A spatial domain identification method and system based on consensus clustering for spatial multi-omics technology

Through spatial multi-omics technology based on consensus clustering, the problems of large-scale spatial multi-omics data processing are solved, and the efficient identification accuracy and computing efficiency are improved, which promotes spatial multi-omics research. development.

CN118800320BActive Publication Date: 2024-11-15SHANDONG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411287761.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-14
Publication Date
2024-11-15
Estimated Expiration
2044-09-14

AI Technical Summary

Technical Problem

When processing large-scale spatial multi-omics data, the computing equipment demands a large amount of time and takes a long processing time, making it difficult to effectively integrate spatial information and multi-dimensional micro-omics data characteristics, resulting in low recognition accuracy and computational efficiency.

Method used

Using spatial multiomics technology based on consensus clustering, the spatial multiomics data of multiple cells is obtained, the Rand coefficient is calculated and corrected, and the similarity matrix is ​​generated. The spectral clustering algorithm and mean square error are used to optimize the similarity matrix to improve the space. The accuracy and computational efficiency of domain identification.

Benefits of technology

It improves the recognition accuracy of the spatial domain, optimizes the computing efficiency, and promotes the development of spatial omics and biomedical research.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118800320B_ABST
    Figure CN118800320B_ABST
Patent Text Reader

Abstract

The present invention proposes a spatial domain identification method and system of spatial multi-omics technology based on consensus clustering, which belongs to the technical field of bioinformatics. The spatial omics data of #imgabs0# cells are obtained, and #imgabs1# spatial domain identification methods are used to identify the spatial omics data to obtain preliminary clustering results; for the #imgabs2# spatial domain identification methods used, the modified Rand coefficients between the prediction results of any two methods are calculated; based on the calculated modified Rand coefficients, #imgabs3# similarity matrices are obtained, and for each identification result obtained by the spatial domain identification method, it is converted into a #imgabs4# similarity matrix, which can effectively combine consensus clustering and spatial omics characteristics, not only can the identification accuracy of the spatial domain be improved, but also the calculation efficiency can be optimized, which has important scientific significance and application value for promoting spatial omics and related biomedical research.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of bioinformatics, and in particular relates to a spatial domain identification method and system based on consensus clustering of spatial multi-omics technology. Background Art

[0002] Consensus clustering technology, as a powerful data analysis tool, improves the stability and accuracy of clustering decisions by integrating multiple clustering results. However, directly applying traditional consensus clustering technology to spatial multi-omics data still faces the problem of how to effectively integrate spatial information and multi-dimensional omics data characteristics. In addition, when processing large-scale spatial multi-omics data, existing methods require a large number of computing devices for processing, and the processing time is long, which is unacceptable in practical applications. Summary of the invention

[0003] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a spatial domain identification method of spatial multi-omics technology based on consensus clustering, which can effectively combine consensus clustering and spatial omics characteristics, not only can improve the recognition accuracy of spatial domain, but also can optimize the computational efficiency, which has important scientific significance and application value for promoting spatial omics and related biomedical research.

[0004] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0005] In a first aspect, a spatial domain identification method of spatial multi-omics technology based on consensus clustering is disclosed, comprising:

[0006] Acquire spatial omics data of multiple cells using A spatial domain recognition method is used to identify the same sample of spatial omics data and obtain preliminary clustering results.

[0007] For the same sample A spatial domain identification method is used to calculate the modified Rand coefficient between the prediction results of any two methods;

[0008] Generate a The similarity matrix ,in Representation method and methods The modified Lander value of Generate a Expected Similarity Matrix ;

[0009] Based on the calculated modified Rand coefficient Similarity matrix, for each recognition result obtained by the spatial domain recognition method, it is converted into a The similarity matrix ;

[0010] for The distance between two cells is calculated based on the position information, that is, the Euclidean distance. The similarity is calculated based on the Euclidean distance, and then the spatial similarity matrix is ​​obtained. ;

[0011] For the similarity matrix B, the spatial similarity matrix and the expected similarity matrix , use the mean square error to calculate the similarity loss function of each method and spatial position, perform weighted summation on the loss function, and iteratively optimize the similarity matrix , so that the loss function is minimized;

[0012] The similarity matrix obtained using the spectral clustering algorithm Perform data analysis and As a pre-trained similarity matrix, the spatial domain is identified.

