liquid chromatography ( UPLC), mass spectrometry (MS), infrared spectroscopy (IR) (eg, Fourier transform infrared spectroscopy (FT-IR)), near infrared spectroscopy (NIR) (eg, Fourier transform near infrared spectroscopy) Analysis (FT-NIR)), nuclear magnetic resonance analysis (NMR) (for example, Fourier transform nuclear magnetic resonance analysis (FT-NMR)) and the like. These instrumental analyzes may be combined, and examples thereof include GC / MS, LC / MS (particularly HPLC / MS, UPLC / MS) and the like. The apparatus used for these instrumental analysis is not specifically limited, If the metabolite (for example, amino acid, organic acid, sugar) contained in green tea can be measured, the apparatus normally used is used. Can be. Measurement conditions can be appropriately set by those skilled in the art so as to be appropriate for the measurement of these substances.

  The pretreatment is performed according to the instrumental analysis as described above in order to make the substance to be analyzed in the green tea into a form suitable for the instrumental analysis. Examples of the pretreatment include treatments such as drying, cutting, pulverization, and extraction. For example, the pulverization can be performed using an appropriate instrument such as a blender or a ball mill. The extraction can be performed using water, an organic solvent, or a mixture of these solvents. Examples of the organic solvent that can be used for extraction include methanol, ethanol, n-propanol, isopropanol, acetone, chloroform, and the like. As the extraction operation, green tea may be heated in water and / or an organic solvent to extract the pyrolyzate. The heating temperature can be appropriately determined according to the solvent used, and is usually a temperature near the boiling point of the solvent under normal pressure. The heating time is also set appropriately. These unit operations are singly or combined to set appropriate preprocessing conditions.

  For example, when performing GC as instrumental analysis, the pretreatment preferably further includes a step of silylating an extract from green tea. The silylation can be performed using a silylation reagent for GC that is usually used by those skilled in the art. Alternatively, as a pretreatment, for example, green tea that has been crushed and then subjected to only thermal decomposition treatment can be a sample to be analyzed. In this case, the pyrolyzate can be directly introduced into the GC without extraction. Thermal decomposition can be performed using a commercially available thermal decomposition apparatus.

  When IR / NIR is performed as instrumental analysis, the pretreatment preferably includes a step of pasting green tea powder with a solvent. The step of obtaining the powder is as described above. As the solvent for making a paste, for example, a solvent that is usually selected according to the sample is used for preparing a sample for IR measurement.

  In the case where NMR is performed as the instrumental analysis, the pretreatment preferably includes a step of obtaining an extract from green tea with water and / or a hydrophilic solvent. Examples of the hydrophilic organic solvent include methanol, ethanol, n-propanol, isopropanol, acetone and the like.

  In the case where LC, particularly HPLC, is performed as the instrumental analysis, the pretreatment preferably includes a step of obtaining an extract from green tea with water and / or a hydrophilic solvent. Examples of the column to be used include an affinity column, a reverse phase column, and an ion exchange column. HPLC conditions (for example, flow rate, detector, mobile phase, etc.) are appropriately selected according to the sample.

  In the case where ultra high performance liquid chromatography (UPLC) is performed as the instrumental analysis, the pretreatment preferably includes a step of extracting green tea with water and / or a hydrophilic solvent to obtain an extract. UPLC passes a high linear velocity mobile phase through small column particles to achieve fast analysis speed, high resolution, and high sensitivity. Column particles used in UPLC are column particles having a smaller diameter (eg, 1.7 μm) than column particles (eg, 5 μm or 3.5 μm) used in HPLC, such as affinity columns, reverse phase columns, and An ion exchange column may be mentioned. The UPLC conditions (for example, flow rate, detector, mobile phase, etc.) are appropriately selected depending on the sample.

  The green tea sample subjected to the above pretreatment is subjected to arbitrary instrumental analysis to obtain an analysis result. The obtained analytical result may be a fingerprint of a green tea sample. Multi-variate analysis is performed by converting this fingerprint into numerical data. Examples of results (variables) obtained by analysis include retention time, wavelength (or wave number), and signal data (or ionic strength), spectral data such as absorbance. Furthermore, the variable includes the ranking (grade) of green tea samples.

  As the multivariate analysis, an analysis tool usually used in chemometrics is employed for analysis of instrumental analysis data. For example, PCA (principal component analysis), HCA (hierarchical cluster analysis), PLS regression analysis (Projection to Latent Structure), discriminate analysis, etc. Multivariate tools. Furthermore, PLS (Partial Least Square Projection to Latent Structure) by partial least squares is used to confirm the relationship between the two groups of related variables; for example, the relationship between the metabolite of green tea and its quality. If necessary, multivariate analysis may be performed in combination with spectral filtering methods, eg, orthogonal signal correction (OSC) to remove interfering components. Many of these analysis tools are commercially available as software, and arbitrary ones are available. Such a commercially available tool is generally provided with an operation manual so that multivariate analysis can be performed without knowledge of difficult mathematics and statistics.

The multivariate analysis may be performed by selecting a certain range of data important for quality prediction, instead of all the obtained data. For example, when green tea powder is made into a paste and analyzed by FT-NIR, it is preferable to perform PCA on data in the wave number range of 5500 to 5200 cm ⁻¹ .

