A method for discriminating daqu categories based on microbial structural composition

Abstract

The present application belongs to the technical field of Daqu identification, and relates to a method for constructing a Daqu category discrimination method based on microbial structure composition, which comprises the following steps: S1 obtaining microbial composition information in a Daqu sample; S2 analyzing the obtained microbial variable information and the category of the corresponding Daqu sample by a partial least squares method, screening and optimizing the microbial composition information; S3 simplifying the index obtained after screening and optimization in step S2 by a stepwise discriminant analysis method to obtain a modeling index; and S4 taking the modeling index as the independent variable and the category of the Daqu sample as the dependent variable, and performing discriminant analysis by a Fisher discriminant analysis method to construct the discriminant model. Based on the method, efficient and accurate classification and evaluation of Daqu are realized, and the method has high accuracy in discriminating Daqu. The constructed Daqu category discrimination method can also provide theoretical support for the dataization and systemization of the brewing process.

Description

Technical Field

[0001] This invention belongs to the field of Daqu identification technology, specifically relating to a method for classifying Daqu based on the composition of microbial structures. Background Technology

[0002] High-temperature Daqu (a type of starter culture) is the fermentation initiator for Maotai-flavor Baijiu (a type of Chinese liquor) and plays a crucial role in the brewing process. Daqu is classified into different types based on its storage period and environment, making the classification of different types of Daqu a hot topic in Maotai-flavor Baijiu research. Currently, the identification of different types of Daqu mainly relies on traditional methods, which are fast, convenient, and applicable, but require highly skilled evaluators. Extensive and repeated training is needed to build an accurate and efficient sensory evaluation team to ensure stable quality control results. High-throughput sequencing, as a routine experimental technique, is widely used to analyze the microbial community structure in brewing systems. Numerous reports have been published on high-throughput sequencing results for Maotai-flavor Baijiu Daqu, but reports on establishing efficient and accurate Daqu classification methods based on differences in microbial composition are scarce. Therefore, this invention aims to establish a high-temperature Daqu classification method based on microbial structural composition. Summary of the Invention

[0003] The purpose of this invention is to provide a method for classifying Daqu (a type of starter culture) based on its microbial structure. This invention employs high-throughput sequencing to detect the microbial composition of Daqu. Using different types of Daqu as samples, 10 OTUs (Optical Characteristic Units) are selected as modeling indicators using PLS-DA combined with stepwise discriminant analysis. Then, Fisher discriminant analysis is used to construct a Daqu classification model, achieving efficient and accurate classification of Daqu. This model achieves a classification accuracy of 96.7%. By establishing this Daqu classification method, theoretical support can be provided for the data-driven and systematic development of the brewing process.

[0004] On the one hand, the present invention provides a method for constructing a method for distinguishing the type of Daqu (a type of starter culture) based on the structural composition of microorganisms, the method comprising the following steps:

[0005] S1 obtains information on the microbial composition of Daqu samples;

[0006] S2 analyzes the obtained microbial variable information and the corresponding Daqu sample category using partial least squares method to screen and optimize the microbial composition information;

[0007] S3 simplifies the indicators obtained after screening and optimization in step S2 using stepwise discriminant analysis to obtain modeling indicators;

[0008] S4 uses the modeling index as the independent variable and the category of the Daqu sample as the dependent variable, and performs discriminant analysis using the Fisher discriminant analysis method to construct a discriminant model.

[0009] In the prior art CN104372075A, which is closest to this invention, a method for constructing a discriminant model to identify the quality of Daqu (a type of starter culture) is disclosed. This method directly uses the indicators obtained by screening and optimizing the microbial information obtained through partial least squares method as modeling indicators, and uses the discriminant model constructed by the quadratic discriminant analysis method based on R language to distinguish between finished Daqu, yellow Daqu and white Daqu.

[0010] Based on this, the present invention further improves and explores the method for classifying Daqu (a type of starter culture). After screening and optimizing the obtained microbial information using partial least squares (PLS), it further explores that simplifying the screened and optimized indicators using stepwise discriminant analysis significantly reduces the number of indicators required to construct the discriminant model. At the same time, the accuracy of the discriminant model constructed based on the reduced indicators in classifying Daqu remains unaffected, reaching as high as 96.7%. Furthermore, in exploring methods to reduce the number of indicators required for modeling, various methods were tried, but not all analytical methods that can simplify indicators can achieve the effects achievable by the present invention. Secondly, the present invention constructs a discriminant model based on the reduced indicators using Fisher discriminant analysis in SPSS 20.0 software. The accuracy of this model in classifying Daqu is significantly higher than that of the discriminant model constructed using the quadratic discriminant analysis method based on R language in existing technologies.