[0013] As a further technical solution, the process of calculating the modified Rand coefficient between the prediction results of any two methods is:

[0014] Given two clustering results and , respectively represent the clustering results obtained by two different methods;

[0015] Constructing a confusion matrix ,in Indicates that in the clustering results Belong to the Class and in the clustering results Belong to the The number of sample pairs of the class;

[0016] Calculate the number of combinations for each row, column and total;

[0017] The modified Rand coefficient is calculated based on the number of combinations.

[0018] As a further technical solution, the following also includes: for the similarity matrix ,in Representation method and methods The ARI value of each method is obtained by averaging each column of the similarity matrix, and the methods with ARI values ​​below the threshold are eliminated. The method to obtain the corresponding method with ARI value higher than the threshold α.

[0019] As a further technical solution, for the similarity matrix The rows and columns represent cells, if two cells are of the same type, then , representation method and methods The ARI value of different categories is .

[0020] As a further technical solution, Expected Similarity Matrix , is a diagonal matrix used to initialize the similarity matrix.

[0021] As a further technical solution, For the The similarity matrix of the methods is , Respectively represent Line Column, and then obtain the similarity loss function of each method and the similarity loss function of the spatial position.

[0022] Similarity loss function for each of the above methods:

[0023]

[0024]

[0025] in, For method Generated similarity matrix and expected similarity matrix The mean square error between is the average value of the similarity loss function of all methods, indicating the average mean square error between the similarity matrix generated by multiple recognition methods and the expected similarity matrix. For the The similarity matrix of the methods is , Respectively represent Line List, Indicates the number of cells, Expected Similarity Matrix In the Line The value of the column.

[0026] Similarity loss function of spatial position:

[0027] .

[0028] in, Representation space similarity matrix In the Line The value of the column, is the spatial similarity matrix Similarity matrix with expectation The mean square error between is the similarity loss function of the spatial position.

[0029] As a further technical solution, it also includes calculating the nuclear norm:

[0030] Generate a random matrix Mapping transfer matrix ;

[0031]

[0032] Indicates the number of cells, Indicates the number of recognition results. is a A random matrix, is the transfer matrix;

[0033] After mapping, the matrix conduct Decompose and get the orthogonal matrix and the upper triangular matrix ;

[0034] When performing SVD decomposition and other matrix decompositions, the orthogonal matrix The introduction of can simplify the calculation process;

[0035]

[0036] yes The matrix after mapping, yes of Orthogonal matrix, yes of The upper triangular matrix, then Perform SVD decomposition;

[0037]

[0038] yes of Left singular matrix, The elements on the diagonal are of Singular values, yes of The right singular matrix is The singular value matrix of is used to calculate an approximation of the nuclear norm.

[0039]

[0040] yes The number of singular values ​​of yes No. Singular values.

[0041] In the second aspect, a spatial domain identification system of spatial multi-omics technology based on consensus clustering is disclosed, comprising:

[0042] The preliminary clustering module is configured to obtain spatial omics data of multiple cells using A spatial domain recognition method is used to identify the same sample of spatial omics data and obtain preliminary clustering results.

[0043] The expected similarity matrix generation module is configured as follows: A spatial domain identification method is used to calculate the modified Rand coefficient between the prediction results of any two methods;

[0044] Generate a The similarity matrix ,in Representation method and methods The modified Lander value of Generate a Expected Similarity Matrix ;

[0045] The similarity matrix generation module is configured to obtain based on the calculated modified Rand coefficient Similarity matrix, for each recognition result obtained by the spatial domain recognition method, it is converted into a The similarity matrix ;

[0046] The spatial similarity matrix generation module is configured as follows: The distance between two cells is calculated based on the position information, that is, the Euclidean distance. The Gaussian kernel function is used to calculate the similarity based on the Euclidean distance, and then the spatial similarity matrix is ​​obtained. ;

[0047] The optimization module is configured as follows: for the similarity matrix B, the spatial similarity matrix and the expected similarity matrix , use the mean square error to calculate the similarity loss function of each method and spatial position, perform weighted summation on the loss function, and iteratively optimize the similarity matrix , so that the loss function is minimized;

[0048] The spatial domain recognition module is configured to: use the spectral clustering algorithm to obtain the similarity matrix Perform data analysis and As a pre-trained similarity matrix, the spatial domain is identified.