  The PLS regression analysis suitably employed in the present invention is an effective technique for creating a calibration curve from spectrum data having a correlation between variables (for example, wave number and wavelength). Normally, if there is a correlation between variables, the regression accuracy will be significantly reduced depending on the combination of variables used. To avoid this, PLS converts the variables into variables that are uncorrelated with each other (latent variables). To do regression. That is, PLS is an analysis method in which data variables are orthogonally transformed and (multiple) regression analysis is performed using the new variables.

  Specifically, when a hydrophilic low-molecular-weight metabolite is extracted from a plurality of green tea samples of known qualities or when quantitative analysis is performed by GC / MS on the pyrolyzate of green tea, the obtained GC / MS chromatogram is obtained. Matrix data is created with the retention time index as the independent variable and the mass spectrometry signal intensity as the dependent variable. A ranking prediction model can be created by the PLS method using a ranking of a plurality of green tea samples with known rankings as explanatory variables and a known sample as a training set. For example, multi-variate analysis of data obtained by similar instrumental analysis from green tea samples with unknown quality, and comparing and collating the analysis results with the prediction model, it is possible to know which ranking it is, so quality can be determined. Can be predicted.

  For example, when metabolites are extracted from green tea and analyzed by GC / MS, the analysis results obtained include retention times, mass spectra, ionic strength, and the like of various metabolites. In this case, in particular, the data for arabinose, glutamine, citric acid, glutamic acid, aspartic acid, inositol, shikimic acid, phosphoric acid, fructose, theanine, malic acid, caffeine, quinic acid, glucose, mannose, ribose, and sucrose It plays an important role in quality prediction (grading classification) of green tea. In accordance with the present invention, relatively high amounts of amino acids, quinic acid, phosphate, ribose, and arabinose are present as metabolites in high ranking green tea samples, while fructose, glucose, mannose are present in low ranking green tea samples. It has become clear that sugars such as have a tendency to exist as main metabolites.

  Moreover, the accuracy of the prediction model obtained by the above method can be increased by accumulating data. Therefore, for example, if the tendency of the relative amount of metabolites in green tea becomes clearer, a fingerprint can be obtained from the analysis result of green tea whose quality is unknown, and quality prediction can be performed based on this fingerprint.

(Example 1: Preparation of analytical sample for GC / MS)
The dried leaves after the processing of 53 kinds of most teas ranked at the 2005 competition were used as samples. These tea samples described in the commercial tea fair were obtained from the Nara Prefectural Agricultural Center Tea Industry Promotion Center. Tea ranking was determined by the overall score of the sensory test, which was leaf appearance, aroma, color, and taste of the tea that was appraised by a professional tea appraiser. In addition, the ranking was given from No. 1 to No. 53 in order from the highest ranking.

  First, dried tea leaves (30 mg) were freeze-dried in a 2 mL Eppendorf tube and pulverized with a Retsch ball mill (20 Hz, 1 minute). 1 mL of a mixture of methanol, water, and chloroform (2.5: 1: 1, v / v / v) is added to the tea leaves, and 60 μL ribitol (Wako Pure Chemical Industries, Ltd .: deionized water) is used as an internal standard. Diluted to a concentration of 0.2 mg / mL), shaken for 5 minutes, and centrifuged at 16000 × g for 3 minutes at 4 ° C. 900 μL of the supernatant was then transferred to a 1.5 mL Eppendorf tube. 400 μL of water purified using a Millipore Milli-Q system (Berdford, Mass.) Was added, vortexed and centrifuged, and then 400 μL of the polar phase was added to another 1.5 mL volume capped with a perforated cap. Moved to Eppendorf tube. The resulting extract (hydrophilic compound) was dried in a vacuum centrifuge drier until dry (overnight).

  Next, 50 μL of methoxyamine hydrochloride (Sigma) in pyridine (20 mg / mL) was added to the extract as the first derivatizing agent. The mixture was incubated at 30 ° C. for 90 minutes. Further, 100 μL of N-methyl-N- (trimethylsilyl) trifluoroacetamide (MSTFA: GL sciences Inc.), which is a second derivatizing agent, was added and incubated at 30 ° C. for 30 minutes. Was derivatized to obtain an analytical sample.

(Example 2: Analysis of green tea sample by GC / MS)
1 μL of the analytical sample obtained in Example 1 above was injected into the GC / MS in split mode (25: 1, v / v). The GC / MS apparatus used in this example was a 0.25 μm CP-SIL 8 CB low bleed (Varian Inc. 689CN (Agilent Co.) equipped with a 30 m × 0.25 mm id fused silica capillary column coated with The injection temperature was 230 ° C. The flow rate of helium gas through the column was 1 mL / min. The column temperature was kept isothermal at 80 ° C. for 2 minutes, then increased to 330 ° C. at 15 ° C./minute and kept isothermal for 6 minutes. The transfer line and ion source temperatures were 250 ° C. and 200 ° C., respectively. Ions were generated by 70 kV electron impact (EI) and 20 scans per second were recorded over the mass range of 85-650 m / z. The acceleration voltage was activated after a solvent delay of 250 seconds.

(Example 3: Analysis of data by GC / MS and fingerprinting of green tea metabolites)
For preprocessing of the data obtained by GC / MS in Example 2 above, the obtained raw chromatographic data (Pegasus file, * .peg) is converted into an ANDI file (analysis data exchange protocol, * .cdf). Converted. Using the ANDI format, it was possible to convert and transfer data between different mass spectral data systems. The converted file (ANDI) was subjected to a data preprocessing procedure and the data points were reprocessed in detail. In addition, data conversion was also performed to obtain the best chromatographic data. Total ion chromatograph data was then extracted and saved as an ANDI format without fragment data. These files were transferred to the commercial software LineUp (Informatrix, Inc.) for retention time multiple alignment. Misaligned peaks were aligned using a correlation optimized warping algorithm.

