Method for rapid detection of three high-temperature actinomyces and application thereof in fermented products

Abstract

The application discloses a method for rapidly detecting three kinds of high-temperature actinomycetes and application thereof in fermented products, and belongs to the field of microbial detection and ecology application. The method is used for Raman detection of the three kinds of high-temperature actinomycetes, and a training model is established by using characteristic wave bands of the high-temperature actinomycetes to identify unknown bacterial strains. The method is applied to samples such as Daqu, fermented grains, compost and fermented feed, and bacterial suspension samples are prepared by using gradient centrifugation, homogenization or vortexing, single-cell Raman detection is performed, the established high-temperature actinomycete training model is used for machine learning prediction of Raman spectra of unknown bacteria to predict the type of the bacterial strain, real-time, rapid and label-free detection of high-temperature actinomycetes in the bacterial suspension of the fermented product is realized, the type of the high-temperature actinomycetes is determined, and the detection and analysis time is greatly shortened.

Description

Technical Field

[0001] This invention relates to a method for rapid detection of three types of thermophilic actinomycetes and their application in fermentation products, belonging to the field of microbial detection and ecological applications. Background Technology

[0002] Thermotropic actinomycetes are considered by the microbiological community to be a group of Gram-positive, chemoheterotrophic bacteria capable of growing and metabolizing in extreme environments. Taxonomically, they fall between bacteria and actinomycetes, morphologically resembling actinomycetes, but molecularly closer to bacteria. Most species in this group can grow at temperatures above 50°C and are widely distributed in terrestrial hydrothermal vents and high-temperature composting environments. Their endospores exhibit good resistance, making them extreme resource microorganisms adapted to high temperatures. They possess physiological, biochemical, and metabolic characteristics not found in ordinary microorganisms, and have significant potential industrial application value and theoretical significance. Current literature has conducted in-depth studies on the community structure and diversity of bacteria in high-temperature Daqu (a type of starter culture) for Baijiu (Chinese liquor), finding that thermotropic actinomycetes are the dominant genus in Daqu bacteria, with common thermotropic actinomycetes being the dominant species. Their metabolites play an indispensable role in the formation of the unique flavor compounds in Baijiu. Therefore, rapid and reliable identification of thermophilic actinomycetes is essential during the fermentation process of fermented products. By identifying the types and content changes of thermophilic actinomycetes, the fermentation process or abnormalities can be monitored, thereby ensuring product quality in the food production process.

[0003] Currently, rapid identification of thermophilic actinomycetes mainly includes standard plate colony counting, polymerase chain reaction (PCR), and enzyme-linked immunosorbent assay (ELISA). Colony counting is the standard method for bacterial detection, but it requires 24–72 hours, is labor-intensive, and time-consuming. ELISA and PCR have high sensitivity, shortening the detection time to a few hours, but cross-reactivity between antibodies can occur in ELISA reactions, and protein samples have short retention times. PCR is prone to false positives in bacterial cases, and both require multiple pretreatment steps, sophisticated instruments, and specialized operation. Therefore, relying on plate colony counting, ELISA, and PCR techniques for the detection of thermophilic actinomycetes is time-consuming and difficult. To overcome these shortcomings, it is necessary to develop methods for detecting thermophilic actinomycetes and spores with shorter analysis times and higher sensitivity.

[0004] Raman spectroscopy is a type of scattering spectroscopy. Based on the Raman scattering effect discovered by Indian scientist CV Raman, Raman spectroscopy analyzes the scattered spectra with frequencies different from the incident light to obtain information about molecular vibrations and rotations, and is applied to molecular structure research. Generally, Raman spectroscopy provides a unique chemical fingerprint for a specific molecule or material. In recent years, it has been discovered that Raman spectroscopy can be applied to the biological field. By analyzing single cells, the Raman spectrum of a single cell is a superposition of the vibrational modes of its intracellular components, composed of Raman peaks corresponding to different types of chemical bonds, reflecting multidimensional information about the composition and relative abundance of metabolites within a single cell. Raman spectroscopy can distinguish between bacteria and fungi, and can identify storage compounds and cytochromes in single microbial cells. Raman spectroscopy detection allows for label-free, rapid, and in-situ rapid detection of microorganisms at the species level. This method identifies strain types with high sensitivity, specificity, and accuracy, and compared to plate culture methods, it can also be used to analyze uncultured microorganisms. The advantage of traditional machine learning methods for identifying strains using Raman spectroscopy lies in handling binary classification problems. However, for multi-classification problems of thermophilic actinomycetes at the species level, corresponding strategies need to be designed to increase the complexity of modeling. This is necessary to achieve accurate identification of key microorganisms such as thermophilic actinomycetes and to determine the types and proportions of thermophilic actinomycetes in the fermentation process. This is of great significance for providing quality assurance measures for food brewing products and reducing pollution in the brewing industry. Summary of the Invention

[0005] Because conventional methods for identifying thermophilic actinomycetes are time-consuming and have poor specificity, they are not suitable for brewing companies to quickly identify the types and contents of thermophilic actinomycetes during fermentation to guide production. Therefore, it is necessary to develop a method for rapid identification and determination of the proportion of thermophilic actinomycetes in fermentation samples rich in thermophilic actinomycetes.

[0006] The purpose of this invention is to provide a method for detecting high-temperature actinomycetes based on Raman spectroscopy. This method pre-treats three high-abundance high-temperature actinomycetes in the traditional brewing process and then performs Raman detection. In the early stage, multiple machine learning algorithms are used to train the three high-temperature actinomycetes to find the three models with the highest scores. After assembling the models, an accurate classification standard can be given. Compared with other strain identification models, the strain identification time is shorter and the accuracy is higher.

[0007] In the fermentation processes of daqu (a type of starter culture), wine mash, feed, and compost, it is necessary to determine the types and content of thermophilic actinomycetes to guide production. This invention employs a special pretreatment method for complex, solid-state fermentation samples rich in thermophilic actinomycetes. Unlike conventional sample treatments, it utilizes gradient centrifugation, homogenization, and vortexing to prepare bacterial suspension samples. The resulting samples exhibit the highest biomass, extremely low background Raman noise, and a high signal-to-noise ratio in single-cell spectral analysis. This invention enables real-time, rapid, and label-free detection of thermophilic actinomycetes in fermentation samples, offering advantages such as high timeliness, high sensitivity, strong specificity, and the ability for large-scale online detection.

