Quantitative analysis model and prediction method for titer, moisture and protein content of biological enzyme

Abstract

The invention discloses a quantitative analysis model and a prediction method for the titer, moisture and protein content of a biological enzyme, and relates to the technical field of near infrared spectrum analysis. The quantitative analysis model provided by the invention has good stability and reliability; according to the model, the titer, the moisture and the protein content of the biological enzyme are detected in real time by establishing a quantitative relationship between the moisture and the protein content and the titer in a biological enzyme sample and a near infrared spectrum. The prediction method provided by the invention does not need solution preparation in advance and other pretreatment, and does not need reagents and consumables. And three indexes of activity / titer, moisture and protein content of the biological enzyme can be rapidly and synchronously detected in real time. And after the sample detection is completed, the sample can be reused without damage.

Description

Technical Field

[0001] The invention relates to the technical field of near infrared spectroscopy analysis, and in particular to a quantitative analysis model and a prediction method for the titer, moisture and protein content of a biological enzyme. Background Art

[0002] Bioactive enzymes are a general term for enzymes with biological activity that are extracted and separated from organisms. Most bioactive enzymes are macromolecules with complex structures and are difficult to standardize. In the quality control of bioactive enzymes, bioactivity / potency is a key quality attribute that reflects the effectiveness of bioactive enzymes. The determination of bioactivity / potency is a very important analytical method in drug development and post-marketing safety and efficacy assessment.

[0003] In the quality control of bioactive enzymes, bioactivity / titer is a key quality attribute that reflects the effectiveness of bioactive enzymes. Other key quality indicators include water content and protein content. The agarose-fibrin plate method is commonly used in the prior art to determine the bioactivity / titer of bioactive enzymes. The detection time of this method requires at least one day and one night, and there are problems such as long time consumption, many reagents, and expensive consumables. In recent years, a parallel line analysis method based on quantitative reaction has been proposed. However, both methods require more than ten hours of incubation time, which obviously affects the efficiency in the actual production process. In addition, there are two methods for determining the titer, namely toluenesulfonyl-L-arginine methyl ester (TAME) and chromogenic substrate assay. In the TAME method, TAME is a reaction substrate, which has a strong cleavage activity on the carboxyl terminus of arginine in lumbrokinase. The effectiveness of the enzyme can be indirectly determined by measuring the amount of methanol produced by lumbrokinase cleavage. In the chromogenic substrate assay, the hydrolysis and cleavage of the substrate amide bond causes a color effect, thereby enabling the effectiveness of the enzyme to be monitored. Both of the above methods require expensive substrates and special fluorescence detectors, which limits their application in actual production processes.

[0004] At present, the moisture content of biologically active enzymes is determined by the second method (drying method) in Part III 0832 Moisture Determination Method of the Chinese Pharmacopoeia 2020 Edition. The protein content is determined by the first method (Kjeldahl method) in Part III 0731 Protein Content Determination Method of the Chinese Pharmacopoeia 2020 Edition. However, the drying method for measuring moisture content and the Kjeldahl method for measuring protein content both have the problems of long time consumption and sample irreversibility during laboratory testing.

[0005] In view of this, the present invention is proposed. Summary of the invention

[0006] The purpose of the present invention is to provide a quantitative analysis model and prediction method for the titer, moisture and protein content of a biological enzyme so as to simultaneously and quickly realize the titer detection, moisture content prediction and protein content prediction of the biological enzyme. The prediction method provided by the present invention does not require the biological enzyme sample to be pretreated into a solution, and near-infrared spectra can be directly collected.