  Subsequently, principal component analysis (PCA) was performed in order to grasp the relationship expressed by similarity or dissimilarity among groups of multivariate data. Commercial software, Pirouette® (Informatrix, Inc.) was applied for this purpose.

  The 53 most ranked tea samples were divided into 3 groups ranked according to their grade, but the classification of these 3 groups was not clear. Since there were about 20 samples in each given group, the variation in each group of samples was so high that the high ranking samples did not show a distinct population separate from the low ranking samples. The low ranking (No. 51) and high ranking (No. 1) samples were then preprocessed again at the mean center without converting the data. As a result, it was found that there was a certain difference between these groups (FIG. 1).

  A prominent compound was identified by comparing its mass spectrum with the spectrum in the library (NIST library and internal library prepared from certified standard chemicals). In addition, a library provided by the Max-Planck Institute of Molecular Plant Physiology, Germany was also used for this purpose (http://www.mpimp-golm.mpg.de/mms-library/index-e.htlm ). The higher ranking samples contained relatively high amounts of amino acids, quinic acid, phosphate, ribose, and arabinose, while the lower ranking samples were primarily sugars (fructose, glucose, and mannose). This indicated that metabolomics analysis could be useful for green tea quality measurement.

  Next, a projection model was created by selecting projection (PLS) (SIMCA-P version 11.0: Umetrics) on the latent structure by the partial least square method. PLS finds a relationship between two sets of variables (measurements and response values). That is, matrix data is created from the obtained GC / MS chromatogram using the retention time index as an independent variable and the mass spectrometry signal intensity as a dependent variable, and the quality ranking ranking of each sample as an explanatory variable and a known sample as a training set. A ranking prediction model was created by the PLS method.

  The relationship between the sample metabolite fingerprint (matrix X) and its quality (matrix Y) was observed. First, the analysis samples were divided into training set and test set groups. Samples with rankings of 2, 12, 22, 32, 42, and 52 were excluded as test sets for model validation. None of the variables were transformed; all of the variables were aggregated and scaled to Pareto variance to reduce chromatographic data noise effects. The completeness of the model, i.e. the number of potential factors in the PLS model, could be determined by cross-validation, and the optimal number was found in the balance between fit (to the model) and predictability. In addition, the PLS model was validated with a test set and the mean square error (RMSEP) of the predicted values was computed. Two significant components were extracted, and according to cross-validation, an 80.7% variation in Y was described (R2Y = 0.807) and a 31.3% variation in Y was predicted (Q2Y = 0.313). (See FIG. 2). The test set was then predicted with the PLS model, and the prediction accuracy of the test sample (RMSEP = 10.23) was obtained for model evaluation based on the training set sample (RMSEP = 6.91) (FIG. 3). A certain degree of prediction was obtained by this PLS model.

  In addition, spectral filtering techniques were applied to enhance the predictive power of the PLS model. OSC is a PLS-based solution that removes X data variability that is not related to Y modeling. In OSC, one component is removed one by one from the matrix X using a non-linear iterative partial least squares (NIPALS) algorithm. By removing the two OSC components from the PLS model, Q2Y increased to 0.500. The total squared residual was 41.40%. Thus, the 58.60% variation in X was not related to Y and was eliminated (see FIG. 4). The predicted value after subjecting the test set to the model is the prediction accuracy for the test sample (RMSEE = 8.13) that is in good agreement with the model evaluation based on the training sample (RMSEP = 6.57) (FIG. 5). . In addition, the highly relevant explanatory variable Y was also identified from the VIP (variable importance in projection) value. Large VIP values above 1 are the most relevant explanatory variable Y. Thus, it was found that quinic acid, amino acids (particularly theanine), and sugars are important for creating a quality prediction model for green tea (FIG. 6).

(Example 4: Green tea metabolism profiling by GC / MS)
Yamato tea, which is accepted as a high-grade tea, was obtained from the Nara Prefectural Agricultural Center Tea Industry Promotion Center. The green tea sample pretreatment method and GC / MS analysis conditions were the same as in Examples 1 and 2 above. Since the chromatogram resolution was satisfactory, the peak area of each metabolite was calculated by total ion count (TIC). Subsequently, all peaks of the identified metabolites were normalized to the internal standard ribitol. From this analysis, it was found that the dried leaves of green tea contain many compounds, mainly organic acids, amino acids and sugars (both monosaccharides and disaccharides) as shown in Table 1 below. The compounds detected in the highest amount were saccharides, especially sucrose and fructose. Theanine, an important amino acid that contributes to the unique taste of tea, was also detected. Several other amino acids could be identified in the extract, but in small quantities. About 23 identified metabolites were identified from the chromatogram.

  This demonstrates that metabolomic analysis using GC / MS in combination with the chemometrics obtained in Example 3 above provides useful information in green tea research.

(Example 5: Preparation and analysis of analytical sample for ¹ H-NMR)
As in Example 1 above, the dried leaves after processing of 53 kinds of most teas ranked (rated) in the 2005 competition were used as samples.

