A method and system for constructing and tailoring a microbial community interaction network and core microbial community markers and applications

The flora interaction network was constructed through probability statistical models and the method was cut, which solved the false positive problem of the study of interbacterial interaction relationships, and successfully discovered the core flora markers of children with functional constipation, achieving more stable network construction and efficient Fungus mining effect.

CN118471344BActive Publication Date: 2024-11-15DIPROBIO (SHANGHAI) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410623361.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-20
Publication Date
2024-11-15
Estimated Expiration
2044-05-20

AI Technical Summary

Technical Problem

When studying the network of interaction between bacteria, the prior art is susceptible to technical parameters such as sample collection, preservation and transportation status, resulting in false positives and false negatives, affecting the discovery and stability of core bacterial flora.

Method used

Using the method of constructing a microbial interaction network based on probability statistical model, the microbial composition structure is obtained through sequencing, the correlation relationship between bacteria is calculated, the initial microecological association network is constructed, and the core microbial structure is found through the cutting method, which is applied to functionality. Research on microecology of children with constipation.

Benefits of technology

A more stable microbial network construction was achieved, the risk of false positives was reduced, 25 core microbial markers related to functional constipation were successfully discovered, and their effectiveness was verified in machine learning models, reaching 85.2% AUC.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118471344B_ABST
    Figure CN118471344B_ABST
Patent Text Reader

Abstract

The present invention provides a method and system for constructing and trimming a microbiota interaction network and an application of a core microbiota structure. The present invention constructs the microbiota composition data of a group by collecting information on the composition of intestinal microorganisms in the group, and compares the co-discovery frequency between bacteria with the theoretical frequency based on probability to determine the beneficial and harmful associations between bacteria and construct a complex network of bacterial interactions. Further, by iterative trimming and optimization of the microbiota interaction network, the core bacterial composition in the microecology of the group is finally determined. The microbiota network analysis study of children with functional constipation using this algorithm found 25 core microbiota markers related to constipation, and the classification model of the 25 markers was verified by a machine learning algorithm to distinguish between children with constipation and normal children, and the AUC of the model can reach 85.2%. The present invention mines the core microbiota markers of a specific population through a network construction and analysis algorithm, providing a new method for clinical application research of intestinal microecology.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of microecological analysis, and in particular to a microbial community interaction network construction, a tailoring method and system, and a core microbial community marker and application. Background Art

[0002] As the second largest organ of humans, intestinal microorganisms play an important role in human health. In recent years, data based on animal studies and clinical studies have proved that intestinal flora is closely related to the occurrence and development of diseases. Intestinal flora participates in the dynamic regulation of human health in terms of forming intestinal barriers, participating in metabolism, and maintaining immune balance. The human intestinal microorganisms are composed of 1,000 to 1,500 species of bacteria or fungi. The specific relationships between these bacteria, including mutualistic symbiosis and inhibitory competition, remain to be studied and explored. Therefore, a deep and comprehensive analysis of the correlation between bacteria is very important for future research on microorganisms and human health. It can not only promote future microecological research, but also promote the clinical transformation of microecological research results.

[0003] At present, most of the network studies on the interaction between bacteria are based on the correlation of bacterial abundance in specific data sets, such as Pearson correlation and Spearman correlation. Then, the core bacterial community structure and characteristics in a specific group are discovered based on the constructed correlation network. However, this method is based on the correlation of bacterial abundance between a small number of samples, and some important bacterial community relationships discovered may no longer exist in other data sets or large data sets. In addition, this type of bacterial community association algorithm is easily affected by technical parameters such as detection algorithms and sample storage and transportation status, which makes the abundance of a certain type of bacteria too high or too low, thereby affecting the false positive and false negative of the relationship between bacterial communities, and then affecting the discovery of core bacterial communities. Therefore, a stable and effective bacterial community network construction method and core bacterial community mining method are urgently needed. Summary of the invention

[0004] Purpose of the invention: The technical problem to be solved by the present invention is to provide a method and system for constructing a microbial interaction network, and to determine the association between bacteria based on a probabilistic co-detection model between microbial communities to construct a microbial community network; the technical problem to be solved by the present invention is to provide a method and system for trimming a complex microecological association network to discover the core microbial community structure in the microecological network.