[0007] The present invention is achieved in that:

[0008] In a first aspect, the present invention provides a method for constructing a quantitative analysis model of the titer, moisture and protein content of a biological enzyme, comprising the following steps:

[0009] (1) collecting near infrared spectra of biological enzyme samples, and detecting the biological enzyme titer, water content and protein content of the biological enzyme samples;

[0010] (2) Screening out a calibration set from the biological enzyme samples for constructing a quantitative analysis model and performing spectral data preprocessing on the near-infrared spectra of the biological enzyme samples;

[0011] The preprocessing method is selected from at least one of first-derivative spectroscopy (1st), second-derivative spectroscopy (2nd), Savitzky golay smoothing (SG), standard normal variate transformation (SNV), multiplicative scatter correction (MSC), baseline offset subtraction, finite impulse response filtering (FIR) and wavelet transform;

[0012] (3) Based on the spectral data after pretreatment in step (2) and the potency, moisture content and protein content of the biological enzyme sample, an analytical method is used to establish a potency model, a moisture content model and a protein content model:

[0013] The analysis methods were multivariate linear regression (MLR), principal component analysis (PCA), partial least squares (PLS), artificial neural network (ANN), and topological;

[0014] The biological enzyme sample is a solid or liquid biological enzyme sample.

[0015] By preprocessing the near-infrared spectra of the collected enzyme samples to filter out noise and remove the interference of high-frequency noise on the signal, the prediction ability (quantitative analysis ability) of the model is improved. Noise includes but is not limited to: high-frequency random noise, baseline offset, signal background, sample inhomogeneity, light scattering, etc.

[0016] The present invention establishes a quantitative relationship between the water content, protein content and potency of a biological enzyme sample and a near-infrared spectrum, thereby detecting the potency (or activity), water content and protein content of the biological enzyme in real time.

[0017] In the process of constructing the above model: the water content of the biologically active enzyme is determined by, for example, the second method (drying method) in Part III 0832 of the "Chinese Pharmacopoeia 2020 Edition" of the water determination method. The protein content is determined by, for example, the first method (Kjeldahl nitrogen determination method) in Part III 0731 of the "Chinese Pharmacopoeia 2020 Edition" of the protein content determination method, and the enzyme titer of the biological enzyme sample is detected by, for example, the agarose-fibrin plate method.

[0018] In a preferred embodiment of the present invention, the preprocessing method includes: the water content model uses the first-order derivative method to preprocess the original spectrum, the protein content model uses the filtering and smoothing method to preprocess the original spectrum, and the potency model uses the filtering and smoothing method to preprocess the original spectrum.

[0019] The inventors compared the effects of various preprocessing methods such as first-order derivative spectroscopy, second-order derivative spectroscopy, filtering and smoothing, standard normal variable transformation, and additional scattering correction on the model performance parameters after spectral preprocessing, and found that the bioenzyme titer was preprocessed using filtering and smoothing (Savitzky golay smoothing, SG), the moisture content data was preprocessed using the first-order derivative 1st, and the protein content was preprocessed using SG smoothing. The model established after preprocessing has better predictive ability, and the use of the above-mentioned preprocessing methods to process spectral data is helpful to improve the performance (stability and reliability) of the quantitative analysis model.

[0020] Among them, the SG smoothing algorithm can filter the noise signal in the spectrum, 1st can eliminate the influence of baseline drift, and SNV can be used to enhance the difference between spectra, making the sample spectrum difference significant, which is helpful to extract the signals between overlapping spectra and is more conducive to the establishment of near-infrared spectral quantitative analysis model.

[0021] In a preferred embodiment of the present invention, the spectral data after pretreatment in step (2), the enzyme titer of the enzyme sample in step (1), and the water content and protein content data of the enzyme sample are imported into The UnscramblerX10.4 software, and an analytical method is used to establish a quantitative analysis model for the titer, water content and protein content of the enzyme.

[0022] In a preferred embodiment of the present invention, the partial least squares method is used to establish a quantitative analysis model for the titer, water content and protein content of the biological enzyme.

[0023] In a preferred embodiment of the present invention, the above-mentioned biological enzyme includes but is not limited to lumbrokinase, and may also be other enzymes with biological enzyme activity extracted and separated from animals, plants or microorganisms.

[0024] In an optional embodiment, the lumbrokinase is in the form of lumbrokinase dry powder. The model provided by the present invention can directly detect and analyze the lumbrokinase dry powder without the need for pre-preparation of solutions or other pre-treatments.

