A method for identifying and analyzing microorganisms in soil

Abstract

The present invention provides a method for identifying and analyzing microorganisms in soil and their interactions, which obtains microbial DNA sequences, establishes a sequence quality scoring function, and performs sequence preprocessing; deeply analyzes the biological properties of microbial DNA sequences, extracts sequence features, and clusters the sequences using a fusion method based on a hybrid metaheuristic algorithm; constructs a probabilistic graph model, identifies microorganisms based on DNA sequences, designs a microbial interaction network construction algorithm, and quantitatively analyzes the interactions between microorganisms; the method solves the problem that the prior art lacks the ability to deeply analyze microbial DNA sequences and is unable to comprehensively extract and analyze the multidimensional features of microorganisms, and can more accurately and comprehensively reveal the multidimensional characteristics, interactions, and community structures of microorganisms, providing important theoretical basis and experimental data for microbial ecology and environmental science research.

Description

Technical Field

[0001] The present invention relates to the field of microorganism identification, and in particular to a method for analyzing microorganism identification and interaction in soil. Background Art

[0002] Soil is a complex ecosystem in which microbial communities play a vital role and have a profound impact on soil health, fertility and ecological functions. Microbial interactions, diversity and functionality are the core content of soil ecological research. However, due to the wide variety of soil microorganisms and complex interactions, traditional microbial research methods often fail to accurately and comprehensively reveal the true state and interactions of microorganisms.

[0003] With the development of molecular biology and computational biology, high-throughput sequencing technology has provided new tools for the study of microbial communities, which can directly obtain a large amount of microbial DNA sequence information from environmental samples without cultivation. However, how to accurately identify microorganisms, analyze their interactions, and reveal their functions in soil from these massive data remains a huge challenge.

[0004] my country's patent application number: CN202110420635.7, publication date: 2021.11.16, discloses a method for identifying key microorganisms for corn straw carbon assimilation in soil, aiming to solve the technical problem that the key microbial groups for straw carbon assimilation and their succession laws in straw decomposition cannot be identified or answered at the species level in the prior art. The present invention is based on stable isotope probe technology, based on the isolated 13C-labeled microbial DNA, selects each OTU representative sequence for taxonomic analysis, and statistically analyzes the community composition at each level, and then performs α- / β-diversity and co-occurrence network analysis of the corn straw carbon assimilation bacterial community to identify and identify the key microbial groups for straw carbon assimilation and their community succession laws, and then reveal the microbiological mechanism of straw decomposition in the target soil. Based on this, targeted and efficient corn straw composting agents can be further developed to promote the rapid decomposition of straw, thereby achieving the purpose of improving soil, saving fertilizer and increasing efficiency.

[0005] However, in the process of implementing the technical solutions of the invention in the embodiments of the present application, the inventors of the present application found that the above-mentioned technology has at least the following technical problems: the existing technology lacks the ability to deeply analyze microbial DNA sequences, and is unable to comprehensively extract and analyze the multidimensional characteristics of microorganisms, resulting in an insufficiently in-depth and comprehensive understanding of the biological properties and functional characteristics of microorganisms; there is a lack of effective methods for quantitatively analyzing microbial interactions, and multidimensional network analysis is not fully utilized to reveal the multidimensional structure and functional characteristics of microbial communities, which limits the in-depth understanding of soil microbial ecosystems. Summary of the invention

[0006] The embodiment of the present application provides a method for identifying and analyzing microorganisms in soil and their interactions, which solves the problem that the prior art lacks the ability to deeply analyze microbial DNA sequences, cannot fully extract and analyze the multidimensional characteristics of microorganisms, and leads to insufficient in-depth and comprehensive understanding of the biological properties and functional characteristics of microorganisms; lacks effective methods for quantitatively analyzing microbial interactions, and does not fully utilize multidimensional network analysis to reveal the multidimensional structure and functional characteristics of microbial communities, which limits the in-depth understanding of soil microbial ecosystems. An advanced method for identifying and analyzing microorganisms in soil and their interactions has been implemented, which can more accurately and comprehensively reveal the multidimensional characteristics, interactions and community structure of microorganisms through deep learning and multidimensional network analysis, and provides important theoretical basis and experimental data for microbial ecology and environmental science research.