[0005] The technical problem that the present invention also aims to solve is to provide core flora markers in the intestinal microecological interaction network of children with functional constipation.

[0006] The technical problem that the present invention needs to solve is to provide a kit that can identify functional constipation in children.

[0007] Technical solution: To solve the above technical problems, the present invention provides a method for constructing a microbial interaction network based on a probability statistical model, and on the basis of network construction, a method for pruning a complex network of microecological associations is proposed to mine the core microbial flora structure of the microecological association network in a specific population. In addition, 25 core microbial flora markers in the microecology of children with functional constipation are mined by applying this method.

[0008] On the one hand, the present invention provides a method for constructing a microbial interaction network based on a probability statistical model, including the following steps:

[0009] The first step: Obtain the microbial composition structure of each individual and the microbial distribution matrix of a population composed of N samples by sequencing;

[0010] The second step: Calculate the proportion P of each bacterium detected in the population to obtain the prevalence of each species in the population;

[0011] The third step: For two bacteria A and B, calculate the number of times F they are co-detected in the population A,B , and the expected number of times Exp that bacteria A and bacteria B are co-detected in the independent state A,B =N*P A *P B ;

[0012] The fourth step: Calculate the difference Delta between the expected number of times that bacteria are co-detected in the independent state and the observed number of times that they are co-detected A,B =Exp A,B -F A,B , and calculate the average M and variance SD of the difference;

[0013] The fifth step: Determine the association relationship according to the average difference and one-fold variance. If the magnitude of Delta A,B is between M - SD and M + SD, it is considered that there is no association between bacteria A and bacteria B. If Delta A,B <M–SD, it is considered that there is a positive association between bacteria A and bacteria B. If Delta A,B >M + SD, it is considered that there is a negative association between bacteria A and bacteria B;

[0014] The sixth step: Define the microecological association network of the population according to the association relationship determined in the fifth step. For the bacteria with positive or negative associations, the association strength is the absolute value of Delta A,B .

[0015] On the other hand, the present invention provides a method for pruning a complex network of microecological associations, including the following steps:

[0016] Step 1: construct an initial undirected microecological association network according to the microbial community interaction network construction method based on the probability statistical model;

[0017] Step 2: Remove the edges with the highest and lowest association strengths in the network respectively, and calculate the eigenvector centrality Eig of the remaining network structure 大 and Eig 小 ;

[0018] Step 3: If Eig 大 Greater than Eig 小 , the network removes the edge with the highest association strength, and vice versa, removes the edge with the lowest association strength, and simultaneously calculates the proximity centrality Clo and betweenness centrality Bet of the network structure;

[0019] Step 4: Continue to loop from step 2 to step 3 until only the preset number of edges remain in the network;

[0020] Step 5: Compare and analyze the proximity centrality Clo and betweenness centrality Bet after subtracting one edge each time, and find the optimal step S for both opt ;

[0021] Step 6: Prune the initial network to the optimal step S opt The final network is determined as the pruned core network, and the bacteria connected by the edges in the core network constitute the core flora.

[0022] As a preference, the optimal step S is determined according to the following rules: opt : Use proximity centrality Clo and betweenness centrality Bet as two-dimensional coordinate axes to draw a graph, and take the trimming step corresponding to the point in the upper right corner area as the optimal step.

[0023] Based on the same inventive concept, the present invention also provides a bacterial community interaction network construction system based on a probabilistic statistical model, comprising:

[0024] The data acquisition module is used to obtain the bacterial flora composition structure of each individual and the bacterial flora distribution matrix of a group composed of N samples through sequencing;

[0025] The probability calculation module is used to calculate the proportion P of each bacterium detected in the group and obtain the prevalence of each species in the group; for two bacteria A and B, calculate the number of times they are detected together in the group F A,B , and the expected number of times bacteria A and bacteria B are detected together in an independent state Exp A,B =N*P A *P B ; and, calculate the difference between the expected number of times bacteria are detected together in an independent state and the observed number of times they are detected together A,B= Exp A,B -F A,B and calculate the average value M and variance SD of the difference;

[0026] An association calculation module, configured to determine an association relationship according to the average difference and one-fold variance. If Delta A,B is between M - SD and M + SD, it is considered that there is no association between bacterium A and bacterium B. If Delta A,B < M – SD, it is considered that there is a positive association between bacterium A and bacterium B. If Delta A,B > M + SD, it is considered that there is a negative association between bacterium A and bacterium B;

[0027] And a network construction module, configured to determine the microecological association network of the population according to the definition of the association relationship. For the association strength between positively or negatively associated bacteria, it is the absolute value of Delta A,B .

