Cell function region division method and device, electronic equipment and storage medium

Abstract

The invention provides a cell function region division method and device, electronic equipment and a storage medium, and relates to the technical field of data processing, and the main technical scheme comprises the following steps: carrying out dimension reduction preprocessing on to-be-processed cell data to obtain target cell data, the target cell data comprising cell expression profile data and position information; determining a reference cluster number corresponding to the target cell data; inputting target cell data containing the cell expression profile data and the position information into a preset clustering model based on the reference clustering number to obtain initial clustering data; and performing classification correction on the initial clustering data to obtain target clustering data, and performing functional region division based on the target clustering data. Compared with the prior art, the method has the advantages that the to-be-processed cell data is preprocessed, the reference clustering number is determined in advance, and the initial clustering data is classified and corrected to divide the functional region, so that the recognition effect and generalization ability of the same kind of cells are improved.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a method and device for dividing cell functional areas, an electronic device, and a storage medium. Background Art

[0002] Complex plant and animal tissues contain many spatially specific cell types, and the function of tissues is the result of the synergistic effects of cells of different regions and types. Understanding the relative positions of cells with different gene expression profiles and their interaction networks is extremely important for the study of functional mechanisms and the treatment of diseases.

[0003] In order to understand the relative positions of cells with different gene expression profiles, it is necessary to cluster cells with similar gene expression profiles through a cell clustering algorithm, and divide the different functional areas of the tissue based on cell clustering for research, such as the identification and division of tumor tissue, normal tissue, and adjacent tissue areas.

[0004] However, in the related art, the method of clustering cells through cell clustering algorithms and dividing different functional areas of tissues requires the pre-extraction of histological information or significant spatially specific genes, and the amount of calculation is large. Therefore, it leads to poor recognition of similar cells and low generalization ability. Summary of the invention

[0005] The present invention provides a method and device for dividing cell functional regions, an electronic device and a storage medium. The main purpose is to solve the problem in the related art that the method of performing cell clustering and dividing different functional regions of tissues by using a cell clustering algorithm will result in poor recognition effect of cells of the same type and low generalization ability.

[0006] According to a first aspect of the present disclosure, a method for dividing cell functional regions is provided, comprising:

[0007] Performing dimension reduction preprocessing on the cell data to be processed to obtain target cell data, wherein the target cell data includes cell expression profile data and position information;

[0008] Determining the reference cluster number corresponding to the target cell data;

[0009] Based on the reference cluster number, the target cell data including the cell expression profile data and location information is input into a preset clustering model to obtain initial clustering data;

[0010] The initial clustering data is classified and corrected to obtain target clustering data, and functional areas are divided based on the target clustering data.

[0011] Optionally, performing dimension reduction preprocessing on the cell data to be processed to obtain target cell data includes:

[0012] Performing dimensionality reduction processing on the cell data to be processed by a first preset algorithm to obtain vector data of a first preset dimension;

[0013] The coordinate data of the cell data to be processed is merged with the first preset dimensional vector data according to a preset merging rule to obtain target cell data, wherein the target cell data is represented by the second preset dimensional vector data.

[0014] Optionally, based on the reference cluster number, the cell expression profile data and location information are input into a preset clustering model to obtain initial clustering data, including:

[0015] Based on the reference cluster number, respectively determine the distribution parameters corresponding to each cell data in the target cell data in each cluster, wherein the distribution parameters include mean vector data, covariance matrix data, and cluster weights;

[0016] Based on the preset clustering model, according to the mean vector data, covariance matrix data and clustering weights in the distribution parameters, respectively calculating the conditional probability of each cell data in the target cell data in each cluster;

[0017] The maximum value of the conditional probability is obtained, and each cell data in the target cell data is divided into clusters corresponding to the maximum value of the conditional probability to obtain the initial cluster data.

[0018] Optionally, the classifying and correcting the initial clustering data to obtain target clustering data includes:

[0019] Based on the initial clustering data, obtaining the minimum value of the distance between the target correction data and other cell data to obtain the standard length, wherein the target correction data is the cell data that needs to be corrected in the initial clustering data, and the other cell data is all cell data in the initial clustering data except the target correction data;

[0020] The standard radius is calculated by multiplying the standard length by the number of clusters;

[0021] Based on the standard radius, determining a correction range with the target correction data as the center;

[0022] The target correction data is classified into the cluster with the most other cell data in the correction range to obtain the target cluster data.