[0007] The present application provides a method for identifying and analyzing microorganisms in soil, which specifically includes the following technical solutions:

[0008] A method for identifying and analyzing microorganisms in soil and their interactions comprises the following steps:

[0009] S100: Obtain microbial DNA sequences, establish sequence quality scoring functions, and perform sequence preprocessing;

[0010] S200: In-depth analysis of the biological properties of microbial DNA sequences, extraction of sequence features, and clustering of sequences using a fusion method based on a hybrid metaheuristic algorithm;

[0011] S300: Build probabilistic graphical models, identify microorganisms based on DNA sequences, design algorithms for building microbial interaction networks, and quantitatively analyze interactions between microorganisms.

[0012] Preferably, the step S100 specifically includes:

[0013] During the sequence preprocessing stage, a sequence quality scoring function was established, which comprehensively considers the variability, purity and complexity of the sequence. The variability score measures the degree of variation of the bases in the sequence, the purity score measures whether there are impurity sequences in the sequence, and the complexity score measures the complexity of the sequence. After sequence preprocessing, the preprocessed sequence is compared with the reference database to eliminate sequences that do not match any known sequence.

[0014] Preferably, the step S200 specifically includes:

[0015] Based on the in-depth analysis of sequence features and comprehensive consideration of the weights of different features, the similarity between sequences is further calculated; considering that cosine similarity can measure the angle between feature vectors, and weighted Euclidean distance can measure the distance between feature vectors, the two are combined to measure the similarity between sequences; by optimizing the weights of different features, the accuracy of similarity calculation is further improved.

[0016] Preferably, the step S200 specifically includes:

[0017] During the optimization process, a fusion method based on a hybrid metaheuristic algorithm is adopted. The fusion method based on the hybrid metaheuristic algorithm introduces a fitness function based on information entropy to measure the pros and cons of clustering schemes; and also introduces a mutation operation based on neighborhood search to enhance the search ability of the algorithm.

[0018] Preferably, the step S200 specifically includes:

[0019] Perform initial clustering based on similarity and set a similarity threshold. If the similarity between two sequences is less than the similarity threshold, they are classified into the same category. Randomly select a candidate clustering scheme C′ from the neighborhood of the initial clustering scheme C, calculate the fitness values ​​of the candidate clustering scheme and the current initial clustering scheme, that is, their information entropy, and decide whether to accept the candidate clustering scheme C′ based on the simulated annealing criterion.

[0020] Preferably, the step S200 specifically includes:

[0021] If the fitness value of the accepted candidate clustering scheme C′ is higher than that of the initial clustering scheme C, or meets the acceptance probability criterion, the candidate clustering scheme C′ is accepted as the new clustering scheme; the specific content of the acceptance probability criterion is: if the fitness value of the accepted candidate clustering scheme C′ is smaller than that of the initial clustering scheme C, the candidate clustering scheme C′ is accepted with the probability of acceptance probability Paccept(C,C′).

[0022] Preferably, the step S300 specifically includes:

[0023] In the probabilistic graphical model classification, the parent node set is obtained based on an in-depth analysis of the relationships and dependencies between microorganisms, and a probabilistic graphical model is constructed to describe the relationships and dependencies between microorganisms; by learning the structure and parameters of the network, the parent node set of each node can be obtained.

[0024] Preferably, the step S300 specifically includes:

[0025] By using deep learning models to learn a high-level representation of the data that can capture complex patterns and structures in the data, the feature vector X of each microorganism at each time point t is i(t) is converted into a new representation Y i (t), model parameters θ g It is learned by optimizing the algorithm on a large amount of training data to minimize the reconstruction error.