[0028] Based on the same inventive concept, the present invention also provides a trimming system for a microecological association complex network, including:

[0029] An initial network construction module, configured to construct an initial undirected human intestinal microecological association network according to the construction method of the human intestinal microecological association network described above;

[0030] A trimming exploration module, configured to respectively remove the edge with the highest association strength and the edge with the lowest association strength in the network, and calculate the eigenvector centrality Eig 大 and Eig 小 of the remaining network structure; if Eig 大 is greater than Eig 小 , the edge with the highest association strength is removed from the network. Otherwise, the edge with the lowest association strength is removed. At the same time, the closeness centrality Clo and betweenness centrality Bet of the network structure are calculated; continuously loop and subtract one edge each time until only a preset number of edges remain in the network;

[0031] And an optimal trimming determination module, configured to compare and analyze the closeness centrality Clo and betweenness centrality Bet after subtracting one edge each time, and find the optimal step S opt ; trim the initial network to the optimal step S opt , and determine the final network as the trimmed core network. The bacteria connected by the edges in the core network form the core flora.

[0032] The present invention also provides a computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of any of the above methods are implemented.

[0033] The present invention also provides a microbial core flora marker obtained by the method for cutting the complex network of human intestinal microecological associations.

[0034] The core flora markers include any one or a combination of Catabacter, Olsenella, Actinomyces, Anaerovoracaceae Family XIII AD3011 group, Abiotrophia, Phascolarctobacterium, Akkermansia, Atopobium, Bacteroides, Acidaminococcus, Neoscardovia, Clostridiumsensu stricto 1, Actinomycetaceae_F0332, Christensenellaceae R-7group, Pseudoscardovia, Mogibacterium, Coriobacteriaceae UCG-002, Family XIII UCG-001, Anaerofustis, Butyricicoccus, [Eubacterium]nodatum group, Granulicatella, Acetobacter, S5-A14a or Barnesiella.

[0035] The present invention includes a kit for identifying functional constipation in children. The kit includes a detection reagent, and the detection reagent includes a microbial flora marker in a detection sample to determine whether the child is constipated. The microbial flora marker includes Catabacter, Olsenella, Actinomyces, Anaerovoracaceae Family XIII AD3011group, Abiotrophia, Phascolarctobacterium, Akkermansia, Atopobium, Bacteroides, Acidaminococcus, Neoscardovia, Clostridium sensu stricto 1, Actinomycetaceae_F0332, Christensenellaceae R-7group, Pseudoscardovia, Mogibacterium, Coriobacteriaceae UCG-002, Family XIII UCG-001, Anaerofustis, Butyricicoccus, [Eubacterium]nodatum group, Granulicatella, Acetobacter, S5-A14a or Barnesiella or a combination of them.

[0036] Wherein, the samples include but are not limited to fecal samples.

[0037] The children are aged between 2 and 5 years old.

[0038] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0039] 1) The present invention proposes for the first time a method for constructing a bacterial community network based on a probability model, which can solve the problem that the traditional network construction model is defined based on the correlation between the abundance of different bacteria, which is easily affected by technical parameters such as sample collection method, storage time, and transportation status, resulting in many false positive correlations. The method provided by the present invention ignores the abundance of the bacteria itself when defining the correlation, and only considers whether the bacteria appear. The core network structure constructed is more stable and less prone to false positives.

[0040] 2) This invention proposes for the first time a method for optimizing and trimming the complex network of microecological flora to find the core flora structure in the microecology of a specific population.