[0008] This invention provides a discriminant analysis classification model for identifying thermophilic actinomycete strains, the model being established according to the following steps:

[0009] (1) Preparation of thermophilic actinomycete suspension:

[0010] Thermoactinomyces vulgaris, Thermoactinomyces intermedius, and Thermoactinomyces daqus strains were inoculated into the culture medium for fermentation culture, and the culture broths with different culture times were collected to obtain thermophilic actinomycete suspensions.

[0011] (2) Pretreatment of thermophilic actinomycete suspension:

[0012] After centrifuging the bacterial suspension prepared in step (1), the precipitate was taken, sterile water was added to the precipitate and centrifuged again. This process was repeated 2 to 3 times to obtain the pretreated high-temperature actinomycete suspension.

[0013] (3) Preparation of gold nanoparticles:

[0014] Using the trisodium citrate heating reduction method, ultrapure water and potassium chloroaurate solution were added to a flask, mixed evenly, and heated to boiling. Then, trisodium citrate solution was quickly added and boiling was maintained. When the solution turned purple-red, heating was stopped to obtain gold nanoparticles with a diameter of 10-50 nm.

[0015] (4) Raman spectroscopy detection:

[0016] The thermophilic actinomycete suspension obtained in step (2) was mixed with the gold nanoparticles obtained in step (3), and then single-cell Raman spectroscopy was performed. The spectral acquisition conditions were as follows: a 532 nm laser was used, and the scanning spectrum was 500–3750 cm⁻¹. -1 The laser intensity was 1–300 mW, the acquisition time was 1–20 s / time, and the cumulative number of acquisitions was 1 time. 50–2000 cells were collected from the thermophilic actinomycete suspensions with different culture times.

[0017] (5) Processing of single-cell Raman spectroscopy data:

[0018] The single-cell Raman spectroscopy data obtained in step (4) are subjected to cosmic ray elimination, background noise removal, baseline correction, Savitzky-Golay smoothing, and normalization of all data.

[0019] (6) Model building:

[0020] Using machine learning to establish a discriminant analysis classification model for thermophilic actinomycete strains:

[0021] Characteristic bands identified: The characteristic spectra of *Thermoactinomyces vulgaris*, *Thermoactinomyces intermedius*, and *Thermoactinomyces daqus* are at 540 cm⁻¹. -1 564cm -1 570cm -1 618cm -1 712cm -1 748cm -1 766cm -1 784cm -1 820cm -1 866cm -1 868cm -1 898cm -1 926cm -1 964cm -1 996cm -1 1052cm -1 1096cm -1 1128cm -1 1164cm -1 1202cm -1 1288cm -1 1310cm -1 1344cm -1 1428cm -1 1476cm -1 1538cm -1 1582cm -1 1600cm -1 1622cm -1 1656cm -1 1662cm -1 ;

[0022] The obtained feature bands were machine learning using random forest, logistic regression, and support vector machine algorithms respectively. Training and detection datasets were set up, with the training dataset consisting of 70% of the collected data and the detection dataset consisting of 30% of the collected data. The trained models of random forest, logistic regression, and support vector machine algorithms were combined, and the Voting algorithm package was used as the framework for model combination.

[0023] The parameters for using the Random Forest algorithm for machine learning are: 100 decision trees, "gini" as the criterion for splitting nodes, 13 as the maximum tree depth, 110 as the minimum number of samples required to split internal nodes, 20 as the minimum number of samples required at leaf nodes, and the rest are default values.

[0024] The parameters for using the logistic regression algorithm for machine learning are: penalty term is l2, maximum number of iterations for algorithm convergence is 100, and the others are default values;

[0025] The parameters for machine learning using the support vector machine algorithm are: regularization parameter is l2, kernel function is "linear", kernel function parameter is 3, penalty coefficient is 100, and the others are default values.

[0026] In one embodiment of the present invention, the thermophilic actinomycetes are: Thermoactinomyces daqusCICC 10681T, Thermoactinomyces vulgaris DSM 43016, and Thermoactinomyces intermedius DSM 43816.

[0027] In one embodiment of the present invention, step (3) preparation of gold nanoparticles: using the trisodium citrate heating reduction method, ultrapure water and potassium chloroaurate solution with a concentration of 10 mg / mL are added to a flask, mixed evenly and heated to boiling, and then 0.1% trisodium citrate solution is quickly added and boiling is continued. When the solution is observed to turn purple-red, heating is stopped to obtain gold nanoparticles with a diameter of 10-50 nm.

[0028] In one embodiment of the present invention, the different culture times in step (1) are culture solutions obtained by culturing for 12h, 24h, and 48h.

[0029] In one embodiment of the present invention, in step (3) Raman spectroscopy detection, the laser intensity is 3mW, the acquisition time is 5s / time, the cumulative number of times is 1, and 100 cells are collected from the high-temperature actinomycete suspensions with different culture times.

[0030] In one embodiment of the present invention, in step (4), the bacterial suspension and gold nanoparticles are mixed and placed on a Raman chip, dried, and then Raman spectroscopy is performed; the volume ratio of gold nanoparticles to bacterial suspension is (1:1) to (1:1×10⁻⁶). 5 ).

[0031] In one embodiment of the present invention, the volume ratio of gold nanoparticles to bacterial suspension is 1:1.

[0032] In one embodiment of the present invention, the Raman chip is an aluminum-plated Raman chip.

[0033] This invention also provides a method for rapid identification and detection of thermophilic actinomycetes in fermentation products, the method comprising the following steps:

[0034] (1) Collect samples;

[0035] (2) Sample pretreatment:

[0036] Add the solution to the sample obtained in step (1), stir, shake with a homogenizer or vortex mixer, let stand, centrifuge and take the supernatant, centrifuge the obtained supernatant and take the precipitate.

[0037] After adding the precipitate to sterile water and centrifuging, repeat the process 2 to 3 times to obtain the pretreated sample.

[0038] (3) Preparation of gold nanoparticles:

[0039] Using the trisodium citrate heating reduction method, ultrapure water and a potassium chloroaurate solution with a concentration of 10 mg / mL were added to a flask, mixed evenly, and heated to boiling. Then, 0.1% trisodium citrate solution was quickly added, and boiling was continued. When the solution turned purple-red, heating was stopped to obtain gold nanoparticles with a diameter of 10–50 nm.