[0026] Preferably, the step S300 specifically includes:

[0027] Based on evolutionary dynamics, a dynamic weight calculation method is proposed. The evolutionary dynamics of microbial abundance data is analyzed using evolutionary game theory and dynamic system theory, so as to calculate the dynamic interaction weights between microorganisms. In order to construct a microbial interaction network, a threshold setting method based on the asymmetric information criterion is introduced. This method calculates the asymmetric information of the interaction network under different thresholds and selects the threshold with the largest asymmetric information as the optimal threshold. The threshold is used to determine the edges in the network, that is, the interactions between microorganisms. By maximizing the asymmetric information, a network that reveals the true interactions between microorganisms can be constructed.

[0028] Based on the constructed microbial interaction network, a network optimization method based on multidimensional analysis was proposed to reveal the multidimensional structure and functional characteristics of the microbial community. The multidimensional characteristics of the network were extracted in different dimensions, and multidimensional network analysis was performed. The network N was projected on each dimension d to obtain the projection of the network on that dimension. The multidimensional data analysis theory was applied to analyze the network projection on each dimension to extract the multidimensional characteristics of the network.

[0029] Beneficial effects:

[0030] The multiple technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0031] 1. By establishing a sequence quality scoring function, the original sequence is quality controlled and filtered to remove low-quality and non-target sequences, thereby ensuring the accuracy and reliability of the sequence; by comparing with the reference database, sequences that do not match any known sequence are eliminated, further ensuring the accuracy of identification and classification.

[0032] 2. By deeply analyzing the biological properties of microbial DNA sequences and extracting a set of multi-dimensional features, we can have a more comprehensive and in-depth understanding of the biological properties and functional characteristics of microorganisms; by clustering the sequences using a fusion method based on a hybrid metaheuristic algorithm, combined with a simulated annealing algorithm and a fitness function based on information entropy, we can cluster and classify microorganisms more efficiently and accurately.

[0033] 3. By constructing a microbial interaction network, we can quantitatively analyze the interactions between microorganisms and reveal the multidimensional structure and functional characteristics of the microbial community, which is of great significance for a deep understanding of the structure and function of the soil microbial ecosystem. A network optimization method based on multidimensional analysis is proposed, which can extract the multidimensional characteristics of the network in different dimensions and perform multidimensional network analysis, thereby revealing the multidimensional structure and functional characteristics of the microbial community in a more comprehensive and in-depth manner.

[0034] 4. The technical solution of this application can effectively solve the problem that the existing technology lacks the ability to deeply analyze microbial DNA sequences, and cannot fully extract and analyze the multidimensional characteristics of microorganisms, resulting in insufficient in-depth and comprehensive understanding of the biological properties and functional characteristics of microorganisms; lacks effective methods for quantitative analysis of microbial interactions, and does not fully utilize multidimensional network analysis to reveal the multidimensional structure and functional characteristics of microbial communities, which limits the in-depth understanding of soil microbial ecosystems. In addition, the above-mentioned system or method has undergone a series of effect surveys and verifications, and finally realized an advanced method for identifying and analyzing microorganisms and interactions in soil. Through deep learning and multidimensional network analysis, it can more accurately and comprehensively reveal the multidimensional characteristics, interactions and community structure of microorganisms, and provide important theoretical basis and experimental data for microbial ecology and environmental science research. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flow chart of a method for identifying and analyzing microorganisms in soil and their interactions described in this application. DETAILED DESCRIPTION

[0036] The embodiments of the present application provide a method for identifying and analyzing microorganisms in soil and their interactions, thereby solving the problems that the prior art lacks the ability to deeply analyze microbial DNA sequences and is unable to comprehensively extract and analyze the multidimensional characteristics of microorganisms, resulting in an insufficiently in-depth and comprehensive understanding of the biological properties and functional characteristics of microorganisms; lacks effective methods for quantitatively analyzing microbial interactions, and does not fully utilize multidimensional network analysis to reveal the multidimensional structure and functional characteristics of microbial communities, limiting an in-depth understanding of soil microbial ecosystems.