[0049] One or more of the above technical solutions have the following beneficial effects:

[0050] The technical solution of the present invention discloses a spatial domain identification method of spatial multi-omics technology, which can effectively combine consensus clustering and spatial omics characteristics, not only can improve the recognition accuracy of the spatial domain, but also can optimize the computing efficiency. It has important scientific significance and application value for promoting spatial omics and related biomedical research.

[0051] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0053] Figure 1 This is a schematic diagram of a method flow of a first embodiment of the present invention;

[0054] Figure 2 The recognition result diagram of the first embodiment of the embodiment of the present invention. DETAILED DESCRIPTION

[0055] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0056] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0057] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0058] Embodiment 1

[0059] like Figure 1 As shown, this embodiment discloses a spatial domain identification method of spatial multi-omics technology based on consensus clustering, including:

[0060] Outlier detection is performed on the spatial omics data of cells, which is implemented by existing advanced spatial domain recognition technologies (such as SCANPY, Leiden, STAGATE, stGCL, SEDR, GraphST, MENDER, SpaceFlow, stLearn, SpaGCN, etc.). Each method can obtain a label similarity matrix;

[0061] The label similarity matrix refers to the preliminary recognition results obtained by using the existing method in step S1, step S2, and step S3, that is, the similarity matrix B i ;

[0062] A similarity matrix is ​​constructed based on the label similarity matrix and the space similarity matrix, as shown in step S7. The specific steps include:

[0063] S1: Perform spatial domain recognition on the spatial omics data of cells to obtain preliminary recognition results.

[0064] In this step of the present embodiment, for For the spatial omics data of individual cells, the existing advanced spatial domain recognition technologies (such as SCANPY, Leiden, STAGATE, stGCL, SEDR, GraphST, MENDER, SpaceFlow, stLearn, SpaGCN, etc.) can be used to obtain preliminary clustering results.

[0065] S2: Method screening.

[0066] For the adopted A spatial domain identification method is used to calculate the modified Rand index (ARI) between the prediction results of any two methods. Given two clustering results and , respectively represent the clustering results obtained by two different methods. Construct a confusion matrix ,in Indicates that in the clustering results Belong to the Class and in the clustering results Belong to the Calculate the number of combinations for each row, column, and total:

[0067]

[0068]

[0069]

[0070]

[0071] in Represents the total number of samples. Indicates the number of completely consistent sample pairs in the clustering results, and Indicates that the clustering results are and The number of sample pairs with consistent clustering in is the number of all possible sample pairs. is from The number of combinations of selecting two samples in the sample. Calculate ARI based on the number of combinations:

[0072]

[0073] Then generate a The similarity matrix ,in Representation method and methods The ARI value of each method is obtained by averaging each column of the similarity matrix, and the methods with ARI values ​​below the threshold are eliminated. The method to obtain the corresponding method for ARI values ​​higher than the threshold α is used in S5 to randomly generate a Expected Similarity Matrix .

[0074]

[0075]

[0076] S3: Calculate the similarity matrix for each method.

[0077] The same sample is applied to a recognition method to obtain the similarity matrix between the recognition result and the true result. There are n similarity matrices for n methods in total, and finally we get Similarity Matrix For each recognition result, convert it into a The similarity matrix , making clear scope makes the results easier to interpret and understand.

[0078]

[0079] The rows and columns represent cells, if two cells are of the same type, then , representation method and methods The ARI value of The different categories are , .

[0080] S4: Calculate the spatial similarity matrix.

[0081] Here, it is assumed that cells that are close in spatial position have similar functions, so a spatial similarity matrix is ​​constructed. . Given a cell and The two-dimensional space coordinates are and . The distance between two cells, namely the Euclidean distance, is calculated based on the position information.

[0082]

[0083] The Gaussian Kernel Function is used to calculate the similarity and obtain the spatial similarity matrix. is the bandwidth parameter of the Gaussian kernel function, which controls the decay speed of the similarity.