[0041] 3) The present invention applied the above two algorithms to the actual discovery of the core flora of children with functional constipation, successfully found 25 core flora related to functional constipation, and applied the 25 core flora markers to verify in the machine learning model, obtaining an AUC of 85.2%. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a network diagram of the intestinal flora of children with functional constipation built based on a probability model. Each node in the diagram represents a genus of bacteria. Light red indicates a positive correlation between two bacteria, and green indicates a negative correlation.

[0043] Figure 2 Distribution graphs of the betweenness centrality and proximity centrality of the network at each step in the network pruning process.

[0044] Figure 3 This is a diagram of the core flora structure and relationships in children with functional constipation (dots represent core bacterial genera, red lines represent the symbiotic relationship between two bacterial genera, and green represents the competitive relationship).

[0045] Figure 4 AUC validation results of the classification model constructed using 25 core bacterial genera as markers. DETAILED DESCRIPTION

[0046] Example 1

[0047] This example is explained by taking the mining of the core flora structure of the intestinal microecology of children with functional constipation as an example.

[0048] This embodiment discloses a method for constructing a microbial community interaction network based on a probabilistic statistical model, which mainly includes processing the original sequencing data and constructing a microbial community-related network based on a probabilistic model. The specific steps are as follows:

[0049] Step 1: Obtain the microbial composition structure of each individual and the microbial distribution matrix of a group consisting of N samples through sequencing.

[0050] In this example, fecal 16S rRNA sequencing data of 61 children with functional constipation were selected, and the sequencing data were processed based on the 16S amplicon analysis process of DADA2 (https: / / benjjneb.github.io / dada2 / tutorial.html). The operation steps are as follows:

[0051] (1) First, the 16S rRNA sequence of each sample was obtained by removing the linker and splitting, and then quality filtering, noise removal, merging and chimera removal were performed to obtain an ASV (Amplicon Sequence Variant) sequence with 100% consistency.

[0052] (2) Align the ASV sequences to the Silva database (v138) to obtain the species names of the sequences.

[0053] (3) Integrate the names of different reference sequences, merge the sequence abundances annotated to the same species into the abundance of the same species, and obtain the microbiota composition data of the samples used.

[0054] Merge the microbiota composition structure data of each sample calculated into the microbiota composition structure matrix of the population, and remove the results that are not identified at the genus level.

[0055] Step 2: Calculate the proportion P of each bacterium detected in children with functional constipation to obtain the prevalence of each species in children with functional constipation, that is, if the abundance of a certain bacterium in the sample is greater than 0, then the bacterium is detected.

[0056] Step 3: For any two bacteria A and B, calculate the number of times F they are co-detected in children with functional constipation A,B , and calculate the expected number of times Exp co-detected under the independent conditions of bacteria A and bacteria B A,B = N * P A * P B ;

[0057] Step 4: Calculate the difference Delta between the observed number of co-detections between bacteria and the theoretically expected number of detections under independent conditions A,B = Exp A,B - F A,B , and calculate the average M and variance SD of the difference.

[0058] Step 5: Determine the association relationship according to the range of the average difference and one-fold variance, as follows:

[0059] If the magnitude of Delta A,B is between M - SD and M + SD, it is considered that there is no association between bacteria A and bacteria B; if Delta A,B < M – SD, it is considered that there is a positive association between bacteria A and bacteria B; if Delta A,B > M + SD, it is considered that there is a negative association between bacteria A and bacteria B.

[0060] Step 6: Construct a gut microbiota-related network for children with functional constipation according to the association relationship defined in Step 5, as Figure 1 shown. The association strength between bacteria with positive or negative associations in the network is the absolute value of Delta A,B .

[0061] Example 2

[0062] This embodiment discloses a method for cutting a microecological complex network. Based on the intestinal flora-related network constructed in Example 1, the microecological flora complex network optimization cutting method is used to find the core flora structure in the microecology of a specific group. The steps are as follows:

[0063] Step 1: construct an initial undirected complex network of intestinal flora of children with functional constipation according to the method of Example 1. The size of the edge in the network is the degree of association between two bacteria, that is, |Delta A,B |.