First, dried tea leaves (50 mg) were freeze-dried in a 1 mL Eppendorf tube and pulverized with a Retsch ball mill (20 Hz, 1 minute). To this tea leaf, 1 mL of heavy water (deuterium atom 99.9%: Cambridge Isotope Laboratories, Inc.) was added. This mixture was incubated for 30 minutes at 60 ° C. and 1400 rpm in Thermomixer comfort (Eppendorf) and centrifuged for 10 minutes at 25 ° C. and 16000 × g (centrifuge 5415R: Eppendorf). The supernatant containing the hydrophilic metabolites was filtered through a 0.45 μm PTFE membrane (Advantec). Next, 400 μL of the filtrate was dissolved in 200 μL of 0.2 M phosphate buffer (pH 7.4) containing 3 mM sodium 3- (trimethylsilyl) -1-propane sulfate (DSS: Aldrich) as an internal standard. Sample preparation was performed one day before the ¹ H-NMR measurement.

The obtained analytical sample was measured by ¹ H-NMR. In this example, a spectrum was recorded at 25 ° C. using a 750 MHz Varian Inova 750 spectrometer with a 5 mm ¹ H { ¹³ C / ¹⁵ N} triple resonance indirect detection probe. ¹ H-NMR measurements were performed with 64 transients and 131K composite data points. Acquisition time and recycle delay were 6.257 seconds and 3.743 seconds per scan, respectively, using a 30 ° pulse angle. A pre-saturation pulse sequence was applied to suppress the residual water signal. All spectra were Fourier transformed with a 0.1 Hz line broadening prior to data processing.

Example 6: Analysis of ¹ H-NMR data, fingerprinting of green tea metabolites, and green tea metabolic profiling
^The spectra obtained by ¹ H-NMR measurements were first phase and baseline corrected by Chenomx NMR Suite 4.6 Software Professional Edition (Chenomx Inc., Canada). It was integrated with a size of 0.04 ppm over a range of 1.0-8.0 ppm. On the other hand, the 4.6 to 5.0 ppm water signal was deleted. All measurements were normalized to the DSS methyl peak area.

FIG. 7 shows the ¹ H-NMR spectrum of a water-soluble compound extracted from the highest quality tea. Metabolite-derived signals were assigned by comparison with signals from known standard compounds analyzed under the same conditions as well as the 750 MHz (pH 4-9) library database of Chenomx NMR Suite 4.6. Of these, about 10 major compounds could be identified. Theanine, amino acids, and quinic acid were mainly observed in the high magnetic field region (δ 0.5-3.0 ppm: FIG. 7A). In the middle magnetic field region (δ 3.0 to 4.5 ppm: FIG. 7B), caffeine, arginine, myo-inositol, chlorogenic acid, and quinic acid were clearly observed except for sucrose and fructose. In the low magnetic field region (δ 5.0-8.0 ppm: FIG. 7C), 2-O-β-L-arabinopyranosyl-myoinositol, p-coumaryl-quinic acid or cinnamic acid, (−) — Epigallocatechin-3-gallic acid (EGCG) and (−)-epicatechin-3-gallic acid (ECG) were mainly observed. (−)-Epigallocatechin (EGC) and (−)-epicatechin (EC) were not observed significantly because of their low water solubility.

FIG. 8 shows all the ¹ H-NMR spectra obtained for the 53 ranked first tea samples superimposed. Although no significant difference was observed in the high magnetic field region (FIG. 8A) and the low magnetic field region (FIG. 8C), several differences were observed in the middle magnetic field region (FIG. 8B). In particular, the signals of δ 3.27 and 3.43 ppm of caffeine were strong in high quality tea samples. Caffeine is known to contribute to tea bitterness and to be important for tea quality. Therefore, it was shown that it is an important factor for creating a quality prediction model in the measurement by NMR.

  Next, in order to grasp the relationship represented by the similarity or dissimilarity in the group of multivariate data, Pirouette (registered trademark) software (Informatrix, Inc.) is used, and a high magnetic field region (δ1.0 The principal component analysis (PCA) was performed on each of the intermediate magnetic field region (δ 3.12-4.34 ppm) and the low magnetic field region (δ 6.15-6.35 ppm). First, the 53 samples were divided into two groups ranked according to their grade (high quality: 1-25 and low quality: 26-53). In the score plot, the difference between the two groups was not clearly seen in the high magnetic field region, but in the middle magnetic field region and the low magnetic field region, it could be roughly divided into two groups of high quality and low quality. (Data not shown). In the loading plot, it was found that signals derived from caffeine, theanine, EGCG, ECG, EGC, and EC contributed to cluster separation (data not shown).

  Furthermore, a projection model for predicting the dependent (response) variable Y from the independent (predictive) variable X by selecting the projection (PLS) (SIMCA-P version 11.0: Umetrics) on the latent structure by the partial least squares method Created. Regression is performed by creating a mathematical model based on system behavior to determine the optimal value for the model parameter for the training sample. The value of the unknown independent variable is then predicted using the resulting training model. In the present example, the quality evaluation ranking was used as a dependent variable. Analytical samples were divided into training set and test set groups. Samples ranked Nos. 5, 15, 25, 35, and 45 were used as test sets for model validation. There were no variables transformed before analysis, data was aggregated and scaled to unit variance.