[0084]

[0085] For all cell pairs, the similarity value is calculated to obtain a The spatial similarity matrix ,in is the total number of cells.

[0086] The above spatial similarity matrix calculates the Euclidean distance between two cells according to the position information, and is used for calculating the similarity loss of the spatial position in the subsequent step S6.

[0087] S5: Initialize the similarity matrix.

[0088] Randomly generate a Expected Similarity Matrix , and is a diagonal matrix used to initialize the similarity matrix. The expected similarity matrix is ​​as follows, which can avoid the initialization matrix being too rigid and provide some randomness.

[0089]

[0090] in, is a The similarity matrix is ​​used to represent the preliminary similarity between cells. is a matrix Middle Line The elements of the column represent the similarity between cells.

[0091] In this step, adding a random perturbation matrix H can prevent the initialization matrix from being too rigid, provide some randomness, and help the algorithm escape from the local optimum; initialization as a diagonal matrix means that each cell is similar only to itself at the beginning, ensuring the simplicity and intuitiveness of the initial state.

[0092] S6: Calculate similarity loss.

[0093] By combining the similarity matrix calculated for each method in step S3 and the spatial similarity matrix calculated in S4, and using the mean square error to calculate the similarity loss function for each method and spatial position, a unified standard is provided to evaluate the quality of similarity matrices generated by different methods.

[0094] For the The similarity matrix of the methods is , Respectively represent Line Columns, build consensus matrix , to characterize the similarity relationship between different cells. The similarity loss function of each method is:

[0095]

[0096]

[0097] Similarity loss function of spatial position:

[0098]

[0099] S7: Calculate the nuclear norm.

[0100] To reduce the impact of noise and reduce model complexity, a random matrix is ​​generated Mapping transfer matrix , through the mapping matrix Moving the matrix to a lower-dimensional space can reduce the impact of noise on the matrix, making subsequent calculations more stable and robust. The introduction of the random matrix O can effectively reduce the dimension of the original matrix, thereby simplifying the calculation process, reducing the consumption of computing resources, and improving the efficiency of processing large-scale data.

[0101]

[0102] Indicates the number of cells, Indicates the number of recognition results. is a random matrix, is the transfer matrix. conduct Decompose and get the orthogonal matrix and the upper triangular matrix The column vectors of the orthogonal matrix Q are unit orthogonal, which means that the column vectors are independent and normalized. This property is very stable in numerical calculations, reducing numerical errors in the calculation process. When performing SVD decomposition and other matrix decompositions, the orthogonal matrix The introduction of can simplify the calculation process.

[0103]

[0104] yes The matrix after mapping, yes of Orthogonal matrix, yes of The upper triangular matrix. Then Although SVD decomposition will lose some information during dimensionality reduction, SVD decomposition can retain the main singular values ​​of the matrix, thereby retaining the core information and structure of the matrix.

[0105]

[0106] yes of Left singular matrix, The elements on the diagonal are of Singular values, yes of The right singular matrix is The singular value matrix of is used to calculate an approximation of the nuclear norm.

[0107]

[0108] yes The number of singular values ​​of yes No. Singular values.

[0109] S8: Model optimization.

[0110] First, the loss function is weighted summed.

[0111]

[0112] in, is the final total loss function, is the similarity matrix based on each method and the expected similarity matrix The loss is calculated by the difference between Based on the spatial position similarity between cells and the expected similarity matrix The loss is calculated by the difference between is based on the expected similarity matrix The loss of the sum of the singular values ​​of is the method similarity loss The weight coefficient of is the spatial position similarity loss The weight coefficient of is the nuclear norm loss The weight coefficient of .

[0113] Select Adam optimizer and iteratively optimize the similarity matrix , so that the loss function of the model is minimized. Taking multiple losses into account through weighted summation enables the model to comprehensively consider the performance of different aspects, thereby obtaining a more comprehensive optimization result.

[0114] The optimized model is obtained through steps S6, S7 and S8.

[0115] S9: Spatial domain identification.

[0116] Use the spectral clustering algorithm to calculate the similarity matrix obtained in step S5 Perform data analysis such as spectral clustering based on the standard Laplace matrix and As a pre-trained similarity matrix, it can more accurately identify the spatial domain, see Appendix Figure 2 As shown, the ARI is 0.88, and there is a high similarity between the clustering results and the true classification.