[0040] (4) Raman spectroscopy detection:

[0041] After mixing the sample obtained in step (2) with the gold nanoparticles obtained in step (3), single-cell Raman spectroscopy was performed. The spectral acquisition conditions were as follows: a 532 nm laser was used, and the scanning spectral range was 500–1800 cm⁻¹. -1 The laser intensity was 1–300 mW, the acquisition time was 1–20 s / time, and the cumulative number of acquisitions was 1 time. 50–2000 cells were collected from the thermophilic actinomycete suspensions with different culture times.

[0042] (5) Processing of single-cell Raman spectroscopy data:

[0043] The single-cell Raman spectroscopy data obtained in step (4) are subjected to cosmic ray elimination, background noise removal, baseline correction, Savitzky-Golay smoothing, and normalization of all data.

[0044] (6) Identification of thermophilic actinomycetes

[0045] The above discriminant analysis classification model is used to identify unknown strains. The characteristic band data of the single-cell Raman spectrum of the sample obtained in step (5) is input into the discriminant analysis classification model to identify the three thermophilic actinomycetes in the sample. According to the model strain classification standard, if the score is greater than or equal to 0.9, the strain with the highest score is identified as the strain; if the score is less than 0.9, it is identified as not belonging to thermophilic actinomycetes, but to other types of strains.

[0046] The characteristic bands of the single-cell Raman spectroscopy data are: 540 cm⁻¹, respectively. -1 564cm -1 570cm -1 618cm -1 712cm -1 748cm -1 766cm -1 784cm -1 820cm -1 866cm -1 868cm -1 898cm -1 926cm -1 964cm -1 996cm -1 1052cm -1 1096cm -1 1128cm -1 1164cm -1 1202cm -1 1288cm -1 1310cm -1 1344cm -1 1428cm -1 1476cm -1 1538cm -1 1582cm -1 1600cm -1 1622cm -1 1656cm -1 1662cm -1 ;

[0047] In one embodiment of the present invention, the thermophilic actinomycetes are: Thermoactinomyces vulgaris, Thermoactinomyces intermedius, and Thermoactinomyces daqus.

[0048] In one embodiment of the present invention, step (2) is: take 5g of the sample obtained in step (1) and add 20mL of solution, mix evenly and shake with a homogenizer or vortex mixer, let stand, centrifuge at 500rpm, and take the supernatant.

[0049] Centrifuge the supernatant at 7000 rpm and collect the precipitate; rinse the precipitate repeatedly with sterile water and centrifuge 3 times.

[0050] In one embodiment of the present invention, the sample in step (1) refers to a mixture of solid or liquid raw materials containing a variety of microorganisms. Block or powder fermentation samples are pulverized into powder using a mortar or grinder, and liquid fermentation samples are filtered using gauze.

[0051] In one embodiment of the present invention, the sample obtained in step (1) is added to a solution, stirred, and homogenized for 10 to 60 minutes at a speed of 3 to 12 times / second, or vortexed for 15 to 90 minutes, then allowed to stand for 2 to 5 minutes, the solution is aspirated, and the solution is centrifuged at 500 rpm for 2 minutes, and the supernatant is retained.

[0052] Centrifuge the supernatant at 7000 rpm for 2 min, discard the supernatant, add 6 mL of sterile water to the precipitate, mix well by pipetting, centrifuge at 7000 rpm for 2 min, discard the supernatant, add 6 mL of sterile water to the precipitate, mix well by pipetting. Rinse repeatedly with sterile water and centrifuge 3 times.

[0053] In one embodiment of the present invention, the sample includes, but is not limited to, daqu (a type of starter culture), fermented mash, brewed food, and fermented food.

[0054] The present invention also provides the application of the above method in detecting the relative content of thermophilic actinomycetes in fermentation samples.

[0055] In one embodiment of the present invention, the thermophilic actinomycetes are: *Thermoactinomyces vulgaris*, *Thermoactinomyces intermedius*, and *Thermoactinomyces daqus*. In one embodiment of the present invention, the application is:

[0056] Following the above method, 300 to 10,000 single-cell Raman spectra were collected from the microorganisms in the fermentation sample community.

[0057] The characteristic bands of the input single-cell Raman spectra are selected and input into the above-mentioned thermophilic actinomycete discrimination analysis and classification model. The unknown single-cell species in the fermentation sample are output. The total number of single-cell categories of the 300 to 10000 single-cell prediction results are counted. The ratio of the number of each thermophilic actinomycete to the total number of single cells is calculated to obtain the relative content of thermophilic actinomycetes in the microorganisms of the fermentation sample.

[0058] In one embodiment of the present invention, the percentage of thermophilic actinomycetes in the fermentation sample microorganisms is calculated. The total number of output single-cell categories is counted, and the ratio of the number of each thermophilic actinomycete to the total number of single cells is calculated to obtain the relative content of thermophilic actinomycetes in the fermentation sample microorganisms.

[0059] The calculation formula is:

[0060]

[0061] Where C(THn) represents the relative content of thermophilic actinomycetes in the microorganisms of the fermentation sample, n(THn) represents the number of thermophilic actinomycetes obtained by predicting unknown single cells using the model, and n represents the total number of single cells collected.

[0062] In one embodiment of the present invention, the percentage of thermophilic actinomycetes in the fermentation sample microorganisms requires a single-cell Raman spectroscopy collection quantity of 300 to 10,000 to calculate the proportion of thermophilic actinomycetes in the fermentation sample microbial community. A larger single-cell Raman spectroscopy collection quantity better represents the fermentation sample microbial community; in this embodiment, the collection quantity is 800 to 1000 single-cell Raman spectra.

[0063] Beneficial effects

[0064] (1) This invention uses gold nanoparticles to establish a Raman spectral standard library of thermophilic actinomycetes with different growth cycles. The thermophilic actinomycetes (including spores) in different fermentation cycles are all included in the thermophilic actinomycete Raman spectral standard library, which has strong adaptability. The prediction model of thermophilic actinomycetes can be reused. Once established, as long as the same Raman spectral measurement conditions are maintained, the single cell to be identified can be directly input into the model for identification.