[0025] In a preferred embodiment of the present invention, the parameters of the near-infrared spectrum of the biological enzyme sample collected in the above step (1) are: wavelength range is 908-1650nm; 13.6ms integration time of each spectrum corresponds to the accumulation of 100 scans; the ambient temperature of the near-infrared scanning is 25℃±0.5℃, and the humidity is 30%±0.5.

[0026] Under the above-mentioned near-infrared spectrum parameter conditions, it can be ensured that spectra of different samples are collected under the same parameter conditions, the original spectra are preprocessed and combined with chemometrics to extract effective spectral information, so as to establish a quantitative relationship between the moisture, protein content and potency in the sample and the near-infrared spectrum. The wavelength range of 908-1650nm is the default wavelength range of the spectrometer. Selecting this wavelength range can extract effective spectral information without causing too much noise. Therefore, this band is selected when collecting spectra, and the information collected in this band is also retained when modeling.

[0027] In a preferred embodiment of the present invention, the screening of the calibration set in the above step (2) includes: analyzing the near-infrared spectra of the biological enzyme samples using principal component analysis (PCA), eliminating abnormal biological enzyme samples, and dividing the biological enzyme samples with outliers eliminated into a calibration set and a validation set through an interactive validation algorithm, wherein the validation set is used to validate the quantitative analysis model.

[0028] The inventors found that the model established by analyzing abnormal enzyme samples through PCA is more stable and reliable.

[0029] In a preferred embodiment of the present invention, the construction method further comprises using a validation set to validate the quantitative analysis model.

[0030] In a second aspect, the present invention also provides a quantitative analysis model for the potency, moisture and protein content of a biological enzyme constructed using the above construction method.

[0031] In a third aspect, the present invention also provides a method for predicting the titer, moisture and protein content of a biological enzyme, comprising the following steps:

[0032] The above quantitative analysis model is used to process the near infrared spectrum of the biological enzyme sample to be tested;

[0033] In an optional implementation, the prediction method is a real-time prediction method.

[0034] The prediction method provided by the present invention does not require the preparation of solutions and other pretreatments in advance, and does not require reagents and consumables. It can detect three indicators of biological enzyme activity / titer, water content and protein content in real time and synchronously. Compared with the analysis method of the prior art (at least a dozen hours), the prediction method provided by the present invention greatly shortens the analysis time, and the analysis process only takes a few seconds to complete. After the sample detection is completed, the sample is intact and can be reused.

[0035] In a preferred embodiment of the present invention, the prediction method comprises first collecting a near infrared spectrum of the biological enzyme sample to be tested, and then using a quantitative analysis model to predict the titer, water content and protein content of the biological enzyme sample;

[0036] The biological enzyme is lumbrokinase;

[0037] In an optional embodiment, the lumbrokinase is a lumbrokinase dry powder.

[0038] A computer-readable storage medium stores a computer program / instruction, which implements the steps of the above prediction method when executed by a processor.

[0039] Computer-readable storage media may be, but are not limited to, random access memory (RAM), read only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable read-only memory (EEPROM), etc.

[0040] In an optional embodiment, an electronic device is also provided, which may include a memory, a processor, a bus and a communication interface, wherein the memory, the processor and the communication interface are electrically connected to each other directly or indirectly to achieve data transmission or interaction. For example, these elements may be electrically connected to each other via one or more buses or signal lines. The processor may process information and / or data related to target identification to perform one or more functions described in the present application.