[0011] In some implementations, in step S1, the category information of the Daqu sample includes: Daqu for normal production after leaving the warehouse and / or Daqu with severe insect infestation after leaving the warehouse.

[0012] In some implementations, obtaining the microbial composition information includes the following steps:

[0013] The microbial composition information of the Daqu sample was obtained by sequencing analysis using high-throughput sequencing technology.

[0014] In some implementations, the sequencing analysis of the Daqu sample using high-throughput sequencing technology includes the following steps:

[0015] The Daqu sample was subjected to DNA extraction, purification, and PCR amplification.

[0016] Using the 454 sequencing platform and employing 16S rDNA hypervariable region sequencing technology, the PCR products of the variable region V4-V5 of the 16S rDNA in the Daqu sample were sequenced.

[0017] The sequencing data was processed and quality controlled. Sequencing adapters and low-quality reads were removed, and the data were assembled to obtain the microbial composition information of the Daqu sample.

[0018] In some implementations, in step S2, the partial least squares method is performed through the Lianchuan Bio Cloud Platform; the principle for screening and optimizing the obtained microbial composition information indicators is that the variable projection importance index (VIP) value is ≥1.

[0019] In some implementations, step S3, the stepwise discriminant analysis is performed using SPSS 20.0 software.

[0020] In some implementations, in step S4, the Fisher discriminant analysis method is implemented using SPSS 20.0 software.

[0021] On the other hand, the present invention also provides a method for distinguishing the types of Daqu (a type of starter culture) based on the microbial structural composition, the method comprising the following steps:

[0022] (1) Obtain information on the microbial composition of the Daqu to be identified;

[0023] (2) Input the obtained microbial composition information into the discrimination model to obtain the category information of the Daqu to be discriminated.

[0024] In some implementations, in step (1), the microbial composition information includes 10 OTUs; the specific information of the 10 OTUs is as follows:

[0025]

[0026] In some implementations, the category information of the Daqu to be determined includes Daqu for normal production and / or Daqu with severe insect infestation.

[0027] In summary, this application includes at least one of the following beneficial technical effects:

[0028] (1) This invention uses high-throughput sequencing to detect the microbial composition in Daqu. Using different types of Daqu as samples, 10 OTUs are screened out as modeling indicators by PLS-DA combined stepwise discriminant analysis. Then, Fisher discriminant analysis is used to construct a Daqu classification and discrimination model, which realizes efficient and accurate classification and evaluation of Daqu. The discrimination model can achieve a discrimination accuracy of up to 96.7% for Daqu.

[0029] (2) The classification and discrimination method of Daqu established by the present invention can provide theoretical support for the data-driven and systematic brewing process. Attached Figure Description

[0030] Figure 1 This is a distribution chart of VIP values ​​for the 59 OTUs obtained by screening in Embodiment 1 of the present invention. Detailed Implementation

[0031] The following specific embodiments further illustrate the technical solution of the present invention. These specific embodiments do not represent a limitation on the scope of protection of the present invention. Non-essential modifications and adjustments made by others based on the concept of the present invention still fall within the scope of protection of the present invention.

[0032] Example 1: A method for classifying Daqu (a type of starter culture) based on microbial structural composition

[0033] The material samples used in this embodiment came from a sauce-flavored liquor production company in Guizhou Province, including 23 batches of starter culture for normal production and 37 batches of starter culture with severe insect infestation.

[0034] I. Obtaining information on the microbial composition of Daqu samples

[0035] The above-mentioned Daqu samples were sequenced and analyzed using high-throughput sequencing technology to obtain information on the microbial composition of all Daqu samples. Specifically, the following steps were performed: DNA was extracted and purified from the Daqu samples, followed by PCR amplification. The resulting samples were sent to the Megamicro 454GS FLX Titanium sequencing platform. Using 16S rDNA hypervariable region sequencing technology, the PCR products of the 16S rDNA variable regions V4-V5 of each Daqu sample were sequenced. The sequencing reads were processed and quality-controlled, removing sequencing adapters and low-quality reads before assembly. The assembly steps included:

[0036] (1) Use MEGAHIT (V1.0.6): https: / / github.com / voutcn / megahit to collect CleanData; then break the assembled Scaffold at N connection points, leaving Scafftigs without N connections; use Megahit to compare the Clean Data of all samples with each Scafftig to obtain the unused PEreads;

[0037] (2) After merging all the unused reads from the previous step, use MEGAHIT software to perform mixed assembly with the same parameters as single sample assembly; break the mixed Scaffolds at N connection points to obtain Scaftigs; screen out fragments less than 500bp from all Scaftigs for statistical analysis.