[0065] (2) Complex substances in fermentation samples are prone to fluorescence Raman interference. By combining sample homogenization with gradient centrifugation, the direct separation of cereal and vegetable substances from microorganisms can be achieved. This invention is easier to use in actual production in fermentation plants, with low cost and simple operation.

[0066] (3) This invention uses an integrated model of the Raman spectral characteristic bands of high-temperature actinomycetes to predict strains. Compared with other strain identification models, the sample detection time can be shortened to about 1 hour, the accuracy can reach more than 90%, the strain identification time is shorter, and the accuracy is higher. Attached Figure Description

[0067] Figure 1 Raman spectra of three types of thermophilic actinomycetes in Example 1 of this invention.

[0068] Figure 2 The model confusion matrix obtained by training the ensemble model in Embodiment 1 of this invention. Detailed Implementation

[0069] The present invention will be further described below with reference to specific embodiments and accompanying drawings. It should be understood that the specific embodiments described herein are for illustrative and explanatory purposes only and are not intended to limit the present invention.

[0070] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0071] The microbial community structure described in the following examples refers to the collection of different single cells in a fermentation sample collected using single-cell Raman spectroscopy within a certain space.

[0072] The Daqu (fermentation starter), mash, compost, and feed involved in the following examples were sampled from: Maotai Town Winery, a local composting plant in Sichuan, and a feed mill in Jiangsu during their fermentation processes.

[0073] The Thermoactinomyces daqus CICC 10681T used in the following examples was purchased from the China Industrial Microbial Culture Collection Center.

[0074] Thermoactinomyces vulgaris DSM 43016 and Thermoactinomyces intermedius DSM 43816 mentioned in the following examples were purchased from the DSMZ Culture Bank in Germany.

[0075] Thermoactinomyces vulgaris, Thermoactinomyces intermedius, and Thermoactinomyces daqus mentioned in the following embodiments are designated as TV, TI, and TD, respectively.

[0076] The culture media involved in the following examples are as follows:

[0077] Modified culture medium: glucose 10 g / L, wheat extract 20 mL / L, sodium chloride 5 g / L, dipotassium hydrogen phosphate 4 g / L, potassium dihydrogen phosphate 2 g / L, magnesium sulfate heptahydrate 0.5 g / L, peptone 4 g / L, pH 7.0–7.2, sterilized at 115℃ for 30 min.

[0078] The following examples involve calculating the percentage of thermophilic actinomycetes in the fermentation samples:

[0079] The total number of output single-cell categories is counted, and the ratio of the number of each type of thermophilic actinomycete to the total number of single cells is calculated to obtain the relative content of thermophilic actinomycetes in the microorganisms of the fermentation sample.

[0080] The calculation formula is:

[0081]

[0082] Where C(THn) represents the relative content of thermophilic actinomycetes in the microorganisms of the fermentation sample, n(THn) represents the number of thermophilic actinomycetes obtained by predicting unknown single cells using the model, and n represents the total number of single cells collected.

[0083] Example 1: Construction of a Discriminant Analysis Classification Model

[0084] (1) Preparation of thermophilic actinomycete suspension

[0085] The strains Thermoactinomyces daqus CICC 10681T, Thermoactinomyces vulgaris DSM43016, and Thermoactinomyces intermedius DSM 43816 were inoculated into modified medium and cultured at 50℃ for 48 h. The culture broths of the thermophilic actinomycetes after 12 h, 24 h, and 48 h of culture were collected to obtain thermophilic actinomycete suspensions.

[0086] (2) Pretreatment of thermophilic actinomycete suspension:

[0087] 1 mL of cultured bacterial suspensions for 12 h, 24 h, and 48 h were respectively transferred to 1.5 mL centrifuge tubes, centrifuged at 7000 rpm for 2 min, and the supernatant was discarded. 500 μL of sterile water was added to the precipitate, and the mixture was mixed by pipetting. The mixture was then centrifuged at 7000 rpm for 2 min, and the supernatant was discarded. 500 μL of sterile water was added to the precipitate, and the mixture was mixed by pipetting. The above steps were repeated 3 times to prepare thermophilic actinomycete suspensions cultured for 6 h, 12 h, and 24 h, respectively.

[0088] (3) Preparation of gold nanoparticles:

[0089] Using the trisodium citrate heating reduction method, 47 mL of ultrapure water and 3 mL of potassium chloroaurate solution with a concentration of 10 mg / mL were added to a flask. After mixing evenly, the mixture was heated to boiling, and then 2 mL of 0.1% trisodium citrate solution was quickly added. The boiling was continued until the solution turned purple-red, at which point heating was stopped. The resulting gold nanoparticles had a diameter of 10–50 nm. After cooling to room temperature, the nanoparticles were stored at 4 °C in the dark.

[0090] (4) Raman spectroscopy detection

[0091] Take the bacterial suspensions obtained in step (2) after 12h, 24h, and 48h of culture, respectively, and mix them with gold nanoparticles at a 1:1 volume ratio. Add 2.5 μL of the mixture to an aluminized Raman chip and let it stand for 10 min to air dry. Collect single-cell Raman spectra using a confocal Raman spectrometer with an excitation wavelength of 532 nm and a scanning spectral range of 500–1800 cm⁻¹. -1 Microscopic power 3mW, scanning time 5s.

[0092] One hundred specimens of each thermophilic actinomycete were collected from bacterial suspensions cultured for 6 h, 12 h, and 24 h, respectively. Raman spectra were analyzed to remove cosmic rays, and baseline correction (iterative adaptive weighted penalized least squares method, airPLS), Savitzky-Golay smoothing, and normalization were performed to obtain a standard Raman spectral database. Figure 1 ).

[0093] (5) Using machine learning to establish a discriminant analysis classification model for thermophilic actinomycete strains: Using linear discriminant analysis, the following values ​​were identified for Thermoactinomyces daqus CICC 10681T, Thermoactinomyces vulgaris DSM43016, and Thermoactinomyces intermedius DSM 43816: 540, 564, 570, 618, 712, 748, 766, 784, 820, 866, 868, 898, 926, 964, 996, 1052, 1096, 1128, 1164, 1202, 1288, 1310, 1344, 1428, 1476, 1538, 1582, 1600, 1622, 1656, 1662 cm⁻¹. -1 Characteristic bands:

[0094] The obtained feature bands were machine learning using random forest, logistic regression, and support vector machine algorithms respectively. Training and detection datasets were set up, with the training dataset consisting of 70% of the collected data and the detection dataset consisting of 30% of the collected data. The trained models of random forest, logistic regression, and support vector machine algorithms were combined, and the Voting algorithm package was used as the framework for model combination.