FIG. 9 shows the PLS relationship between the measured value and the predicted value of the green tea sample. The samples No. 2 and No. 8 were excluded from the model creation because of the large noise. The performance of the PLS regression model was verified by the correlation error R ² and the cross-validated correlation coefficient Q ² and the validation error between the measured value and the predicted value of the mean square error (RMSEP) of the predicted value. The PLS regression model for green tea showed R ² = 0.987 and Q ² = 0.671 (RMSEP = 1.82) (FIG. 9). This shows that it is excellent in fit and predictability. The predictability of the model for predicting the ranking ranking of green tea was tested with a test set (circled in FIG. 11). The prediction results were distributed from the ideal diagonal (RMSEP = 9.22). A certain degree of prediction was obtained by this PLS model.

Furthermore, the predictive power of the PLS model was improved by simplifying the variation using orthogonal signal correction (OSC). By removing the two OSC components from the PLS model, Q ² increased from 0.671 to 0.982 and the predictability improved by 31% (FIG. 10). The predictability of the PLS-OSC regression model was verified again by the test set (circled in FIG. 10). The RMSEP value decreased significantly from 9.22 to 5.91. According to the regression plot shown in FIG. 10, indicating that the increase and decrease of RMSEP Q ² 'is the ability of the predictive model by a signal correction is dramatically improved.

Wavelet transform is a signal processing tool used to compress complex signals to remove noise and extract only relevant information. It is known that combination with OSC can improve the quality of multivariate tests without compromising predictive power. Therefore, also in this embodiment, a combination of OSC and wavelet transform (OSCW) was applied. One-component OSC filtering removed 63.42% of unrelated X variables. Using the 506 wavelet coefficients, 99.50% of the sum of the squares was retained. The obtained PLS-OSCW regression is shown in FIG. R ² and Q ² were 0.982 and 0.944, respectively. The predictive ability of the PLS regression model was improved by the execution of OSCW, and RMSEP decreased from 5.91 to 5.53 in the test set verification (circled in FIG. 11). The number of wavelet coefficients used in the PLS model to capture useful variables was very high and only 0.5% unrelated variables were removed. This indicates that almost all Y variables have been captured by OSC filtering, and only some variation resulting from noise remained. Wavelet analysis enhances pre-processing performance by removing the remaining variables without losing predictability.

Thus, when comparing R ² , Q ² , and RMSEP values for PLS, PLS-OSC, and PLS-OSCW regression, both R ² and Q ² values are greater than 0.9 for all models, Shows that it can be used to predict the quality of green tea with good compatibility and good predictability.

(Example 7: Preparation and analysis of analysis sample for FT-NIR)
Thirteen kinds of dried leaves after processing selected from 64 kinds of first teas ranked (rated) at the 2005 competition were used as samples. These browns are samples of rankings 4, 8, 14, 18, 24, 28, 34, 38, 44, 48, 54, 58, and 64. These tea samples described in the commercial tea fair were obtained from the Nara Prefectural Agricultural Center Tea Industry Promotion Center. Tea ranking was determined by the overall score of the sensory test, which was leaf appearance, aroma, color, and taste of the tea that was appraised by a professional tea appraiser.

  Dry tea leaves (200 mg) were ground in a 2 mL Eppendorf tube with a Retsch ball mill (20 Hz, 10 min) and 600 μL of glycerol, hexane, acetonitrile, or mineral oil was added. Each of these mixtures was homogenized again with a ball mill (20 Hz, 10 minutes) to obtain a pasty analytical sample. Each of these analytical samples was transferred to a 2 mL glass vial.

The obtained pasty analytical sample was measured by FT-NIR. The FT-NIR measuring apparatus used in this example was NICOLET 6700 FT-IR (Thermo Electron Co., Ltd.) equipped with Smart Near-IR UpDRIFT, CaF ₂ beam splitter, and cooled InGaAs detector. The FT-NIR spectrum was recorded from 10000 to 4000 cm ⁻¹ at intervals of 3.857 cm ⁻¹ . The mirror speed was 1.2659 cm / s and the resolution was 8 cm ⁻¹ . The total number of data points was 1557 for each spectrum. Each sample was measured four times.

  When FT-NIR was measured for samples prepared using various solvents, the variation in measurement data was minimal when glycerol (B) was used (FIG. 12). Therefore, the following analysis was performed on samples prepared using glycerol.

(Example 8: Analysis of FT-NIR data, fingerprinting of green tea metabolites, and quality prediction of green tea)
Each data (ie, measured value) obtained by FT-NIR was averaged multiplied by 100,000. All data were preprocessed and transformed by second derivative and standard normal variables (SNV).

Principal component analysis (PCA) was performed using Pirouette (registered trademark) (Informatrix, Inc.), which is commercially available software. In PCA factor 3, 13 tea samples could be divided into 3 groups ranked according to their grade. The variation with factor 3 was 2.38% of the total variation and very low. From this Factor 3 loading plot, a wave number range of 5500-5200 cm ⁻¹ was predicted to be associated with tea grade (FIG. 13). Therefore, PCA was performed again on data in the wave number range of 5500 to 5200 cm ⁻¹ . As a result, grouping was possible by factor 1, and it was found that this wave number range was highly related to tea grade (FIG. 14).