[0037] The technical solution in the embodiment of the present application is to solve the above problems, and the overall idea is as follows:

[0038] By establishing a sequence quality scoring function, the original sequence is quality controlled and filtered, and low-quality and non-target sequences are removed, thereby ensuring the accuracy and reliability of the sequence; by comparing with the reference database, sequences that do not match any known sequence are eliminated, further ensuring the accuracy of identification and classification; by deeply analyzing the biological properties of microbial DNA sequences and extracting a set of multidimensional features, a more comprehensive and in-depth understanding of the biological properties and functional characteristics of microorganisms can be achieved; a fusion method based on a hybrid metaheuristic algorithm is used to cluster the sequences, combined with a simulated annealing algorithm and a fitness function based on information entropy, which can more efficiently and accurately cluster and classify microorganisms; by constructing a microbial interaction network, the interactions between microorganisms can be quantitatively analyzed, revealing the multidimensional structure and functional characteristics of the microbial community, which is of great significance for a deep understanding of the structure and function of the soil microbial ecosystem; a network optimization method based on multidimensional analysis is proposed, which can extract the multidimensional characteristics of the network in different dimensions and perform multidimensional network analysis, thereby revealing the multidimensional structure and functional characteristics of the microbial community in a more comprehensive and in-depth manner.

[0039] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0040] Refer to the attached Figure 1 The method for identifying and analyzing microorganisms in soil and their interactions described in the present application comprises the following steps:

[0041] S100: Obtain microbial DNA sequences, establish sequence quality scoring functions, and perform sequence preprocessing;

[0042] Soil samples were collected and microbial DNA was extracted from them. Microbial DNA sequences were obtained using high-throughput sequencing technology. After the raw sequences were obtained, sequence preprocessing was performed to control the quality and filter the raw sequences to remove low-quality and non-target sequences.

[0043] In the sequence preprocessing stage, a sequence quality scoring function was established, which comprehensively considers the variability, purity and complexity of the sequence. The specific formula is:

[0044] Q(s i )=η·V(s i )+θ·P(s i )+ι·H(s i )

[0045] Where Q is the sequence quality score, s i is the original sequence, η, θ, ι are weight parameters, which are used to adjust the weights of variability score V, purity score P and complexity score H in the total score.i ) measures the degree of variation of bases in the sequence and defines the variability score function:

[0046]

[0047] Where N is the number of base types in the sequence. For DNA sequences, N = 4 (A, T, C, G), which are adenine, thymine, cytosine, and guanine respectively; n k is the number of bases of type k in the sequence; L is the length of the sequence; and σ 2 are the mean and variance of the number of bases, respectively. The variability score function statistically analyzes the base distribution and introduces an exponential term to more accurately measure the variability of the sequence. i ) measures whether there is an impurity sequence in the sequence and defines the purity score function:

[0048]

[0049] Where M is the number of possible impurity sequences; p m is the proportion of the mth impurity sequence in the sequence; λ is the weight parameter. The purity score function measures the purity of the sequence more comprehensively through entropy calculation and Gini coefficient in information theory. i ) measures the complexity of the sequence and defines the complexity score function:

[0050]

[0051] Where L is the length of the sequence; f l is the base frequency at the lth position in the sequence; φ is the weight parameter.

[0052] The variability score, purity score and complexity score are all normalized and range between [0,1]. The larger the value, the higher the variability, purity and complexity. According to the experimental purpose and data quality requirements, a quality score threshold is set. Only sequences with a quality score higher than this threshold will be retained, and sequences below this threshold will be eliminated. For example, if the threshold is set to 0.7, all sequences with a quality score lower than 0.7 will be eliminated. The remaining sequences are length filtered to remove sequences that are too long or too short. Through these scores, the quality of the sequence can be more accurately evaluated, and low-quality sequences can be effectively filtered out to ensure the accuracy of subsequent alignment and classification.