[0095] The parameters for using the Random Forest algorithm for machine learning are: 100 decision trees, "gini" as the splitting criterion, 13 as the maximum tree depth, 110 as the minimum number of samples required to split internal nodes, 20 as the minimum number of samples required at leaf nodes, and the rest are default values. The parameters for using the Logistic Regression algorithm for machine learning are: l2 as the penalty term, 100 as the maximum number of iterations for convergence, and the rest are default values. The parameters for using the Support Vector Machine algorithm for machine learning are: l2 as the regularization parameter, "linear" as the kernel function, 3 as the kernel parameter, 100 as the penalty coefficient, and the rest are default values.

[0096] The model evaluation parameters and confusion matrices of three bacteria, *Thermoactinomyces vulgaris*, *Thermoactinomyces intermedius*, and *Thermoactinomyces daqus*, were obtained. Figure 2 The model obtained is the best.

[0097] Example 2: Identification of three thermophilic actinomycetes in Daqu (a type of Chinese liquor)

[0098] The fermentation starter culture (Daqu) for baijiu (Chinese liquor) is mostly made from wheat. The wheat is crushed with water, mixed with mother starter culture, pressed into blocks, and then stacked in a storage facility to form Daqu containing various bacteria and enzymes. Daqu plays a crucial role in baijiu brewing, including inoculating microorganisms, feeding grains, saccharification and fermentation, and aroma development. Researching the types of thermophilic actinomycetes can help better regulate the flavor compounds in baijiu, improve traditional processes, produce higher-quality baijiu, and enhance the overall level of baijiu brewing technology.

[0099] (1) Sample processing

[0100] Three types of Daqu samples were collected from different Daqu workshops. The Daqu blocks were crushed using a mortar and pestle to a size of 1-10 mm. After mixing the powders, 5 g was weighed and 20 mL of sterile water was added. The mixture was then homogenized in a homogenizer for 10 min at a speed of 10 times / second, with a 5-second interval between working and 5-second intervals, to ensure that the microorganisms were fully dispersed in the sterile aqueous solvent.

[0101] Then let it stand for 5 minutes, aspirate 6 mL of solution, centrifuge the solution at 500 rpm for 2 minutes, and retain the supernatant;

[0102] Centrifuge the supernatant at 7000 rpm for 2 min, discard the supernatant, add 6 mL of sterile water to the precipitate, mix well by pipetting, centrifuge at 7000 rpm for 2 min, discard the supernatant, add 6 mL of sterile water to the precipitate, mix well by pipetting. Rinse repeatedly with sterile water and centrifuge 3 times.

[0103] Pretreated Daqu samples 1 to 3 were prepared respectively.

[0104] Preparation of gold nanoparticles:

[0105] Using the trisodium citrate heating reduction method, 47 mL of ultrapure water and 3 mL of potassium chloroaurate solution with a concentration of 10 mg / mL were added to a flask. After mixing evenly, the mixture was heated to boiling, and then 2 mL of 0.1% trisodium citrate solution was quickly added. The boiling was continued until the solution turned purple-red, at which point heating was stopped. The resulting gold nanoparticles had a diameter of 10–50 nm. After cooling to room temperature, the nanoparticles were stored at 4 °C in the dark.

[0106] (2) Take 2.5 μL of pretreated Daqu sample 1-3 and mix it with gold nanoparticles at a 1:1 volume ratio on an aluminized Raman chip, let it stand for 10 min to air dry, and collect single-cell Raman spectra using a confocal Raman spectrometer with an excitation wavelength of 532 nm and a scanning spectral range of 500-1800 cm⁻¹. -1 The microscopic power was 3mW, the scanning time was 5s, and the cumulative number of scans was 1. Then, the data was processed to remove cosmic rays, perform baseline correction (iterative adaptive weighted penalized least squares method, airPLS), Savitzky-Golay smoothing, and normalize the Raman data to obtain the Raman spectra of Daqu microorganisms. 1000 single-cell Raman spectra were collected for each type of Daqu.

[0107] (3) The discriminant analysis classification model obtained in Example 1 above was used to identify the unknown strains. The characteristic bands of the single-cell Raman spectra of the samples obtained in step (2) were used to identify the unknown strains. -1 The sample can be identified by inputting the data into the discriminant analysis classification model obtained in Example 1. According to the model strain classification criteria, if the score is greater than or equal to 0.9, the strain with the highest score is identified as that strain; if the score is less than 0.9, it is identified as not belonging to thermophilic actinomycetes, but to other types of strains.

[0108] Based on the classification criteria for model strains, a discriminant model was used to identify 3000 collected thermophilic actinomycetes. The results are shown in Table 1.

[0109] Table 1: Proportion of thermophilic actinomycetes in different types of Daqu (a type of starter culture) in Example 2

[0110]

[0111] The results showed that by statistically analyzing the identification species of 1000 single cells in each Daqu (a type of starter culture), the proportion of thermophilic actinomycetes in the Daqu microbial community could be obtained.

[0112] Furthermore, using the method of this invention, the detection and analysis time for a single cell is 2 seconds, while the detection and analysis time for 1000 single cells in each Daqu (a type of Chinese liquor) is 2000 seconds, or 0.33 hours.

[0113] Example 3: Identification of three thermophilic actinomycetes in fermented mash

[0114] The brewing process of baijiu (Chinese liquor) involves a pile fermentation process. Pile fermentation refers to the process where raw materials or mash are cooked and gelatinized at high temperatures, then high-temperature koji (fermentation starter) is added, the mash is piled up, and left to stand on the ground for a period of time before being placed in fermentation pits. During the pile fermentation process, functional microorganisms are enriched. The identification of the types and contents of thermophilic actinomycetes during this process allows for better monitoring of the dynamic changes in microorganisms during pile fermentation.

[0115] (1) Sample processing

[0116] Two types of fermented mash samples were collected from the first and second batches of Maotai-flavor liquor. The mash was crushed using a mortar and pestle to a size of 1-10 mm. After mixing the powder, 5 g of the powder was weighed out and 20 mL of sterile water was added. The mixture was homogenized in a homogenizer for 10 min at a speed of 10 times / second for 5 seconds, followed by a 5-second interval, to ensure that the microorganisms were fully dispersed in the sterile aqueous solution.