[0064] Step 2: Regardless of whether the correlation between the two bacteria is positive or negative, remove the edge with the highest and lowest correlation strength in the network, and calculate the eigenvector centrality Eig of the remaining network structure. 大 and Eig 小 Here the eigenvector centrality of the network structure is defined as:

[0065] Eig=Σ v (max(eig v )-eig v )

[0066] Among them, v is a node in the network, eig v is the eigenvector centrality of node v in the network, which measures the value of the point based on the importance of its neighboring points. max(eig v ) represents the maximum value of the eigenvector centrality of all nodes.

[0067] Step 3: If Eig 大 Greater than Eig 小 , remove the edge with the highest association strength in the network, and vice versa, remove the edge with the lowest association strength, and at the same time calculate the proximity centrality Clo and betweenness centrality Bet of the network structure.

[0068] Here the proximity centrality of the network structure is defined as:

[0069] Clo=Σ v (max(clo v )-clo v )

[0070] Among them, clo v is the proximity centrality of node v in the network, which measures the inverse of the sum of the distances between the point and other points in the network. The larger the value, the closer it is to the center and the faster it can reach other points. max(clo v ) represents the maximum value of the proximity centrality of all nodes.

[0071] Here the betweenness centrality of the network structure is defined as:

[0072] Bet=∑v (max(bet v )-bet v )

[0073] Among them, bet v is the betweenness centrality of node v in the network, indicating whether the shortest distance passes through this point. If so, it means that this point is very important. max(bet v ) represents the maximum value of the betweenness centrality of all nodes.

[0074] Step 4: Continue to loop from step 2 to step 3 until there are only 5 edges left in the network.

[0075] Step 5: Compare and analyze the proximity centrality Clo and betweenness centrality Bet after subtracting one edge each time, and find the optimal step S opt ; Find the optimal steps according to the changes of proximity centrality Clo and betweenness centrality Bet, such as Figure 2 As shown, this embodiment uses proximity centrality Clo and betweenness centrality Bet as two-dimensional coordinate axes to draw a graph, and the trimming step corresponding to the point in the upper right corner area is the optimal step, such as step 204.

[0076] Step 6: Prune the initial network to the optimal S opt , and determine the final network as the pruned core network, such as Figure 3 The core intestinal flora structure of 25 children with functional constipation was obtained, as shown in Table 1.

[0077] Table 1 Figure 2 The number of each node in the

[0078]

[0079] Example 3

[0080] This embodiment is based on the same inventive concept as the first embodiment, and discloses a bacterial community interaction network construction system based on a probabilistic statistical model, including:

[0081] The data acquisition module is used to obtain the bacterial flora composition structure of each individual and the bacterial flora distribution matrix of a group composed of N samples through sequencing;

[0082] The probability calculation module is used to calculate the proportion P of each bacterium detected in the group and obtain the prevalence of each species in the group; for two bacteria A and B, calculate the number of times they are detected together in the group F A,B , and the expected number of times bacteria A and bacteria B are detected together in an independent state Exp A,B =N*P A *P B; and calculating the difference Delta between the expected number of times co-detected in the independent state between bacteria and the observed number of times co-detected A,B = Exp A,B - F A,B , and calculating the average M and variance SD of the differences;

[0083] An association calculation module, configured to determine an association relationship according to the average difference and one-fold variance. If the magnitude of Delta A,B is between M - SD and M + SD, it is considered that there is no association between bacterium A and bacterium B. If Delta A,B < M – SD, it is considered that there is a positive association between bacterium A and bacterium B. If Delta A,B > M + SD, it is considered that there is a negative association between bacterium A and bacterium B;

[0084] and a network construction module, configured to determine the microecological association network of the population according to the definition of the association relationship. For the association strength between positively or negatively associated bacteria, it is the absolute value of Delta A,B .