A projection model of tea sample quality was then created by selecting a projection (PLS) on the latent structure by partial least squares. Analytical samples were divided into training set and test set groups. Samples Nos. 14, 34, and 54 were excluded as a test set for model validation. Second order differentiation and standard normal variate (SNV) were performed on all data in the spectral range of 5500-5200 cm ⁻¹ . The number of potential factors in the PLS model was determined by cross validation. The optimal number was 5 in the training set model. The standard error (SEP) of the predicted value of the training set was 1.59, and the SEP of the test set was 3.05 (FIG. 15). The training set studentized residual was 0.997 and the test set was 0.998. These numbers indicate the accuracy of the prediction model. That is, if the value is close to “1”, it means that the result is accurate. These facts suggest that the prediction model is very accurate. From the test set results, this model is accurate for the prediction of low quality samples, but not very accurate for the prediction of high quality samples (FIG. 15). This result suggests that when the tea quality is high, the difference in quality is small, making prediction difficult. This study showed that tea quality prediction by metabolite fingerprinting using FT-NIR is accurate and realistic.

(Example 9: Preparation and analysis of pyrolysis treatment analysis sample for GC / MS)
As in Example 1 above, the dried leaves after processing of 53 kinds of most teas ranked (rated) in the 2005 competition were used as samples. These tea leaves were pulverized into powder using a blender in order to make the leaves and petiole uniform, and stored at -20 ° C.

  1 ± 0.005 mg of the ground sample was weighed and placed in Eco-cup S (Frontier Lab Co., Ltd.) and introduced into PY-2020iD Double Shot Pyrolyzer (Frontier Lab Co., Ltd.). The pyrolyzer oven temperature was set to 500 ° C and the interface set to 250 ° C. The pyrolysis time was set at 5 minutes for each sample.

  The GC / MS device was directly connected to the pyrolyzer via a GC inlet. The GC / MS apparatus used in this example is equipped with a THERMO TR-WaxMS column (60 m × 0.25 mm i.d. × 0.25 μm) (Thermo electron Co.) and a TRACE DSQ mass spectrometer (Thermo electron Co.). And TRACE GC gas chromatograph (Thermo electron Co.). The injection temperature was 230 ° C. High purity helium gas was used as a carrier gas at a flow rate of 1 mL / min. The split ratio of the GC inlet was 20: 1. The analysis was performed with the following oven temperature program: kept isothermal at 70 ° C. for 2 minutes, then ramped from 70 ° C. to 260 ° C. at 7 ° C./minute and held at 260 ° C. for 20 minutes. The transport line and ion source temperature were both 250 ° C. Ions were generated by 70 kV electron impact (EI) and 20 scans per second were recorded over the mass range of 50-650 m / z. The acceleration voltage was activated after every GC scan time.

(Example 10: GC / MS data analysis of pyrolysis-treated green tea samples, fingerprinting of green tea metabolites, and green tea metabolic profiling)
For pre-processing of the data obtained by GC / MS in Example 9 above, the obtained raw chromatographic data (Xcalibur file type, * .raw) is used using the file conversion menu of Xcalibur software (Thermofinnigan). Converted to an ANDI file (analysis data exchange protocol, * .cdf). The chromatogram data points were automatically adjusted to scan for intensity every 0.01 seconds and stored as AIA data without fragment data. Data were transferred to the commercial software LineUp (Informatrix, Inc.) for retention time alignment and aligned using a correlation optimized warping algorithm.

  The relationship between the sample metabolite fingerprint (matrix X) and its quality (matrix Y) was observed. First, the analysis samples were divided into training set and test set groups. Samples with rankings of 2, 12, 22, 32, 42, and 52 were excluded as test sets for model validation. A regression model was created using the partial least-squares projection of potential structure (PLS) (SIMCA-P version 11.0: Umetrics) with orthogonal signal correction (OSC), a spectral filtering technique, and the metabolite profile of green tea ( X variable) and quality ranking (Y variable) were examined. The validity and reliability of the PLS regression model is determined by the model (R2Y), the predictability parameter (Q2), the mean square error (RMSEE) of the estimate of the training set, and the mean square error (RMSEP) of the estimate of the test set Evaluated by the explained rate of variation. The validity and reliability of the PLS regression model was verified with a training set and validated with a test set.

  The two OSC components were removed, leaving 52.11% of the relevant X variable for the Y variable in the model. After constructing the PLS model, according to cross-validation, a 77.9% variation in Y was reflected (R2Y = 0.799) and a 66.9% variation in Y was predicted (Q2 = 0.669). The model created by this training set had an RMSEE value of 7.57. A good correlation was observed by plotting between the first summaries of X (t1) and Y (u1). However, the 46th sample was excluded because the predicted values were very different. When the prediction model was constructed again excluding No. 46, R2Y was improved to 0.876 and Q2 = 0.778, and a good correlation was observed between X and Y (FIG. 16). RMSEE also decreased to 5.36. Next, when verified with a test set, it was in good agreement with the quality ranking (FIG. 16).

  Furthermore, the highly relevant explanatory variable Y important for model construction was identified by computer analysis from VIP (variable importance in projection) values. Large VIP values above 1 are the most relevant explanatory variable Y. Since the molecules are randomly destroyed, it is not feasible to identify metabolites from the chromatogram, but judging from the VIP values, important metabolites could be characterized as retention times (FIG. 17). ). The variable for retention time showed an intensity difference between high ranking green tea and low ranking green tea in the original chromatogram. Therefore, it was found that the quality of green tea can be predicted by ranking the predictor variables in order of the degree of contribution (for example, retention time) to the final model.