[0038] Using the above method, 295 OTU microbial composition information were finally obtained from the Daqu samples.

[0039] II. Screening and Optimization of Microbial Composition Information

[0040] The microbial composition information and corresponding Daqu sample categories obtained above were analyzed using partial least squares analysis (PLS-DA) on the Lianchuan Bio Cloud Platform (https: / / www.omicstudio.cn / tool). The results are shown in Table 1. Figure 1 As shown. According to Figure 1 The results showed that by using the above PLS-DA analysis method, the obtained microbial composition information indicators were screened and optimized, with the principle that the variable projection importance index (VIP) value ≥1. Finally, a total of 59 OTUs were obtained as indicators for subsequent construction of the discriminant model.

[0041] Table 1. Data information table corresponding to 59 OTUs

[0042]

[0043]

[0044] III. Simplification of Microbiological Indicators

[0045] Furthermore, using the above method, 59 indicators were finally selected. To further simplify the model, the 59 OTUs selected in step two were further optimized using stepwise discriminant analysis. This was done using SPSS 20.0 software, and the specific operation was as follows:

[0046] 1. Import the data of the 59 selected microorganisms and their corresponding Daqu sample classification information into SPSS 20.0 software;

[0047] 2. Click “Analyze” - “Classify” - “Discriminative”;

[0048] 3. Select "Category" as the grouping variable and set the definition range. Select the remaining indicators as independent variables and click "Use Step Method" below.

[0049] 4. Click "Statistics" on the right, and check "Average" and "Fischer" and "Unstandardized" under "Function Coefficients";

[0050] 5. Click the "Category" button on the right, and check the "Summary Table" and the "Merge Group" and "Domain Map" below the figure;

[0051] 6. Click the "Save" button on the right, check "Predict Group Members", and finally click OK to perform the calculation. The results are shown in Table 2.

[0052] Table 2. Simplified index results of stepwise discriminant analysis.

[0053]

[0054] Based on the results presented in Table 2, OTU_14, OTU_4, OTU_5, OTU_9, OTU_16, OTU_13, OTU_24, OTU_83, OTU_98, and OTU_65 from the Daqu sample were ultimately selected as the modeling indicators for constructing the subsequent discriminant model. The specific information for the 10 OTUs obtained above is shown in Table 3.

[0055] Table 3. OTU Species Classification Table (Modeling)

[0056]

[0057] IV. Establishing a discriminant model

[0058] Using SPSS 20.0 software and Fisher's discriminant analysis method, with the 10 types of microbial composition data obtained above as independent variables and the types of Daqu samples (normal production Daqu and severely insect-infested Daqu) as dependent variables, a discriminant analysis was conducted to construct a discriminant model for distinguishing Daqu types.

[0059] Table 4 Fisher's discrimination results

[0060]

[0061]

[0062]

[0063] Table 5. Cross-validation results of modeled samples

[0064]

[0065] Note: CK grade koji is for normal production use; T grade koji is for severely insect-infested koji.

[0066] As can be seen from Table 5 above, a total of 60 Daqu samples were used for modeling, including 23 Daqu samples for normal production after leaving the warehouse and 37 Daqu samples with severe insect infestation after leaving the warehouse. The discrimination model above achieved an accuracy rate of 58 / 60 = 96.7% in identifying the type of Daqu samples, indicating that the discrimination model established above is quite effective in identifying the type of Daqu.