[0023] According to a second aspect of the present disclosure, a method for training a preset clustering model is provided, which includes:

[0024] Performing dimensionality reduction preprocessing on the training cell data to be processed to obtain target cell data for training, wherein the target cell data for training includes cell expression profile data and position information;

[0025] Determining the number of training reference clusters corresponding to the training target cell data;

[0026] Based on the training reference cluster number, inputting the training target cell data including the cell expression profile data and position information into a preset clustering model to obtain an initial conditional probability for training;

[0027] Based on a preset optimization algorithm, the preset clustering model is optimized according to the initial condition probability for training to obtain a trained preset clustering model.

[0028] Optionally, performing dimension reduction preprocessing on the to-be-processed cell data for training to obtain target cell data for training includes:

[0029] Performing dimensionality reduction processing on the training cell data to be processed by a first preset algorithm to obtain training first preset dimensional vector data;

[0030] The training coordinate data of the training cell data to be processed and the training first preset dimensional vector data are merged by a preset merging rule to obtain training target cell data, wherein the training target cell data is represented by the training second preset dimensional vector data.

[0031] Optionally, the step of inputting the training target cell data into a preset clustering model based on the reference cluster number to obtain the initial conditional probability for training includes:

[0032] Based on the reference number of training clusters, determining initial training distribution parameters corresponding to each cluster, wherein the initial training distribution parameters include initial training mean vector data, initial training covariance matrix data, and initial training cluster weights;

[0033] Deploying the preset clustering model based on the initial distribution parameters for training to obtain an initial preset clustering model;

[0034] Based on the initial preset clustering model, calculation is performed according to the initial training mean vector data, the initial training covariance matrix data and the initial training clustering weights in the initial training distribution parameters to obtain the initial training conditional probability corresponding to each cell data in the training target cell data in each cluster.

[0035] Optionally, optimizing the preset clustering model based on the preset optimization algorithm according to the initial condition probability for training to obtain the trained preset clustering model includes:

[0036] Calculating according to the training initial condition probability and the number of clusters to obtain training update distribution parameters corresponding to each cluster, the training update distribution parameters including training update mean vector data, training update covariance matrix data and training update cluster weights;

[0037] Updating the initial clustering model based on the training updated distribution parameters to obtain an updated preset clustering model;

[0038] Based on the updated preset clustering model, calculation is performed according to the training updated mean vector data, the training updated covariance matrix data and the training updated clustering weights in the training updated distribution parameters to obtain the training updated conditional probability corresponding to each cell data in the training target cell data in each cluster;

[0039] Based on a preset iteration condition and a preset convergence condition, according to the training update conditional probability and the training update distribution parameter, the updated preset clustering model is converged to obtain the training expected distribution parameter;

[0040] The updated preset clustering model is updated based on the training expected distribution parameters to obtain a trained preset clustering model.

[0041] According to a third aspect of the present disclosure, a device for dividing cell functional regions is provided, comprising:

[0042] A processing unit, used for performing dimension reduction preprocessing on the cell data to be processed to obtain target cell data, wherein the target cell data includes cell expression profile data and position information;

[0043] A determination unit, used to determine the reference cluster number corresponding to the target cell data;

[0044] An input unit, configured to input the target cell data including the cell expression profile data and position information into a preset clustering model based on the reference cluster number to obtain initial clustering data;

[0045] A correction unit, used for classifying and correcting the initial clustering data to obtain target clustering data;

[0046] A division unit is used to perform functional area division based on the target clustering data.

[0047] Optionally, the processing unit includes:

[0048] A processing module, used for performing dimensionality reduction processing on the cell data to be processed by a first preset algorithm to obtain vector data of a first preset dimension;

[0049] A merging module is used to merge the coordinate data of the cell data to be processed with the first preset dimensional vector data according to a preset merging rule to obtain target cell data, wherein the target cell data is represented by the second preset dimensional vector data.

[0050] Optionally, the input unit includes:

[0051] A determination module, configured to determine, based on the reference cluster number, distribution parameters corresponding to each cell data in the target cell data in each cluster, wherein the distribution parameters include mean vector data, covariance matrix data, and cluster weights;

[0052] A calculation module, configured to calculate, based on the preset clustering model and according to the mean vector data, the covariance matrix data and the clustering weight in the distribution parameters, the conditional probability corresponding to each cell data in the target cell data in each cluster;

[0053] The partitioning module is used to obtain the maximum value of the conditional probability, and to partition each cell data in the target cell data into the cluster corresponding to the maximum value of the conditional probability, so as to obtain the initial clustering data.