[0117] Specifically, in the spectral clustering algorithm, the data to be identified is finally input into the K-means algorithm. In the previous step, the feature matrix X is obtained by performing eigenvalue decomposition on the standard Laplace matrix L. This matrix X contains the eigenvectors corresponding to the first k smallest eigenvalues, which are used as the input data of K-means clustering. The K-means algorithm uses these eigenvectors to perform cluster analysis on the data and finally obtain the cluster to which each cell belongs.

[0118] Use the expected similarity matrix H randomly generated in step S5 to construct the standard Laplacian matrix . The standard Laplacian matrix is ​​defined as:

[0119]

[0120] in, is the identity matrix, is the degree matrix, , Indicates The degree of a node is the sum of the connection strength or adjacency relationship between the node and other nodes. Representation Node and nodes The similarity between and Represents the index of the corresponding row and column in the matrix, usually used to represent two nodes in the graph. The standard Laplacian matrix is ​​used to measure the quality of the graph segmentation and capture the potential clustering structure in the data.

[0121] For the standard Laplacian matrix Perform eigenvalue decomposition to obtain eigenvalues ​​and eigenvectors. The eigenvectors corresponding to the smallest eigenvalues ​​constitute the characteristic matrix :

[0122]

[0123] in, By the standard Laplace matrix The eigenvector matrix obtained after eigenvalue decomposition.

[0124] choose The corresponding front Columns, forming a matrix ,in Each column of is a feature vector.

[0125] The feature matrix As data input, K-means algorithm is used for clustering. The goal of K-means clustering is to divide data points into clusters so that the data points in each cluster have the minimum square error. The formula is as follows:

[0126]

[0127] in, is the indicator function, It is By K-means clustering, each cell is assigned to The final cluster label is obtained by selecting one of the clusters. Similarity loss, spatial loss and low-rank loss are constructed.

[0128] Embodiment 2

[0129] The purpose of this embodiment is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the program.

[0130] Embodiment 3

[0131] The purpose of this embodiment is to provide a computer-readable storage medium.

[0132] A computer-readable storage medium stores a computer program, which executes the steps of the above method when executed by a processor.

[0133] Embodiment 4

[0134] The purpose of this embodiment is to provide a spatial domain identification system based on consensus clustering using spatial multi-omics technology, including:

[0135] The preliminary clustering module is configured to obtain spatial omics data of multiple cells using A spatial domain recognition method is used to identify the same sample of spatial omics data and obtain preliminary clustering results.

[0136] The expected similarity matrix generation module is configured as follows: A spatial domain identification method is used to calculate the modified Rand coefficient between the prediction results of any two methods;

[0137] Generate a The similarity matrix ,in Representation method and methods The modified Lander value of Generate a Expected Similarity Matrix ;

[0138] The similarity matrix generation module is configured to obtain based on the calculated modified Rand coefficient Similarity matrix, for each recognition result obtained by the spatial domain recognition method, it is converted into a The similarity matrix ;

[0139] The spatial similarity matrix generation module is configured as follows: The distance between two cells is calculated based on the position information, that is, the Euclidean distance. The Gaussian kernel function is used to calculate the similarity based on the Euclidean distance, and then the spatial similarity matrix is ​​obtained. ;

[0140] The optimization module is configured as follows: for the similarity matrix B, the spatial similarity matrix and the expected similarity matrix , use the mean square error to calculate the similarity loss function of each method and spatial position, perform weighted summation on the loss function, and iteratively optimize the similarity matrix , so that the loss function is minimized;

[0141] The spatial domain recognition module is configured to: use the spectral clustering algorithm to obtain the similarity matrix Perform data analysis and As a pre-trained similarity matrix, the spatial domain is identified.

[0142] Embodiment 5

[0143] The purpose of this embodiment is to provide a computer program product containing instructions, which, when running on a computer, enables the computer to execute the methods and functions involved in any of the above embodiments.