[0085] Example 4

[0086] This example is based on the same inventive concept as Example 1 and discloses a trimming system for a microecological association complex network, including:

[0087] An initial network construction module, configured to construct an initial undirected human intestinal microecological association network according to the construction method of the human intestinal microecological association network described above;

[0088] A trimming exploration module, configured to remove the edge with the highest association strength and the edge with the lowest association strength in the network respectively, and calculate the eigenvector centrality Eig 大 and Eig 小 ; if Eig 大 is greater than Eig 小 , the edge with the highest association strength is removed from the network. Otherwise, the edge with the lowest association strength is removed. At the same time, the closeness centrality Clo and betweenness centrality Bet of the network structure are calculated; continuously loop and subtract one edge each time until only a preset number of edges remain in the network;

[0089] and an optimal trimming determination module, configured to compare and analyze the closeness centrality Clo and betweenness centrality Bet after subtracting one edge each time, and find the optimal step S opt ; trim the initial network to the optimal step S opt , and determine the final network as the trimmed core network. The bacteria connected by the edges in the core network constitute the core flora.

[0090] Example 5

[0091] This embodiment discloses a computer program product, including a computer program / instruction, which implements the steps of the above-mentioned embodiment 1 or 2 when executed by a processor.

[0092] Example 6 Prediction of children's constipation symptoms based on the core flora of children with functional constipation

[0093] This example uses the 25 core bacteria detected in Example 2 as characteristic bacteria to evaluate their effectiveness in distinguishing children with functional constipation from normal children. The main steps include data collation, model training, and model evaluation.

[0094] 1. Data collation: The abundance data of the above 25 core strains in 61 children with functional constipation and the abundance data of the 25 core strains in 61 normal children were selected to form a complete data set used in this example;

[0095] 2. Model training: Split the data in the first step into training set and test set in a ratio of 8.5:1.5. Without going through the feature selection stage, directly train on the training set and then verify on the test set.

[0096] 3. Model training: Using the 25 core bacteria in the second step training set as features, a binary classification model was trained to distinguish between children with functional constipation and normal children. The logistic regression model was selected for model training. According to the probability output by the model, a probability greater than 50% was selected as allergy, and a probability less than or equal to 50% was selected as normal.

[0097] 4. Model evaluation: Evaluate the trained model on the test set. The AUC of the model on the test set is as follows: Figure 3 It was proved that the use of these 25 core bacteria as markers can achieve a good effect of distinguishing children with functional constipation from normal children, and the model AUC reached 85.2%.

Claims

1. A method for constructing a microbial community interaction network based on a probabilistic statistical model, characterized in that: The following steps are involved: Step 1: Obtain the bacterial community structure of each individual and the bacterial community distribution matrix of a group consisting of N samples through sequencing; Step 2: Calculate the proportion P of each bacterium detected in the population to obtain the prevalence of each species in the population; Step 3: For two bacteria A and B, calculate the number of times they are detected together in the population F A,B , and the expected number of times bacteria A and bacteria B are detected together in an independent state Exp A,B =N*P A *P B ; Step 4: Calculate the difference between the expected number of times bacteria are detected together in an independent state and the observed number of times they are detected together. A,B =Exp A,B -F A,B , and calculate the mean M and variance SD of the difference; Step 5: Determine the correlation based on the difference mean and one standard deviation. If Delta A,B is between M - SD and M + SD, it is considered that there is no correlation between bacterium A and bacterium B. If Delta A,B < M – SD, it is considered that there is a positive correlation between bacterium A and bacterium B. If Delta A,B > M + SD, it is considered that there is a negative correlation between bacterium A and bacterium B; Step 6: Determine the microecological association network of the group based on the association relationship defined in step 5. The association strength between positively or negatively associated bacteria is Delta. A,B The absolute value of Step 7: Remove the edges with the highest and lowest association strengths in the network respectively, and calculate the eigenvector centrality Eig of the remaining network structure 大 and Eig 小 ; The definition of the eigenvector centrality of the network structure is: Eig=∑ v (max(eig v )-eig v ) Among them, v is a node in the network, eig v is the eigenvector centrality of node v in the network, max(eig v ) represents the maximum value of the eigenvector centrality of all nodes; Step 8: If Eig 大 Greater than Eig 小 , the network removes the edge with the highest association strength, and vice versa, removes the edge with the lowest association strength, and calculates the proximity centrality Clo and betweenness centrality Bet of the network structure at the same time; the proximity centrality of the network structure is defined as: Clo=Σ v (max(clo v )-clo v ) Among them, clo v is the proximity centrality of node v in the network, max(Clo v ) represents the maximum value of the proximity centrality of all nodes; The definition of betweenness centrality of a network structure is: Bet=Σ v (max(bet v )-bet v ) Among them, bet v is the betweenness centrality of node v in the network, max(bet v ) represents the maximum value of the betweenness centrality of all nodes; Step 9: Continue to repeat steps 7 to 8 until only the preset number of edges remain in the network; Step 10: Compare and analyze the proximity centrality Clo and betweenness centrality Bet after subtracting one edge each time, and find the optimal step S for both opt ; Step 11: Prune the initial network to the optimal step S opt The final network is determined as the pruned core network, and the bacteria connected by the edges in the core network constitute the core bacterial community; Determine the optimal step S according to the following rules opt : Use proximity centrality Clo and betweenness centrality Bet as two-dimensional coordinate axes to draw a graph, and take the trimming step corresponding to the point in the upper right corner area as the optimal step.