[0041] The present invention has the following beneficial effects:

[0042] (1) The present invention performs spectral data preprocessing on the near-infrared spectra of the collected biological enzyme samples to filter out noise and remove the interference of high-frequency noise on the signal, thereby improving the prediction ability (quantitative analysis ability) of the model;

[0043] (2) The quantitative analysis model provided by the present invention has good stability and reliability; the model establishes a quantitative relationship between the water content, protein content and potency of the biological enzyme sample and the near-infrared spectrum, thereby detecting the potency (or activity), water content and protein content of the biological enzyme in real time;

[0044] (3) The prediction method provided by the present invention does not require the preparation of solutions or other pretreatments, and does not require reagents or consumables. It can simultaneously detect three indicators of biological enzyme activity / titer, water content and protein content in real time. Compared with the analysis method of the prior art (at least a dozen hours), the prediction method provided by the present invention greatly shortens the analysis time, and the analysis process only takes a few seconds to complete. After the sample test is completed, the sample can be reused without loss. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0046] Figure 1 This is the original spectrum of lumbrokinase dry powder;

[0047] Figure 2 This is the noise graph of lumbrokinase dry powder;

[0048] Figure 3 This is a model diagram of near infrared spectroscopy quantitative analysis of water content, protein and potency;

[0049] Figure 4 Instrument and method configuration page diagram;

[0050] Figure 5 To configure the workflow page diagram;

[0051] Figure 6 This is the Unscrambler model page diagram;

[0052] Figure 7 To output the page map;

[0053] Figure 8 It is a prediction window diagram for the water content, protein and potency index of the sample to be tested. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present invention will be described clearly and completely below. If the specific conditions are not specified in the embodiments, they are carried out according to conventional conditions or conditions recommended by the manufacturer. If the manufacturer of the reagents or instruments used is not specified, they are all conventional products that can be purchased commercially.

[0055] The features and performance of the present invention are further described in detail below in conjunction with the embodiments.

[0056] The reagents, experimental materials, instruments and software information involved in the following examples are as follows:

[0057] Table 1 Reagents

[0058]

[0059] The samples were from batch 129 of lumbrokinase dry powder produced by Jiangzhong Pharmaceutical Co., Ltd. (Nanchang, China).

[0060] Table 2 Instruments and software

[0061]

[0062]

[0063] Example 1

[0064] This embodiment takes lumbrokinase dry powder as an example and provides a method for constructing a quantitative analysis model for the potency, moisture and protein content of lumbrokinase.

[0065] 1. Collection of Near-Infrared Spectra

[0066] In this example, a Micro NIR 1700 spectrometer was used to collect spectra of 129 batches of lumbrokinase dry powder in the range of 908-1650 nm. Before collecting the spectra, the instrument was stabilized for 30 minutes to ensure that the temperature fluctuations of the instrument were within a stable range. The parameters were set as follows: the 13.6 ms integration time for each spectrum corresponded to the accumulation of 100 scans. The lumbrokinase dry powder was directly loaded into the accessory vial and fixed, and three spectra were taken for each sample to reduce operating errors, and their average spectrum was used for further analysis. During the spectral acquisition, the indoor ambient temperature was maintained at 25°C and the humidity was approximately 30%.

[0067] The original near-infrared absorption spectra of all lumbrokinase dry powder samples are shown in Figure 1 As shown in the figure, due to a large amount of overlap, no obvious signal can be found in the original spectrum, and the absorption peak is unclear. Therefore, it is necessary to preprocess the original spectrum and combine chemometrics to extract effective spectral information in order to establish a quantitative relationship between the moisture content, protein content and potency of lumbrokinase dry powder and the near-infrared spectrum.

[0068] 2. Determination of sample standard values, including moisture content, protein content and potency.

[0069] After collecting the NIR spectrum, the moisture content of each sample was determined using the second method (drying method) in Part III 0832 Moisture Determination Method of the Chinese Pharmacopoeia 2020 Edition.

[0070] The protein content of each sample was determined using the first method (Kjeldahl nitrogen method) in Part III 0731 Protein Content Determination Method of the "Chinese Pharmacopoeia 2020 Edition".

[0071] The titer determination of each sample was performed by agarose-fibrin plate method for reference analysis. The detailed steps of the agarose-fibrin plate method are as follows: add an appropriate amount of agarose to a working solution containing pH 7.8 phosphate buffer and 0.9% sodium chloride solution in a ratio of 1:17, add fibrinogen and thrombin when the agarose solution is 55°C, mix well, and then quickly pour the solution into a horizontal plate. After cooling and solidification at room temperature, an agarose-fibrin plate is obtained. The lumbrokinase standard with a concentration of 2000U / ml, 4000U / ml, 6000U / ml, 8000U / ml and 10000U / ml and the sample to be tested are applied to the holes that have been punched on the fibrin plate. The agarose-fibrin plate is placed in a 37°C incubator for 18 hours.