(Example 11: Preparation and analysis of analytical sample for UPLC / MS)
The dried leaves after the processing of 56 kinds of the most teas ranked (rated) at the 2006 competition were used as samples. These tea samples described in the commercial tea fair were obtained from the Nara Prefectural Agricultural Center Tea Industry Promotion Center. Tea ranking was determined by the overall score of the sensory test, which was leaf appearance, aroma, color, and taste of the tea that was appraised by a professional tea appraiser. In addition, ranking was given from No. 1 to No. 56 in order from the highest ranking.

  First, dried tea leaves (10 mg) were freeze-dried in a 2 mL Eppendorf tube and pulverized with a Retsch ball mill (20 Hz, 1 minute). The hydrophilic primary metabolite of green tea was extracted using a mixture of methanol, water, and chloroform (2.5: 1: 1, v / v / v). The resulting mixture was incubated for 30 minutes at 37 ° C. in a thermomixer. The solution was then centrifuged at 16000 × g for 10 minutes at 4 ° C. 200 μL of the supernatant was then transferred to a 1.5 mL Eppendorf tube capped with a perforated cap. The resulting extract was dried in a vacuum centrifuge dryer to dryness (overnight). 200 μL of methanol / water (4: 1, v / v) (containing 0.1% (v / v) formic acid) was added to the obtained dried product, and filtered through a PTFE filter to obtain an analytical sample.

4 μL of the analytical sample was subjected to UPLC / MS. In this example, an ACQUITY UPLC ^™ system (Waters) equipped with a binary solvent manager, column heater, and sample manager was used, and detection was performed using LCT premier ^™ XE mass spectrometry (Waters). This apparatus was equipped with a 2.1 × 150 mm, 1.7 μm Acquity UPLC BEH C18 column and operated at 40 ° C. The mobile phase conditions were as follows: mobile phase A with water (containing 0.1% (v / v) formic acid) and mobile phase B with acetonitrile (0.1% (v / v) at a flow rate of 0.3 mL / min. v) containing formic acid) and a linear gradient increasing mobile phase B from 0% (v / v) to 100% (v / v) in 7 minutes, then maintained for 7 minutes. MS used an electrospray ionization (ESI) source in negative mode. The ionization parameters were capillary voltage: 2.0 kV, cone voltage: 0.035 kV, desolvation gas flow rate 500 L / hour, desolvation temperature: 350 ° C., and source temperature: 100 ° C.

(Example 12: UPLC / MS data analysis and green tea metabolic profiling of green tea samples)
For preprocessing of the data obtained by UPLC / MS in Example 11 above, the obtained raw chromatographic data was converted into an ANDI file (analysis data exchange protocol, * .cdf). The converted file (AIA) was preprocessed with data processing software MZmine for peak detection, alignment, gap filtering, and standardization.

  FIG. 18 shows a total ion chromatogram of the first ranking green tea sample among the ranked first tea samples. All rank samples showed similar chromatogram patterns. However, as shown in FIG. 19, a difference in peak intensity was observed due to a difference in tea quality.

  Multivariate analysis was then performed to reveal important compounds and information from this complex data. Principal component analysis (PCA) (SIMCA-P version 11.0: Umetrics) was selected in order to grasp the relationship represented by similarity or dissimilarity among groups of multivariate data. An unsupervised PCA was performed to grasp the whole. For the nine principal components, R2X = 0.767 and Q2 = 0.508, and low and high grade teas were separated separately along the PC1 axis (FIG. 20). That is, high quality green tea samples were collected on the negative side of PC1, and low quality green tea samples were collected on the positive side of PC1. A loading plot corresponding to the score plot is shown in FIG. Differences in predicted quality indicated differences in metabolic profiling of green tea samples.

  Next, in order to examine the relationship between green tea quality and its metabolome, a projection model was created by selecting projections (PLS) (SIMCA-P version 11.0: Umetrics) on the latent structure by the partial least squares method. The 56 analysis samples obtained in Example 11 were divided into a training set (50 samples) and a test set (6 samples). Samples Nos. 2, 12, 22, 32, 42, and 52 were excluded as test sets for model validation. The X matrix of all variables was scaled to the center. Two prominent components were extracted and, according to cross validation, described 83.3% variation in Y (R2Y = 0.833) and predicted 71.4% variation in Y (Q2Y = 0.714) (FIG. 22). The test set was then predicted with the PLS model, and the prediction accuracy of the test sample (RMSEP = 7.22) was obtained for model evaluation based on the training set sample (RMSEE = 6.75) (FIG. 22).

  According to the present invention, it is possible to predict the overall quality of product green tea, which has been difficult in the past, by a simple method. That is, quality can be predicted quickly and beneficially by screening green tea by various instrumental analyzes such as chromatography and spectroscopic analysis.

  Traditional instrumental analysis of green tea focuses on a specific group of compounds that are considered important for a specific function, such as antioxidant compounds for medical purposes, and catechins and other phenolic compounds for flavor and aroma. It was hit. However, the characterization of green tea can come from combinations rather than from a single metabolite. Therefore, if the method of the present invention by analyzing a plurality of components at once is used, more accurate quality prediction is possible, and further improvement of the green tea production process or the green tea distribution and storage process can be expected.