[0144] The steps involved in the apparatus of the above embodiment correspond to the method embodiment 1, and the specific implementation method can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0145] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0146] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A spatial domain identification method based on consensus clustering of spatial multi-omics technology, characterized by: include: Acquire spatial omics data of multiple cells using A spatial domain recognition method is used to identify the same sample of spatial omics data and obtain preliminary clustering results. For the same sample A spatial domain identification method is used to calculate the modified Rand coefficient between the prediction results of any two methods; Generate a The similarity matrix ,in Representation method and methods The modified Lander value of Generate a Expected Similarity Matrix ; Based on the calculated modified Rand coefficient Similarity matrix, for each recognition result obtained by the spatial domain recognition method, it is converted into a The similarity matrix ; for The distance between two cells is calculated based on the position information, that is, the Euclidean distance. The Gaussian kernel function is used to calculate the similarity based on the Euclidean distance, and then the spatial similarity matrix is ​​obtained. ; For the similarity matrix B, the spatial similarity matrix and the expected similarity matrix , use the mean square error to calculate the similarity loss function of each method and spatial position, perform weighted summation on the loss function, and iteratively optimize the similarity matrix , so that the loss function is minimized; The similarity matrix obtained using the spectral clustering algorithm Perform data analysis and As a pre-trained similarity matrix, the spatial domain is identified.

2. The spatial domain identification method of spatial multi-omics technology based on consensus clustering as claimed in claim 1, characterized in that: The procedure for calculating the modified Rand coefficient between the predictions of any two methods is: Given two clustering results and , respectively represent the clustering results obtained by two different methods; Constructing a confusion matrix ,in Indicates that in the clustering results Belong to the Class and in the clustering results Belong to the The number of sample pairs of the class; Calculate the number of combinations for each row, column and total; The modified Rand coefficient is calculated based on the number of combinations.

3. The spatial domain identification method of spatial multi-omics technology based on consensus clustering as claimed in claim 1, characterized in that: Also included: For the similarity matrix ,in Representation method and methods The ARI value of each method is obtained by averaging each column of the similarity matrix, and the methods with ARI values ​​below the threshold are eliminated. The method to obtain the corresponding method with ARI value higher than the threshold α.

4. The spatial domain identification method of spatial multi-omics technology based on consensus clustering as claimed in claim 1, characterized in that: For the similarity matrix The rows and columns represent cells, if two cells are of the same type, then , representation method and methods The ARI value of different categories is .

5. The spatial domain identification method of spatial multi-omics technology based on consensus clustering as claimed in claim 1, characterized in that: Expected Similarity Matrix , is a diagonal matrix used to initialize the similarity matrix.

6. The spatial domain identification method of spatial multi-omics technology based on consensus clustering as claimed in claim 1, characterized in that: For the The similarity matrix of the methods is , Respectively represent Line Column, and then obtain the similarity loss function of each method and the similarity loss function of the spatial position.

7. A spatial domain identification system based on consensus clustering of spatial multi-omics technology, characterized by: include: The preliminary clustering module is configured to obtain spatial omics data of multiple cells using A spatial domain recognition method is used to identify the same sample of spatial omics data and obtain preliminary clustering results. The expected similarity matrix generation module is configured as follows: A spatial domain identification method is used to calculate the modified Rand coefficient between the prediction results of any two methods; Generate a The similarity matrix ,in Representation method and methods The modified Lander value of Generate a Expected Similarity Matrix ; The similarity matrix generation module is configured to obtain based on the calculated modified Rand coefficient Similarity matrix, for each recognition result obtained by the spatial domain recognition method, it is converted into a The similarity matrix ; The spatial similarity matrix generation module is configured as follows: The distance between two cells is calculated based on the position information, that is, the Euclidean distance. The Gaussian kernel function is used to calculate the similarity based on the Euclidean distance, and then the spatial similarity matrix is ​​obtained. ; The optimization module is configured as follows: for the similarity matrix B, the spatial similarity matrix and the expected similarity matrix , use the mean square error to calculate the similarity loss function of each method and spatial position, perform weighted summation on the loss function, and iteratively optimize the similarity matrix , so that the loss function is minimized; The spatial domain recognition module is configured to: use the spectral clustering algorithm to obtain the similarity matrix Perform data analysis and As a pre-trained similarity matrix, the spatial domain is identified.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 6 are performed.