[0067] V. Identification

[0068] The discrimination model constructed above was used to distinguish the types of 27 external test Daqu samples. These 27 external test Daqu samples included 8 samples for normal production use and 19 samples with severe insect infestation. The specific discrimination process is as follows:

[0069] Following the method described in step one, we obtained the microbial information of 10 OTUs from 27 external test Daqu samples, specifically including: OTU_14, OTU_4, OTU_5, OTU_9, OTU_16, OTU_13, OTU_24, OTU_83, OTU_98, and OTU_65;

[0070] Substituting the obtained microbial information into the discrimination model constructed in step three, the output result is the type of Daqu (fermented starter culture) of the tested Daqu sample. The Daqu type includes Daqu for normal production after leaving the warehouse and Daqu with severe insect infestation after leaving the warehouse. The obtained discrimination results are shown in Table 6.

[0071] Table 6 Identification Results

[0072]

[0073]

[0074] As can be seen from Table 6, the discrimination accuracy of the discrimination model constructed above for external test samples is 18 / 19 = 94.7%, which further indicates that the established discrimination model has a good discrimination effect on the Daqu type.

[0075] Comparative Example 1

[0076] Comparative Example 1 is based on the previous example, but differs from Example 1 in that: in Example 1, step three, which optimizes the 59 OTUs selected in step two, uses stepwise discriminant analysis based on SPSS, while Comparative Example 1 uses PCA analysis, as detailed below:

[0077] Based on the 59 microbial indicators screened in step two of Example 1, the modeling indicators were further optimized. Principal component analysis (PCA) was performed on the 59 microbial indicators obtained through the Lianchuan Bio Cloud Platform (https: / / www.omicstudio.cn / tool), and the results are shown in Table 7.

[0078] Table 7 PCA Principal Component Loadings

[0079]

[0080]

[0081] As can be seen from the results in Table 7 above, principal component analysis and factor analysis were performed on 59 microbial indicators, and 59 principal components were obtained. Among them, the cumulative contribution rate of PC1-PC5 reached 91.57%. It can be seen from the principal component loadings that PCA analysis failed to screen out characteristic indicators from the 59 indicators.

[0082] Comparative Example 2

[0083] Comparative Example 2 is based on Example 1, but differs from Example 1 in that: in Example 1, the optimization of the 59 OTUs selected in Step 2 using SPSS-based stepwise discriminant analysis was employed in Step 3, while Comparative Example 1 uses analysis of variance, as detailed below:

[0084] Based on the 59 OTU indicators initially screened in step two of Example 1, further modeling indicators were selected. The 59 OTU indicators were analyzed and optimized using the analysis of variance method in SPSS 20.0 software. The results are shown in Table 8.

[0085] Table 8. Significance Indicators for Analysis of Variance

[0086]

[0087]

[0088]

[0089] According to the analysis of variance results in Table 8 above, further analysis and optimization of the 59 OTU indicators using the analysis of variance method revealed that 34 OTU indicators showed significant differences (p<0.05), which failed to achieve the goal of narrowing the range of indicators.

[0090] Comparative Example 3

[0091] Comparative Example 3 is based on Example 1, but differs from Example 1 in that: in step four of Example 1, the method used to establish the discriminant model was as follows: using SPSS 20.0 software and Fisher's discriminant analysis, the 10 types of microbial data obtained in step three of Example 1 were used as independent variables, and the types of koji used for normal production and koji with severe insect infestation were used as dependent variables to construct a discriminant model for distinguishing koji types; while in Comparative Example 3, using SPSS 20.0 software and Fisher's discriminant analysis, the 59 OTU data obtained in step two of Example 1 were used as independent variables, and the types of koji used for normal production and koji with severe insect infestation were used as dependent variables to construct a discriminant model for distinguishing koji types. The results are shown in Tables 9 and 10.

[0092] Table 9. Fisher discrimination results for 59 OTU microbial indicators.

[0093]

[0094]

[0095]

[0096]

[0097] Table 10 Cross-validation results of 59 OTU microbial indicator modeling samples

[0098]

[0099] Note: CK grade koji is for normal production use; T grade koji is for severely insect-infested koji.

[0100] Table 10 shows that a total of 60 samples were used for modeling: 23 samples of koji for normal production and 37 samples of koji with severe insect infestation. The discrimination model constructed using the method in Example 1 achieved a discrimination accuracy of 58 / 60 = 96.7% for the above-mentioned koji samples. In contrast, the modeling in Comparative Example 3, which directly used 59 OTUs of microbial data as indicators to distinguish the types of koji samples, achieved the same accuracy as the modeling using 10 OTUs. Therefore, the indicator optimization method in Example 1 not only simplified the microbial information indicators in the modeling process, but also achieved the same accuracy as the final discrimination model.