FIG. 4 shows PCA analysis results, score plots for high and low ranking green tea samples. It is a figure which shows the relationship between the measurement and prediction green tea quality (ranking) of the PLS model about 46 types of green tea samples as a training set. FIG. 6 shows the relationship between measured and predicted green tea quality (ranking) of PLS models for all 53 green tea samples of both test set (shown in circles) and training set. It is a figure which shows the relationship between the measurement and prediction green tea quality (ranking) of the PLS model about 46 types of green tea samples as a training set using orthogonal signal correction | amendment (OSC). FIG. 6 shows the relationship between the measured and predicted green tea quality (ranking) of the PLS model for all 53 green tea samples in both test set (indicated by circles) and training set using orthogonal signal correction (OSC). is there. It is a graph which shows the influence of the variable used in order to produce | generate the quality predicted value of green tea. ¹ shows a ¹ H-NMR spectrum (750 MHz, D ₂ O) of a sample extracted from the highest quality tea. The ¹ H-NMR spectra obtained for all of the 53 ranked first tea samples are superimposed. It is a diagram showing the relationship between green tea quality measured and predicted (ranking) for both PLS model calculated from ^{1 H-NMR} data of the green tea samples of the training set and the test set (indicated by circles). It is a diagram showing the relationship between green tea quality measured and predicted (ranking) for both PLS-OSC model calculated from ^{1 H-NMR} data of the green tea samples of the training set and the test set (indicated by circles). It is a diagram showing the relationship between green tea quality measured and predicted (ranking) for both PLS-OSCW model calculated from ^{1 H-NMR} data of the green tea samples of the training set and the test set (indicated by circles). It is a FT-NIR spectrum about the sample prepared with various solvent, A shows a case where A is a blank, B is glycerol, C is mineral oil, D is hexane, and E is acetonitrile. It is a figure which shows a PCA analysis result, and is a factor 3 loading plot. FIG. 7 shows PCA analysis results for data in the spectral range of 5500-5200 cm ⁻¹ , and is a score plot for high and low ranking green tea samples. FIG. 6 shows the relationship between measured and predicted green tea quality (ranking) of PLS models for all 53 green tea samples of both test set (shown in circles) and training set. In graph showing measured and relationship between green tea quality prediction (ranking) for PLS-OSC model calculated from ^{1 H-NMR} data of the green tea samples of the training set and the test set (indicated by circles) except for the sample of 46th is there. It is a graph which shows the VIP value for every holding time characterizing important metabolism. It is a total ion chromatogram of UPLC of the green tea sample of the 1st ranking. It is a chromatogram which shows the ionic strength of UPLC of high quality green tea (No. 1 and 2) and low quality green tea (No. 55 and 56). Figure 2 is a PCA score plot by UPLC / MS analysis of all ranking green tea samples. It is a figure which shows the PCA analysis result of UPLC / MS analysis of a green tea sample, and is a loading plot of PC1. FIG. 6 shows the relationship between the measured and predicted green tea quality (ranking) of the PLS model in UPLC / MS analysis of all 56 green tea samples of both the test set (shown in circles) and the training set.

Claims (13)

A quality prediction method for green tea,
Pre-treating green tea to obtain an analytical sample;
Subjecting the analysis sample to instrumental analysis to obtain an analysis result;
Converting the analysis result into numerical data and performing multivariate analysis; and predicting quality from the obtained analysis result;
Including a method.   The method of claim 1, wherein the multivariate analysis is a PLS regression analysis.   The method according to claim 1 or 2, wherein the pretreatment includes a step of pyrolyzing the green tea, and the instrumental analysis is a combination of gas chromatography and mass spectrometry.   The method according to claim 3, wherein the analysis result is quality predicted based on a holding time.   The method according to claim 1 or 2, wherein the pretreatment includes a step of extracting a hydrophilic compound from the green tea to obtain an extract.   6. The method of claim 5, wherein the instrumental analysis is a combination of gas chromatography and mass spectrometry, and the pretreatment further comprises silylating the extract.   Among the analysis results, data on arabinose, glutamine, citric acid, glutamic acid, aspartic acid, inositol, shikimic acid, phosphoric acid, fructose, theanine, malic acid, caffeine, quinic acid, glucose, mannose, ribose, and sucrose The method of claim 6, wherein   The method according to claim 5, wherein the instrumental analysis is a combination of liquid chromatography and mass spectrometry, or nuclear magnetic resonance analysis.   The method according to claim 1 or 2, wherein the pretreatment includes a step of pulverizing the green tea to obtain a powder, and a step of pasting the powder with a solvent.   The method according to claim 9, wherein the instrumental analysis is infrared spectroscopy or near infrared spectroscopy. The instrumental analysis is Fourier transform near-infrared spectroscopic analysis, and in the multivariate analysis step, the result of 5500-5200 cm ⁻¹ is converted in the analysis result, and the converted data is further converted into two. The method of claim 10, wherein the second derivative is performed. Pre-treating a plurality of green teas of known quality to obtain individual analytical samples;
Subjecting the individual analysis samples to instrumental analysis to obtain individual analysis results; and converting the relationship between the individual analysis results and the quality into numerical data for multivariate analysis;
A green tea quality prediction model obtained by A quality prediction method for green tea,
Pre-treating green tea to obtain an analytical sample;
Subjecting the analysis sample to instrumental analysis to obtain an analysis result;
Converting the analysis result into numerical data and performing multivariate analysis; and collating the obtained analysis result with the quality prediction model according to claim 12;
Including a method.

Images