[0053] After sequence preprocessing, the preprocessed sequences are compared with the reference database to remove sequences that do not match any known sequences. The DNA sequence-based microbial identification algorithm is used to extract features of the DNA sequences of microorganisms in soil samples to achieve accurate identification and classification of microorganisms, solving the problem of soil microbial diversity and accurate assessment.

[0054] S200: In-depth analysis of the biological properties of microbial DNA sequences, extraction of sequence features, and clustering of sequences using a fusion method based on a hybrid metaheuristic algorithm;

[0055] By deeply analyzing the biological properties of microbial DNA sequences, a set of multi-dimensional features are extracted. For example, the GC content, base frequency, k-mer frequency, etc. of the sequence. The feature vector is set as:

[0056] F(s i )=[GC(s i ),Freq A (s i ),Freq T (s i ),Freq C (s i ),Freq G (s i ),…]

[0057] Among them, F(s i ) represents the sequence s i The characteristic vector of GC(s i )、Freq A (s i )、Freq T (s i )、Freq C (s i ) and Freq G (s i ) represent the sequence s i The GC content and frequency of each base. Each feature is derived from biological knowledge. For example, the GC content is calculated as:

[0058]

[0059] Among them, Count G (s i ) and Count C (s i ) represent the sequence s i The number of G and C bases in Length(s i ) represents the sequence s i Length.

[0060] On the basis of feature extraction, the similarity between sequences is further calculated based on the in-depth analysis of sequence features and the comprehensive consideration of the weights of different features. Considering that cosine similarity can measure the angle between feature vectors, and weighted Euclidean distance can measure the distance between feature vectors, the combination of the two can more comprehensively measure the similarity between sequences. By optimizing the weights of different features, the accuracy of similarity calculation can be further improved. Therefore, the formula for similarity calculation is:

[0061]

[0062] Among them, S(s i ,s k ) represents the sequence s i and sequence s k The similarity between them, ω is the weight parameter of cosine similarity and weighted Euclidean distance, ω m is the weight parameter of the mth feature, φ m (s i ) represents the sequence s i The mth feature of .

[0063] Based on the similarity calculation, the sequences are clustered. In the dynamic clustering optimization, a fusion method based on a hybrid metaheuristic algorithm is adopted. The fusion method based on the hybrid metaheuristic algorithm has the advantages of the simulated annealing algorithm, and also introduces a fitness function based on information entropy to measure the pros and cons of the clustering scheme. A mutation operation based on neighborhood search is introduced to enhance the search ability of the algorithm.

[0064] Specifically, firstly, perform initial clustering according to similarity, set a similarity threshold, and if the similarity between two sequences is less than the similarity threshold, they are classified into the same category. Randomly select a candidate clustering scheme C' from the neighborhood of the initial clustering scheme C, calculate the fitness values ​​of the candidate clustering scheme and the current initial clustering scheme, that is, their information entropy, and decide whether to accept the candidate clustering scheme C' according to the simulated annealing criterion. If the fitness value of the accepted candidate clustering scheme C' is higher than that of the initial clustering scheme C, or meets the acceptance probability criterion, the candidate clustering scheme C' is accepted as a new clustering scheme. The specific content of the acceptance probability criterion is: if the fitness value of the accepted candidate clustering scheme C' is smaller than that of the initial clustering scheme C, the candidate clustering scheme C' is accepted with the probability of acceptance probability Paccept(C,C').

[0065] The information entropy fitness function can consider the sequence distribution in each cluster more carefully, and can introduce the frequency of occurrence of each sequence in each cluster, so as to obtain a more accurate fitness value. The fitness function is expressed as:

[0066]

[0067] Among them, E(C) represents the information entropy of clustering scheme C, which is used to measure the quality of clustering scheme. c is a cluster in C, which contains multiple sequences s i , p(s i |c) represents the sequence s i The probability of occurrence in cluster c can be calculated by dividing the frequency of the sequence in the cluster by the total number of sequences in the cluster. The neighborhood search formula is:

[0068] C′=argmin C′∈N(C,O) E(C′)