[0117] Then let it stand for 5 minutes, aspirate 6 mL of solution, centrifuge the solution at 500 rpm for 2 minutes, and retain the supernatant;

[0118] Centrifuge the supernatant at 7000 rpm for 2 min, discard the supernatant, add 6 mL of sterile water to the precipitate, mix well by pipetting, centrifuge at 7000 rpm for 2 min, discard the supernatant, add 6 mL of sterile water to the precipitate, mix well by pipetting. Rinse repeatedly with sterile water and centrifuge 3 times.

[0119] Pretreated mash samples 1 and 2 were prepared respectively.

[0120] Preparation of gold nanoparticles:

[0121] Using the trisodium citrate heating reduction method, 47 mL of ultrapure water and 3 mL of potassium chloroaurate solution with a concentration of 10 mg / mL were added to a flask. After mixing evenly, the mixture was heated to boiling, and then 2 mL of 0.1% trisodium citrate solution was quickly added. The boiling was continued until the solution turned purple-red, at which point heating was stopped. The resulting gold nanoparticles had a diameter of 10–50 nm. After cooling to room temperature, the nanoparticles were stored at 4 °C in the dark.

[0122] (2) After pretreatment, samples 1-2 of the fermented mash were mixed with gold nanoparticles at a 1:1 volume ratio. 2.5 μL of the mixture was then placed on an aluminized Raman chip, allowed to stand for 10 min to air dry, and single-cell Raman spectra were acquired using a confocal Raman spectrometer. The excitation wavelength was 532 nm, and the scanning spectral range was 500–3750 cm⁻¹. -1 The microscopic power was 3mW, the scanning time was 5s, and the cumulative number of scans was 1. Then, the data was processed to remove cosmic rays, perform baseline correction (iterative adaptive weighted penalized least squares method, airPLS), Savitzky-Golay smoothing, and normalize the Raman data to obtain the Raman spectra of Daqu microorganisms. 1000 single-cell Raman spectra were collected for each mash.

[0123] (3) The unknown strains were identified using the discriminant analysis classification model described above. The characteristic bands (540, 564, 570, 618, 712, 748, 766, 784, 820, 866, 868, 898, 926, 964, 996, 1052, 1096, 1128, 1164, 1202, 1288, 1310, 1344, 1428, 1476, 1538, 1582, 1600, 1622, 1656, 1662 cm⁻¹) of the single-cell Raman spectra of the samples obtained in step (2) were selected. -1 The sample can be identified by inputting the data into the discriminant analysis classification model obtained in Example 1. According to the model strain classification criteria, if the score is greater than or equal to 0.9, the strain with the highest score is identified as that strain; if the score is less than 0.9, it is identified as not belonging to thermophilic actinomycetes, but to other types of strains.

[0124] Based on the classification criteria for model strains, a discriminant model was used to identify 2000 collected thermophilic actinomycetes. The results are shown in Table 2.

[0125] Table 2: Proportion of thermophilic actinomycetes in different types of fermented mash in Example 3

[0126]

[0127] The results showed that by statistically analyzing the identification species of 1000 single cells from each mash, the proportion of thermophilic actinomycetes in the mash microbial community could be obtained.

[0128] Furthermore, using the method of this invention, the detection and analysis time for a single cell is 2 seconds, while the detection and analysis time for 1000 single cells per batch of fermented mash is 2000 seconds, or 0.33 hours.

[0129] Example 4: Identification of three thermophilic actinomycetes in compost

[0130] Composting is a process in which organic matter is degraded by aerobic microorganisms under aerated conditions. Studies have shown that thermophilic actinomycetes are an important component of the microbial community in composting, thus enabling rapid identification of thermophilic actinomycetes in compost.

[0131] (1) Sample processing

[0132] Two compost samples were collected from a local chicken manure aerobic composting plant in Chengdu, Sichuan. The samples were filtered through three layers of sterile gauze. After filtration, 5g of the liquid was mixed and weighed out. 20mL of sterile water was added, and the mixture was homogenized and shaken for 10 minutes at a speed of 10 times / second for 5 seconds, followed by a 5-second interval, to ensure that the microorganisms were fully dispersed in the sterile aqueous solution.

[0133] Then let it stand for 5 minutes, aspirate 6 mL of solution, centrifuge the solution at 500 rpm for 2 minutes, and retain the supernatant;

[0134] Centrifuge the supernatant at 7000 rpm for 2 min, discard the supernatant, add 6 mL of sterile water to the precipitate, mix well by pipetting, centrifuge at 7000 rpm for 2 min, discard the supernatant, add 6 mL of sterile water to the precipitate, mix well by pipetting. Rinse repeatedly with sterile water and centrifuge 3 times.

[0135] Pretreated compost samples 1 and 2 were prepared respectively.

[0136] (2) Take 2.5 μL of pretreated compost sample 1–2 and mix it with gold nanoparticles at a 1:1 volume ratio. Then, take 2.5 μL of the mixture into an aluminized Raman chip, let it stand for 10 min to air dry, and collect single-cell Raman spectra using confocal Raman spectroscopy. The excitation wavelength is 532 nm, and the scanning spectral range is 500–3750 cm⁻¹. -1 The microscopic power was 3mW, the scanning time was 5s, and the cumulative number of scans was 1. Then, the data was processed to remove cosmic rays, perform baseline correction (iterative adaptive weighted penalized least squares method, airPLS), Savitzky-Golay smoothing, and normalize the Raman data to obtain the Raman spectra of Daqu microorganisms. 800 single-cell Raman spectra were collected for each compost.

[0137] (3) The unknown strains were identified using the discriminant analysis classification model described above. The characteristic bands (540, 564, 570, 618, 712, 748, 766, 784, 820, 866, 868, 898, 926, 964, 996, 1052, 1096, 1128, 1164, 1202, 1288, 1310, 1344, 1428, 1476, 1538, 1582, 1600, 1622, 1656, 1662 cm⁻¹) of the single-cell Raman spectra of the samples obtained in step (2) were selected. -1 The sample can be identified by inputting the data into the discriminant analysis classification model obtained in Example 1. According to the model strain classification criteria, if the score is greater than or equal to 0.9, the strain with the highest score is identified as that strain; if the score is less than 0.9, it is identified as not belonging to thermophilic actinomycetes, but to other types of strains.