2. A bacterial community interaction network construction system based on a probabilistic statistical model, characterized in that: include: The data acquisition module is used to obtain the bacterial flora composition structure of each individual and the bacterial flora distribution matrix of a group composed of N samples through sequencing; The probability calculation module is used to calculate the proportion P of each bacterium detected in the group and obtain the prevalence of each species in the group; for two bacteria A and B, calculate the number of times they are detected together in the group F A,B , and the expected number of times bacteria A and bacteria B are detected together in an independent state Exp A,B =N*P A *P B ; and, calculate the difference between the expected number of times bacteria are detected together in an independent state and the observed number of times they are detected together A,B =Exp A,B -F A,B , and calculate the mean M and variance SD of the difference; An association calculation module, which is used to determine the association relationship according to the average difference and the standard deviation. If Delta A,B is between M - SD and M + SD, it is considered that there is no association between bacterium A and bacterium B. If Delta A,B < M - SD, it is considered that there is a positive association between bacterium A and bacterium B. If Delta A,B > M + SD, it is considered that there is a negative association between bacterium A and bacterium B; And, the network construction module is used to determine the microecological association network of the group according to the definition of the association relationship. The association strength between positively or negatively associated bacteria is Delta A,B The absolute value of The trimming exploration module is used to remove the edges with the highest and lowest association strengths in the network and calculate the eigenvector centrality Eig of the remaining network structure. 大 and Eig 小 ; If Eig 大 Greater than Eig 小 , remove the edge with the highest association strength from the network, and vice versa, remove the edge with the lowest association strength, and calculate the proximity centrality Clo and betweenness centrality Bet of the network structure at the same time; continue to cycle and subtract one edge each time until only the preset number of edges remain in the network; the eigenvector centrality of the network structure is defined as: Eig=Σ v (max(own) v )-own v ) Among them, v is a node in the network, eig v is the eigenvector centrality of node v in the network, max(eig v ) represents the maximum value of the eigenvector centrality of all nodes; The definition of proximity centrality of the network structure is: Clo=Σ v (max(clo v )-clo v ) Among them, clo v is the proximity centrality of node v in the network, max(clo v ) represents the maximum value of the proximity centrality of all nodes; The definition of betweenness centrality of a network structure is: Bet=Σ v (max(bet v )-bet v ) Among them, bet v is the betweenness centrality of node v in the network, max(bet v ) represents the maximum value of the betweenness centrality of all nodes; And, the optimal trimming determination module is used to compare and analyze the proximity centrality Clo and the betweenness centrality Bet after subtracting an edge each time, and find the optimal step S of the two opt ; Prune the initial network to the optimal step S opt The final network is determined as the pruned core network, and the bacteria connected by the edges in the core network constitute the core flora; the optimal step S is determined according to the following rules opt : Use proximity centrality Clo and betweenness centrality Bet as two-dimensional coordinate axes to draw a graph, and take the trimming step corresponding to the point in the upper right corner area as the optimal step.

3. A computer program product comprising a computer program / instructions, characterized in that The computer program / instructions implement the steps of the method according to claim 1 when executed by a processor.