[0069] Among them, C' represents the new clustering scheme obtained after neighborhood search, N(C,O) represents the domain of the clustering scheme C obtained through a series of domain operations O, O includes a combination of operations such as exchange, insertion and inversion, and E(C') represents the information entropy of the new clustering scheme C'. The acceptance probability formula of simulated annealing is:

[0070]

[0071] Among them, Paccept(C,C′) represents the acceptance probability of transferring from the current clustering scheme C to the new clustering scheme C′, H represents the external field strength, and M represents the magnetic susceptibility. These two parameters can be used to regulate the acceptance probability, thereby better controlling the search process of the algorithm, and T represents the current temperature.

[0072] S300: Build probabilistic graphical models, identify microorganisms based on DNA sequences, design algorithms for building microbial interaction networks, and quantitatively analyze interactions between microorganisms.

[0073] Microorganisms are identified and classified based on clustering. In the probabilistic graph model classification, the parent node set is obtained based on an in-depth analysis of the relationships and dependencies between microorganisms. By constructing a probabilistic graph model, the relationships and dependencies between microorganisms can be described more accurately. For example, a Bayesian network can be constructed, in which nodes represent microbial categories and edges represent dependencies between microorganisms. By learning the structure and parameters of the network, the parent node set of each node can be obtained. Therefore, the formula for probabilistic graph model classification is:

[0074]

[0075] Among them, P(R|C′) represents the probability of the microbial category set R under a given clustering scheme C′, r i Represents the sequence s i Microbial categories, P(r i |Pa(r i)) means given its parent node Pa(r i ) under the condition that node r i The conditional probability, Pa(r i ) represents the category r i This enables accurate identification and classification of microorganisms, providing a new perspective to understand the diversity and richness of microorganisms.

[0076] A microbial interaction network construction algorithm is designed to quantitatively analyze the interactions between microorganisms. The input microbial feature data is preprocessed by a deep generative model to convert the microbial abundance data into a new representation form to reveal the potential nonlinear structure in the data. Specifically, a high-level representation of the data is learned by using a deep learning model, which is able to capture the complex patterns and structures in the data. The feature vector X of each microorganism at each time point t is i (t) is converted into a new representation Y i (t), model parameters θ g It is learned through optimization algorithm on a large amount of training data to minimize the reconstruction error. The specific formula is:

[0077] Y i (t) = DeepGen(X i (t);θ g )

[0078] Among them, DeepGen represents the deep generation model.

[0079] Based on evolutionary dynamics, a dynamic weight calculation method is proposed. By using evolutionary game theory and dynamic system theory, the evolutionary dynamics of microbial abundance data is analyzed to calculate the dynamic interaction weights between microorganisms. This weight calculation process can be expressed by the following formula:

[0080] W ij (t) = sigmoid(β·ED(Y i (t),Y j (t),S i ,S j )+γ)

[0081] Among them, sigmoid is the activation function, β is the weight adjustment parameter, S i and S j Represent the evolutionary strategy parameters of microorganisms i and j respectively, ED represents the evolutionary dynamics model, which is used to describe the evolutionary process of the microbial population, and γ is the bias term. The above formula is obtained through in-depth research on evolutionary dynamics and multiple experiments, and it can more accurately reflect the interaction intensity between microorganisms.

[0082] In order to construct the microbial interaction network, a threshold setting method based on the asymmetric information criterion is introduced. This method calculates the asymmetric information of the interaction network under different thresholds and selects the threshold with the largest asymmetric information as the optimal threshold. This process can be expressed by the following formula:

[0083] ε * = argmax ε (Af(ε)+λ·ND(ε))

[0084] Among them, ε * represents the optimal threshold for constructing a microbial interaction network, ε is the threshold, Af(ε) is the asymmetric information of the interaction network under the threshold ε, ND(ε) is the density of the network under the threshold ε, and λ is a weight parameter used to balance the asymmetric information of the network and the density of the network. The threshold is used to determine the edges in the network, i.e., the interactions between microorganisms. By maximizing the asymmetric information, a network that reveals the true interactions between microorganisms can be constructed.