[0138] Based on the classification criteria for model strains, a discriminant model was used to identify 1600 collected thermophilic actinomycetes. The results are shown in Table 3.

[0139] Table 3: Proportion of thermophilic actinomycetes in different types of compost in Example 4

[0140]

[0141] The results showed that the proportion of thermophilic actinomycetes in the compost microbial community could be obtained by statistically analyzing the identified species of 800 single cells in each compost.

[0142] Furthermore, using the method of this invention, the detection and analysis time for a single cell is 2 seconds, while the detection and analysis time for 1000 single cells in each compost is 2000 seconds, or 0.33 hours.

[0143] Example 5: Identification of three thermophilic actinomycetes in fermented feed

[0144] Moldy feed can produce toxins that cause poisoning in animals, which may then harm humans through the food chain. Thermostable actinomycetes are one of the strains that cause feed spoilage, so the identification of thermophilic actinomycetes in feed plays an important role in controlling feed quality.

[0145] (1) Sample processing

[0146] Four types of feed samples were collected from a livestock and poultry feed factory in Wuxi, Jiangsu Province. The feed was pulverized using a mortar and pestle to a size of 1-10 mm. After mixing the powders, 5 g of each sample was weighed and 20 mL of sterile water was added. The mixture was then homogenized in a homogenizer for 10 min at a speed of 10 times / second, with a 5-second interval between each 5-second interval, to ensure that the microorganisms were fully dispersed in the sterile aqueous solution.

[0147] Then let it stand for 5 minutes, aspirate 6 mL of solution, centrifuge the solution at 500 rpm for 2 minutes, and retain the supernatant;

[0148] Centrifuge the supernatant at 7000 rpm for 2 min, discard the supernatant, add 6 mL of sterile water to the precipitate, mix well by pipetting, centrifuge at 7000 rpm for 2 min, discard the supernatant, add 6 mL of sterile water to the precipitate, mix well by pipetting. Rinse repeatedly with sterile water and centrifuge 3 times.

[0149] Pretreated feed samples 1–4 were prepared respectively.

[0150] Preparation of gold nanoparticles:

[0151] Using the trisodium citrate heating reduction method, 47 mL of ultrapure water and 3 mL of potassium chloroaurate solution with a concentration of 10 mg / mL were added to a flask. After mixing evenly, the mixture was heated to boiling, and then 2 mL of 0.1% trisodium citrate solution was quickly added. The boiling was continued until the solution turned purple-red, at which point heating was stopped. The resulting gold nanoparticles had a diameter of 10–50 nm. After cooling to room temperature, the nanoparticles were stored at 4 °C in the dark.

[0152] (2) After pretreatment, feed samples 1-4 were mixed with gold nanoparticles at a 1:1 volume ratio, and 2.5 μL was placed on an aluminized Raman chip. The mixture was allowed to stand for 10 min and then air-dried. Single-cell Raman spectra were collected using a confocal Raman spectrometer with an excitation wavelength of 532 nm and a scanning spectral range of 500–3750 cm⁻¹. -1 The power under the microscope was 3mW, the scanning time was 5s, and the cumulative number of scans was 1. Then, the Raman data was processed to remove cosmic rays, perform baseline correction (iterative adaptive weighted penalized least squares method, airPLS), Savitzky-Golay smoothing, and normalization to obtain the Raman spectra of feed microorganisms. 1000 single-cell Raman spectra were collected for each feed sample.

[0153] (3) The unknown strains were identified using the discriminant analysis classification model described above. The characteristic bands (540, 564, 570, 618, 712, 748, 766, 784, 820, 866, 868, 898, 926, 964, 996, 1052, 1096, 1128, 1164, 1202, 1288, 1310, 1344, 1428, 1476, 1538, 1582, 1600, 1622, 1656, 1662 cm⁻¹) of the single-cell Raman spectra of the samples obtained in step (2) were selected. -1 The sample can be identified by inputting the data into the discriminant analysis classification model obtained in Example 1. According to the model strain classification criteria, if the score is greater than or equal to 0.9, the strain with the highest score is identified as that strain; if the score is less than 0.9, it is identified as not belonging to thermophilic actinomycetes, but to other types of strains.

[0154] Based on the classification criteria for model strains, a discriminant model was used to identify 4000 collected thermophilic actinomycetes. The results are shown in Table 4.

[0155] Table 4: Proportion of thermophilic actinomycetes in different types of feed in Example 5

[0156]

[0157] The results showed that by statistically analyzing the identification species of 1000 single cells from each feed, the proportion of thermophilic actinomycetes in the feed microbial community could be obtained.

[0158] Furthermore, using the method of this invention, the detection and analysis time for a single cell is 2 seconds, while the detection and analysis time for 1000 single cells per feed is 2000 seconds, or 0.33 hours.

[0159] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make various modifications and alterations without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be determined by the claims.