[0072] After incubation, use a vernier caliper to measure the two vertical diameters of the lysis zone, use the logarithm of the standard U number of lumbrokinase as the horizontal coordinate, and the logarithm of the product of the two vertical diameters as the vertical coordinate, calculate the standard curve regression equation, and substitute the logarithm of the product of the two vertical diameters of the test sample into the standard curve regression equation to calculate the test sample potency U number. Two points are made for each standard sample and test sample, and the average value is taken as the potency of lumbrokinase.

[0073] In the process of determining the moisture content, protein content and potency of lumbrokinase dry powder, in order to reduce the detection error and ensure the accuracy of near-infrared detection, the moisture content, protein content and potency of 129 samples were operated by the same person, and the stability of the laboratory environment temperature and the consistency of the test operation were ensured. The statistical information of the 129 samples is shown in Table 3.

[0074] Table 3 Statistical results of lumbrokinase dry powder sample data

[0075]

[0076] 3. Selection of preprocessing methods for near infrared spectroscopy quantitative analysis model

[0077] The irregularity of solid powder introduces noise, which in turn affects the near-infrared spectrum. Figure 2 In addition, during the acquisition of near-infrared transmittance and reflectance spectra, background noise and specific physical factors will also affect NIRS, which may cause the shift of the spectral baseline and the amplification of the noise signal. Therefore, before performing spectral analysis, the collected original spectrum should be preprocessed as necessary.

[0078] This embodiment selects the wavelength of 908-1650nm for modeling. This embodiment uses five preprocessing methods: first-derivative spectroscopy (1st), second-derivative spectroscopy (2nd), Savitzky golay smoothing (SG), standard normal variate transformation (SNV), and multiplicative scatter correction (MSC) to preprocess the spectral data collected in step 1. The five preprocessing methods are compared for R 2 c , R 2 p , RMSEC and RMSEP and the influence of the number of main factors.

[0079] Table 4 Pretreatment results of NIRS analysis model for protein, moisture and potency of lumbrokinase dry powder

[0080]

[0081] Note: MSC means multivariate scatter correction; SNV means standard normal variate transformation; 1st means first-order derivative processing; 2nd means second-order derivative processing; SG means Savitzky-Golay filter smoothing processing.

[0082] Table 4 compares the data and lists the advantages and disadvantages of the models obtained by five different preprocessing methods. 2 c , prediction coefficient of determination R 2 p , root mean square error of calibration (RMSEC), and root mean square error of prediction (RMSEP) are the model evaluation indicators. 2 c and R 2 p The larger the value and the smaller the difference, the smaller the RMSEC and RMSEP values ​​and the smaller the difference, the better the corresponding model performance.

[0083] From the data in Table 4, it can be seen that among the five pretreatment methods, the water content is pretreated using the first-order derivative 1st, and the model evaluation index R 2 c and R 2 p The value is the largest and the difference is the least, the RMSEC and RMSEP values ​​are the smallest and the difference is the least, and the corresponding model performance is the best, so the first-order derivative 1st is used for preprocessing of water content.

[0084] Similarly, protein content was pretreated with SG, and the model evaluation index R 2 c and R 2 p The value is the largest and the difference is the least, the RMSEC and RMSEP values ​​are the smallest and the difference is the least, and the corresponding model performance is the best, so SG smoothing is used for preprocessing of protein content.

[0085] SG was used for preprocessing of potency, and the model evaluation index R 2 c and R 2 p The value is the largest and the difference is the least, the RMSEC and RMSEP values ​​are the smallest and the difference is the least, and the corresponding model performance is the best, so SG smoothing is used for preprocessing of the potency.