[0085] Based on the constructed microbial interaction network, a network optimization method based on multidimensional analysis is proposed to reveal the multidimensional structure and functional characteristics of the microbial community. The multidimensional characteristics of the network are extracted in different dimensions, and multidimensional network analysis is performed. Specifically, the network N is projected on each dimension d to obtain the projection N of the network on that dimension. d . Applying the theory of multidimensional data analysis, the network projection on each dimension is analyzed to extract the multidimensional characteristics of the network. This process can be expressed by the following formula:

[0086]

[0087] Among them, MDA is the result of multidimensional analysis, which is used to reveal the multidimensional structure and functional characteristics of microbial communities, φ is the weight parameter of each dimension, which can be learned through optimization algorithms, and DA is the analysis result of network projection on dimension d. Therefore, the microbial interaction network can be used to deeply study the structure and function of microbial communities, and provide important theoretical basis and experimental data for microbial ecology and environmental science.

[0088] In summary, the method for identifying and analyzing microorganisms in soil and their interactions described in this application has been completed.

[0089] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0090] 1. By setting up a sequence quality scoring function, the original sequence is quality controlled and filtered to remove low-quality and non-target sequences, thereby ensuring the accuracy and reliability of the sequence; by comparing with the reference database, sequences that do not match any known sequence are eliminated, further ensuring the accuracy of identification and classification;

[0091] 2. By deeply analyzing the biological properties of microbial DNA sequences and extracting a set of multi-dimensional features, we can have a more comprehensive and in-depth understanding of the biological properties and functional characteristics of microorganisms. We use a fusion method based on a hybrid meta-heuristic algorithm to cluster sequences, combined with a simulated annealing algorithm and a fitness function based on information entropy, to cluster and classify microorganisms more efficiently and accurately.

[0092] 3. By constructing a microbial interaction network, we can quantitatively analyze the interactions between microorganisms and reveal the multidimensional structure and functional characteristics of the microbial community, which is of great significance for a deep understanding of the structure and function of the soil microbial ecosystem. A network optimization method based on multidimensional analysis is proposed, which can extract the multidimensional characteristics of the network in different dimensions and perform multidimensional network analysis, thereby revealing the multidimensional structure and functional characteristics of the microbial community in a more comprehensive and in-depth manner.

[0093] Effect research:

[0094] The technical solution of this application can effectively solve the problem that the existing technology lacks the ability to deeply analyze microbial DNA sequences, and cannot fully extract and analyze the multidimensional characteristics of microorganisms, resulting in insufficient in-depth and comprehensive understanding of the biological properties and functional characteristics of microorganisms; lacks effective methods for quantitative analysis of microbial interactions, and does not fully utilize multidimensional network analysis to reveal the multidimensional structure and functional characteristics of microbial communities, which limits the in-depth understanding of soil microbial ecosystems. In addition, the above-mentioned system or method has undergone a series of effect surveys and verifications, and finally realized an advanced method for identifying and analyzing microorganisms and interactions in soil. Through deep learning and multidimensional network analysis, it can more accurately and comprehensively reveal the multidimensional characteristics, interactions and community structure of microorganisms, and provide important theoretical basis and experimental data for microbial ecology and environmental science research.