[0101] Comparative Example 4

[0102] Comparative Example 4 is based on Example 1, but differs from Example 1 in that: in step four of Example 1, SPSS 20.0 was used in conjunction with Fisher's discriminant analysis, with the 10 OTU microbial data obtained in step three of Example 1 as independent variables and normal production koji and severely insect-infested koji as dependent variables, to construct a discriminant model for koji type identification; in Comparative Example 4, R language software was used in conjunction with secondary discriminant analysis, with the 10 OTU microbial data obtained in step three of Example 1 as independent variables and normal production koji and severely insect-infested koji as dependent variables, to construct a discriminant model for koji type identification. The results are shown in Table 11.

[0103] Table 11. Cross-validation results of samples modeled by quadratic discriminant analysis.

[0104]

[0105]

[0106] Note: CK grade koji is for normal production use; T grade koji is for severely insect-infested koji.

[0107] Table 11 shows that a total of 60 samples were used for modeling: 23 samples of koji for normal production and 37 samples of koji with severe insect infestation. The accuracy of the discrimination model constructed using the method in Comparative Example 4 in identifying the type of koji samples was 55 / 60 = 91.7%. Comparative analysis shows that the discrimination model constructed using R language software and the quadratic discriminant analysis method in Comparative Example 4 significantly reduced the accuracy of identifying the type of koji samples.

[0108] It is understood that the present invention has been described through some embodiments, and those skilled in the art will recognize that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the invention. Furthermore, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of the present invention.

Claims

1. A method for constructing a system for distinguishing the types of Daqu (a type of starter culture) based on microbial structural composition, characterized in that, The method includes the following steps: S1 obtains information on the microbial composition of OTUs in Daqu samples; S2 analyzes the obtained OTU microbial composition information and the corresponding Daqu sample category using partial least squares method to screen and optimize the OTU microbial composition information; S3 simplifies the indicators obtained after screening and optimization in step S2 using stepwise discriminant analysis to obtain modeling indicators; S4 uses the modeling index as the independent variable and the category of the Daqu sample as the dependent variable, and performs discriminant analysis using the Fisher discriminant analysis method to construct a discriminant model.

2. The method as described in claim 1, characterized in that, In step S1, the category information of the Daqu sample includes: Daqu for normal production after leaving the warehouse and / or Daqu with severe insect infestation after leaving the warehouse.

3. The method as described in claim 1, characterized in that, The acquisition of the OTU microbial composition information includes the following steps: The OTU microbial composition information of the Daqu sample was obtained by sequencing analysis using high-throughput sequencing technology.

4. The method as described in claim 3, characterized in that, The sequencing analysis of the Daqu sample using high-throughput sequencing technology includes the following steps: The Daqu sample was subjected to DNA extraction, purification, and PCR amplification. Using the 454 sequencing platform and employing 16S rDNA hypervariable region sequencing technology, the PCR products of the 16S rDNA variable region V4-V5 of the Daqu sample were sequenced. After sequencing, the data were processed and quality controlled. Sequencing adapters and low-quality reads were removed, and the data were assembled to obtain the OTU microbial composition information of the Daqu sample.

5. The method as described in claim 1, characterized in that, In step S2, the partial least squares method is performed through the Lianchuan Bio Cloud Platform; the principle for screening and optimizing the OTU microbial composition information is that the variable projection importance index (VIP) value is ≥1.

6. The method as described in claim 1, characterized in that, In step S3, the stepwise discriminant analysis method is implemented using SPSS 20.0 software.

7. The method as described in claim 1, characterized in that, In step S4, the Fisher discriminant analysis method is implemented using SPSS 20.0 software.

8. A method for classifying Daqu (a type of starter culture) based on the structural composition of microorganisms, characterized in that, The discrimination method includes the following steps: (1) Obtain information on the microbial composition of the Daqu to be identified; (2) Input the obtained microbial composition information into the discrimination model as described in any one of claims 1-7 to obtain the category information of the Daqu to be discriminated; In step (1), the microbial composition information includes 10 OTUs; the specific information of the 10 OTUs is as follows: 。 9. The discrimination method as described in claim 8, characterized in that, The category information of the Daqu to be judged includes Daqu for normal production after leaving the warehouse and / or Daqu with severe insect infestation after leaving the warehouse.

Images