[0086] The SG smoothing algorithm can filter out the noise signal in the spectrum, and first eliminate the influence of baseline drift. SNV can be used to enhance the difference between spectra. The spectral differences between samples are significant, so that the signals between overlapping spectra can be extracted, which is more conducive to the establishment of NIRS quantitative model.

[0087] 4. Establishment of near infrared spectroscopy quantitative analysis model

[0088] The NIRS data of 129 lumbrokinase dry powder samples and their corresponding potency, moisture and protein content were imported into The Unscrambler X 10.4 software. The optimal preprocessing method selected in step 3 was selected for spectral data preprocessing. The PLSR method [1] was used for calculation and the cross-validation method was used to establish a quantitative model between the near-infrared spectrum and various parameters.

[0089] Model evaluation: evaluate the correlation between the predicted value and the measured value of each parameter to determine whether the established NIRS quantitative correction model is reasonable:

[0090] Corrected determination coefficient R 2 c , prediction coefficient of determination R 2 p , root mean square error of calibration (RMSEC), and root mean square error of prediction (RMSEP) are the model evaluation indicators. 2 c and R 2 p The larger the value and the smaller the difference, the smaller the RMSEC and RMSEP values ​​and the smaller the difference, the better the corresponding model performance.

[0091] The results of the PLS quantitative model are as follows: 2 c , R 2 p , RMSEC and RMSEP were 0.9240, 0.9173, 0.1213% and 0.1275% respectively; the R 2 c , R 2 p , RMSEC and RMSEP were 0.9615, 0.9553, 0.0020 and 0.0022 respectively; R 2 c , R 2 p, RMSEC and RMSEP were 0.9632, 0.9591, 26.9713 IU / mg, and 28.5876 IU / mg, respectively. The model performance is good. Figure 3 .

[0092] As shown in Table 3, the moisture content model can detect the moisture range from 2.1124% to 4.6016%, the protein model can detect the protein content range from 0.5973g / g to 0.6373g / g, and the potency model can detect the potency range from 28786IU / mg to 29364IU / mg.

[0093] 5. Validation of near infrared spectroscopy quantitative analysis model

[0094] 10 external independent prediction sets of 10 samples were obtained, and three spectra were collected for each sample. The water content, protein and potency content of lumbrokinase dry powder in the samples were predicted by the newly established quantitative analysis model. The correlation coefficient R of water content prediction 2 =0.9522; Correlation coefficient R for protein prediction 2 =0.9385; correlation coefficient of potency prediction R 2 =0.9251, the results show that the prediction results are good and the model is feasible.

[0095] Example 2

[0096] This example takes lumbrokinase dry powder as an example to demonstrate a method for simultaneously predicting the three indicators of water content, protein and potency of a sample to be tested by using an established prediction model.

[0097] 1. Method setting and model import

[0098] Log in to MicroNIR TM Pro v3.1 software, click "Tools", "Instrument and Method Configuration" icon buttons in turn to enter the "Instrument and Method Configuration" page, create a new "Lumbrokinase Dry Powder" method setting and spectrometer configuration, and select diffuse reflectance, 13.6ms, and 100 times for mode, integration time, and scan times, as shown below Figure 4 .

[0099] Click the "Workflow" button to enter the workflow configuration page, and select "CAMO Prediction" for sample scanning, as shown below Figure 5 .

[0100] Click "CAMO Prediction" to jump to the Unscrambler model selection page, import the water content, protein, and potency prediction models constructed in Example 1 in sequence, and click OK. Figure 6 .

[0101] Click the "Output" button to jump to the output setting page, set the output of water content, protein, and potency at the same time, and save the method settings. Figure 7 .

[0102] 2. Collection of near infrared spectra of samples to be tested

[0103] Micro NIR 1700 spectrometer was used to collect spectra of a batch of lumbrokinase dry powder samples in the range of 908-1650 nm. The collection method was the same as that used in model construction in Example 1.