[0095] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0096] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0097] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0098] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for identifying and analyzing microorganisms in soil and their interactions, characterized in that: The following steps are involved: S100: Obtain microbial DNA sequences, establish sequence quality scoring functions, and perform sequence preprocessing; S200: By deeply analyzing the biological properties of microbial DNA sequences, extracting sequence features, and setting the feature vector as: ， in, Representation sequence The characteristic vector of , , , and Respectively represent the sequence of The content and frequency of each base; Based on the analysis of sequence features and comprehensive consideration of the weights of different features, the similarity between sequences is calculated; the formula for similarity calculation is: ， in, Representation sequence and sequence The similarity between is the weight parameter of cosine similarity and weighted Euclidean distance, It is The weight parameter of each feature, Representation sequence No. Features On the basis of similarity calculation, a fusion method based on a hybrid metaheuristic algorithm is used to cluster the sequences; the fusion method based on a hybrid metaheuristic algorithm introduces a fitness function based on information entropy and a mutation operation based on neighborhood search; the formula of the fitness function based on information entropy is as follows: ， in, Represents a clustering scheme The information entropy of It is a clustering scheme A cluster in Representation sequence In clustering The probability of occurrence in ; Specifically, initial clustering is performed based on similarity, and a similarity threshold is set. When the similarity of two sequences is less than the similarity threshold, they are classified into the same category. In the neighborhood of , calculate the information entropy of the candidate clustering scheme and the current initial clustering scheme, and decide whether to accept the candidate clustering scheme according to the simulated annealing criterion ; When accepting the candidate clustering solution The fitness value is better than the initial clustering solution High, or if the acceptance probability criterion is met, the candidate clustering solution is accepted. As a new clustering scheme; the specific content of the acceptance probability criterion is: when accepting the candidate clustering scheme The fitness value is better than the initial clustering solution hours to obtain the current clustering solution Moving to a new clustering scheme The acceptance probability of accepting the candidate clustering solution ; S300: Construct a probabilistic graph model based on a Bayesian network. The nodes represent the microbial categories, and the edges represent the dependencies between microorganisms. By learning the structure and parameters of the network, the parent node set of each node is obtained. The classification formula of the probabilistic graph model is: ， in, Represents a given clustering solution Microorganism category collection The probability of Representation sequence Types of microorganisms, Indicates that given its parent node Under the condition that The conditional probability of Indicates category The parent node set of It is a clustering scheme A cluster in Design a microbial interaction network construction algorithm to quantitatively analyze the interactions between microorganisms; each microorganism at each time point The eigenvector of Transformed into a new representation , the specific formula is: ， in, represents a deep generative model, are model parameters; Based on evolutionary dynamics, a dynamic weight calculation method is proposed. By using evolutionary game theory and dynamic system theory, the evolutionary dynamics of microbial abundance data is analyzed to calculate the dynamic interaction weights between microorganisms. The dynamic interaction weight calculation formula is: ， in, is the dynamic interaction weight, is the activation function, is the weight adjustment parameter, and Respectively represent microorganisms and The evolution strategy parameters, represents the evolutionary dynamics model, is the bias term; The threshold setting method based on asymmetric information criterion was introduced to construct the microbial interaction network, the asymmetric information of the interaction network under different thresholds was calculated, and the threshold with the largest asymmetric information was selected as the optimal threshold; the optimal threshold formula is as follows: ， in, represents the optimal threshold, is the threshold value, is at the threshold Under the asymmetric information of the interaction network, is at the threshold Next, the density of the network, is the weight parameter; Based on the constructed microbial interaction network, a network optimization method based on multidimensional analysis is proposed; the multidimensional characteristics of the network are extracted in different dimensions, and multidimensional network analysis is performed. On the network Perform projection to obtain the projection of the network on this dimension ; Apply multidimensional data analysis theory to analyze the network projection on each dimension and extract the multidimensional characteristics of the network; this process can be expressed by the following formula: ， in, It is the result of multi-dimensional analysis. is the weight parameter of each dimension, It is in the dimension Analysis results of network projection on .

2. A method for identifying and analyzing microorganisms in soil and their interactions as claimed in claim 1, characterized in that: The step S100 specifically includes: During the sequence preprocessing stage, a sequence quality scoring function was established, which comprehensively considers the variability, purity and complexity of the sequence. The variability score measures the degree of variation of the bases in the sequence, the purity score measures whether there are impurity sequences in the sequence, and the complexity score measures the complexity of the sequence. After sequence preprocessing, the preprocessed sequence is compared with the reference database to eliminate sequences that do not match any known sequence.

Images