Claims

1. A discriminant analysis classification model for identifying thermophilic actinomycete strains, characterized in that, The model was established according to the following steps: (1) Preparation of thermophilic actinomycete suspension: Each Thermoactinomyces vulgaris , Thermoactinomyces intermedius , Thermoactinomyces daqus The strain was inoculated into the culture medium for fermentation culture, and the culture medium was taken at different times of 12 h, 24 h and 48 h to obtain the bacterial suspension. (2) Pretreatment of thermophilic actinomycete suspension: After centrifuging the bacterial suspension prepared in step (1), take the precipitate, add sterile water to the precipitate and centrifuge again. Repeat this process 2-3 times to obtain the pretreated high-temperature actinomycete suspension. (3) Preparation of gold nanoparticles: Using the trisodium citrate heating reduction method, ultrapure water and potassium chloroaurate solution were added to a flask, mixed evenly, and heated to boiling. Then, trisodium citrate solution was quickly added and boiling was continued. When the solution turned purple-red, heating was stopped to obtain gold nanoparticles with a diameter of 10~50 nm. (4) Raman spectroscopy detection: The thermophilic actinomycete suspension obtained in step (2) was mixed with the gold nanoparticles obtained in step (3) at a 1:1 volume ratio, and then single-cell Raman spectroscopy was performed. The spectral acquisition conditions were as follows: a 532 nm laser was used, and the spectral range was 500~1800 cm⁻¹. -1 The grating was 600g, the laser intensity was 1~300 mW, the acquisition time was 1~20 s / time, and the cumulative number of acquisitions was 1 time. 50~2000 cells were collected from the thermophilic actinomycete suspensions with different culture times. (5) Processing of single-cell Raman spectroscopy data: The single-cell Raman spectroscopy data obtained in step (4) are subjected to cosmic ray elimination, background noise removal, baseline correction, Savitzky-Golay smoothing, and normalization of all data. (6) Model building: Using machine learning to establish a discriminant analysis classification model for thermophilic actinomycete strains: Identify characteristic bands: Thermoactinomyces vulgaris , Thermoactinomyces intermedius , Thermoactinomyces daqus The characteristic spectra are at 540 cm⁻¹. -1 564 cm -1 570 cm -1 618 cm -1 712cm -1 748 cm -1 766 cm -1 784 cm -1 820 cm -1 866 cm -1 868 cm -1 898 cm -1 926 cm -1 964cm -1 996 cm -1 1052 cm -1 1096 cm -1 1128 cm -1 1164 cm -1 1202 cm -1 1288 cm -1 1310 cm -1 1344 cm -1 1428 cm -1 1476 cm -1 1538 cm -1 1582 cm -1 1600 cm -1 1622 cm -1 1656 cm -1 1662 cm -1 ; The obtained feature bands were machine learned using random forest, logistic regression, and support vector machine algorithms respectively. Training and detection datasets were set up, with the training dataset consisting of 70% of the collected data and the detection dataset consisting of 30% of the collected data. The trained models of random forest, logistic regression, and support vector machine algorithms were combined, and the Voting algorithm package was used as the framework for model combination. The parameters for using the random forest algorithm for machine learning are: 100 decision trees, "gini" as the criterion for splitting nodes, 13 as the maximum tree depth, 110 as the minimum number of samples required to split internal nodes, and 20 as the minimum number of samples required at leaf nodes. The parameters for using the logistic regression algorithm for machine learning are: a penalty term of l2, and a maximum number of iterations for the algorithm to converge to 100. The parameters for machine learning using the support vector machine algorithm are: regularization parameter l2, kernel function "linear", kernel function parameter 3, and penalty coefficient 100.

2. The discriminant analysis classification model as described in claim 1, characterized in that, The thermophilic actinomycetes are: Thermoactinomyces daqus CICC 10681T, Thermoactinomyces vulgaris DSM 43016 and Thermoactinomyces intermedius DSM 43816.

3. The discriminant analysis classification model as described in claim 1, characterized in that, In step (4), the bacterial suspension was mixed with gold nanoparticles and placed on a Raman chip. After drying, Raman spectroscopy was performed.

4. The discriminant analysis classification model as described in claim 3, characterized in that, In step (4) Raman spectroscopy detection, the laser intensity was 3 mW, the acquisition time was 5 s / time, and the cumulative number of times was 1. 100 cells were collected from the high-temperature actinomycete suspensions with different culture times.

5. A method for rapid identification and detection of thermophilic actinomycetes in fermentation products, characterized in that, The method includes the following steps: (1) Collect samples; (2) Sample pretreatment: Add the solution to the sample obtained in step (1), stir, shake with a homogenizer or vortex mixer, let stand, centrifuge and take the supernatant, centrifuge the obtained supernatant and take the precipitate. After adding the precipitate to sterile water and centrifuging, repeat 2-3 times to obtain the pretreated sample. (3) Preparation of gold nanoparticles: Using the trisodium citrate heating reduction method, ultrapure water and potassium chloroaurate solution were added to a flask, mixed evenly, and heated to boiling. Then, trisodium citrate solution was quickly added and boiling was maintained. When the solution turned purple-red, heating was stopped to obtain gold nanoparticles with a diameter of 10~50 nm. (4) Raman spectroscopy detection: After mixing the sample obtained in step (2) with the gold nanoparticles obtained in step (3), single-cell Raman spectroscopy was performed. The spectral acquisition conditions were as follows: a 532 nm laser was used, and the scanning spectral range was 500~1800 cm⁻¹. -1 The laser intensity was 1~300mW, the acquisition time was 1~20 s / time, and the cumulative number of times was 1. 50~2000 cells were collected from the thermophilic actinomycete suspensions with different culture times. (5) Processing of single-cell Raman spectroscopy data: The single-cell Raman spectroscopy data obtained in step (4) are subjected to cosmic ray elimination, background noise removal, baseline correction, Savitzky-Golay smoothing, and normalization of all data. (6) Identification of thermophilic actinomycetes: Identification of unknown strains using the discriminant analysis classification model described in any one of claims 1 to 4: Input the characteristic band data of the single-cell Raman spectrum of the sample obtained in step (5) into the discriminant analysis classification model to identify the three thermophilic actinomycetes in the sample; according to the model strain classification standard, if the score is greater than or equal to 0.9, the strain with the highest score is identified as the strain; if the score is less than 0.9, it is identified as not belonging to thermophilic actinomycetes, but to other types of strains; The characteristic bands of the single-cell Raman spectral data are: 540 cm⁻¹. -1 564 cm -1 570 cm -1 618 cm -1 712 cm -1 748 cm -1 766 cm -1 784 cm -1 820 cm -1 866 cm -1 868 cm -1 898 cm -1 926 cm -1 964 cm -1 996 cm -1 1052 cm -1 1096 cm -1 1128 cm -1 1164 cm -1 1202 cm -1 1288 cm -1 1310cm -1 1344 cm -1 1428 cm -1 1476 cm -1 1538 cm -1 1582 cm -1 1600 cm -1 1622 cm -1 1656 cm -1 1662 cm -1 .

6. The method as described in claim 5, characterized in that, The thermophilic actinomycetes are: Thermoactinomyces vulgaris , Thermoactinomyces intermedius and Thermoactinomyces daqus .

7. The application of the method according to any one of claims 5 to 6 in detecting the relative content of thermophilic actinomycetes in fermentation samples.

8. The application as described in claim 7, characterized in that, The thermophilic actinomycetes are: Thermoactinomyces vulgaris , Thermoactinomyces intermedius and Thermoactinomyces daqus .

Images