[0104] 3. Prediction of water content, protein content and potency of the sample to be tested

[0105] Click the "Search Data" button to enter the search data page. Click "Batch Prediction" on the left side of the page. Select the method "Lumbrokinase Dry Powder" set in step 1 in the "Select Method" selection box. Select the batch number of the lumbrokinase dry powder sample to be tested in the item selection box, and then click "Search". The water content (3.5361%), protein (0.6190g / g), and potency prediction value (28888.0879IU / mg) of the batch of samples to be tested will be displayed at the same time. Figure 8 . Thus, the prediction of the three indicators of water content, protein and potency of the sample to be tested is completed.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for constructing a quantitative analysis model for the titer, moisture and protein content of a biological enzyme, characterized in that: It includes the following steps: (1) collecting near infrared spectra of biological enzyme samples, and detecting the titer, water content and protein content of the biological enzyme samples respectively; (2) selecting a calibration set from the biological enzyme sample for constructing a quantitative analysis model, and performing spectral data preprocessing on the near-infrared spectrum of the biological enzyme sample; The preprocessing method is selected from at least one of first-order derivative spectroscopy, second-order derivative spectroscopy, filtering and smoothing, standard normal variable transformation, additional scatter correction, baseline offset subtraction, finite impulse response filtering and wavelet transformation; (3) Based on the spectral data after pretreatment in step (2) and the potency, water content and protein content of the biological enzyme sample, an analytical method is used to establish a potency model, a water content model and a protein content model: The analysis methods are multiple linear regression, principal component analysis, partial least squares, artificial neural network and topology; The biological enzyme sample is a solid or liquid biological enzyme sample.

2. The construction method according to claim 1, characterized in that: The preprocessing method includes: the water content model uses the first-order derivative method to preprocess the original spectrum, the protein content model uses the filtering and smoothing method to preprocess the original spectrum, and the potency model uses the filtering and smoothing method to preprocess the original spectrum; Preferably, the spectral data after pretreatment in step (2), the bioenzyme titer of the bioenzyme sample in step (1), and the water content and protein content data of the bioenzyme sample are imported into The Unscrambler X10.4 software, and the quantitative analysis model of the bioenzyme titer, water content and protein content is established by using the analysis method; Preferably, the partial least squares method is used to establish a quantitative analysis model for the titer, moisture and protein content of the biological enzyme.

3. The construction method according to claim 2, characterized in that: The biological enzyme is lumbrokinase; Preferably, the lumbrokinase is lumbrokinase dry powder.

4. The construction method according to claim 1, characterized in that: The parameters of the near infrared spectrum of the biological enzyme sample collected in step (1) are: wavelength range 908-1650nm; 13.6ms integration time of each spectrum corresponds to the accumulation of 100 scans; the ambient temperature of the near infrared scan is 25℃±0.5℃, and the humidity is 30%±0.

5.

5. The construction method according to claim 1, characterized in that: The screening of the calibration set in step (2) includes: analyzing the near-infrared spectra of the biological enzyme samples by principal component analysis, eliminating abnormal biological enzyme samples, and dividing the biological enzyme samples with abnormal values ​​eliminated into the calibration set and the verification set by an interactive verification algorithm, wherein the verification set is used to verify the quantitative analysis model.

6. The construction method according to claim 5, characterized in that: The construction method also includes using the verification set to verify the quantitative analysis model.

7. A quantitative analysis model for the titer, moisture and protein content of a biological enzyme constructed by the construction method according to any one of claims 1 to 6.

8. A method for predicting the titer, water content and protein content of a biological enzyme, characterized in that: It includes the following steps: The quantitative analysis model of claim 7 is used to process the near infrared spectrum of the biological enzyme sample to be tested; Preferably, the prediction method is a real-time prediction method.

9. The prediction method according to claim 8, characterized in that: The prediction method comprises first collecting a near infrared spectrum of a biological enzyme sample to be tested, and then using the quantitative analysis model to predict the titer, water content and protein content of the biological enzyme sample; The biological enzyme is lumbrokinase; Preferably, the lumbrokinase is lumbrokinase dry powder.

10. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the steps of the prediction method described in any one of claims 8-9 are implemented.

Images