[0054] The correction unit comprises:

[0055] An acquisition module, used for acquiring the minimum value of the distance between the target correction data and other cell data based on the initial clustering data to obtain a standard length, wherein the target correction data is the cell data that needs to be corrected in the initial clustering data, and the other cell data is all cell data in the initial clustering data except the target correction data;

[0056] A calculation module, used for calculating the standard radius by multiplying the standard length by the number of clusters;

[0057] A determination module, configured to determine a correction range based on the standard radius and centered on the target correction data;

[0058] A classification module is used to classify the target correction data into the cluster with the most other cell data in the correction range to obtain the target cluster data.

[0059] According to a fourth aspect of the present disclosure, a training device for a preset clustering model is provided, comprising:

[0060] A processing unit, used to perform dimensionality reduction preprocessing on the to-be-processed cell data for training to obtain target cell data for training, wherein the target cell data for training includes cell expression profile data and position information;

[0061] A determination unit, used to determine the number of training reference clusters corresponding to the training target cell data;

[0062] An input unit, configured to input the target cell data for training including the cell expression profile data and position information into a preset clustering model based on the reference cluster number for training to obtain an initial conditional probability for training;

[0063] The optimization unit is used to optimize the preset clustering model based on the preset optimization algorithm and the initial condition probability for training to obtain a trained preset clustering model.

[0064] Optionally, the processing unit includes:

[0065] A processing module, used for performing dimensionality reduction processing on the training cell data to be processed by a first preset algorithm to obtain training first preset dimensional vector data;

[0066] A merging module is used to merge the training coordinate data of the training cell data to be processed with the training first preset dimensional vector data according to a preset merging rule to obtain training target cell data, wherein the training target cell data is represented by the training second preset dimensional vector data.

[0067] Optionally, the input unit includes:

[0068] A determination module, used to determine the initial training distribution parameters corresponding to each cluster based on the reference number of training clusters, wherein the initial training distribution parameters include initial training mean vector data, initial training covariance matrix data, and initial training cluster weights;

[0069] A deployment module, used to deploy the preset clustering model based on the initial distribution parameters for training to obtain an initial preset clustering model;

[0070] A calculation module is used to calculate based on the initial preset clustering model, according to the initial training mean vector data, the initial training covariance matrix data and the initial training clustering weights in the initial training distribution parameters, to obtain the initial training conditional probability corresponding to each cell data in the training target cell data in each cluster.

[0071] Optionally, the optimization unit includes:

[0072] A first calculation module, used to calculate according to the training initial condition probability and the number of clusters to obtain training update distribution parameters corresponding to each cluster, wherein the training update distribution parameters include training update mean vector data, training update covariance matrix data and training update cluster weights;

[0073] A first updating module, used for updating the initial clustering model based on the training update distribution parameter to obtain an updated preset clustering model;

[0074] A second calculation module is used to calculate, based on the updated preset clustering model, according to the training updated mean vector data, the training updated covariance matrix data and the training updated clustering weights in the training updated distribution parameters, to obtain the training updated conditional probability corresponding to each cell data in the training target cell data in each cluster;

[0075] A convergence module, configured to converge the updated preset clustering model based on a preset iteration condition and a preset convergence condition, according to the training update conditional probability and the training update distribution parameter, to obtain the training expected distribution parameter;

[0076] The second updating module is used to update the updated preset clustering model based on the expected distribution parameters for training to obtain a trained preset clustering model.

[0077] According to a fifth aspect of the present disclosure, there is provided an electronic device, including:

[0078] at least one processor; and

[0079] a memory communicatively connected to the at least one processor; wherein,

[0080] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method described in the first aspect or the method described in the second aspect.

[0081] According to a sixth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method described in the first aspect or the method described in the second aspect.

[0082] According to a seventh aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the method described in the first aspect or the method described in the second aspect.

[0083] The method and device, electronic device and storage medium for dividing the cell functional regions provided by the present disclosure obtain target cell data by performing dimension reduction preprocessing on the cell data to be processed, wherein the target cell data includes cell expression spectrum data and location information; the reference cluster number corresponding to the target cell data is determined; based on the reference cluster number, the target cell data including the cell expression spectrum data and location information is input into a preset clustering model to obtain initial cluster data; the initial cluster data is classified and corrected to obtain target cluster data, and functional regions are divided based on the target cluster data. Compared with the related art, the embodiment of the present disclosure can eliminate the need to extract histological information or significant spatially specific genes in advance by preprocessing the cell data to be processed and determining the reference cluster number in advance, and classifying and correcting the initial cluster data to divide the functional regions, and the amount of calculation is small, thereby improving the recognition effect and generalization ability of similar cells.

[0084] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0086] Figure 1 A schematic diagram of a flow chart of a method for dividing cell functional areas provided in an embodiment of the present disclosure;

[0087] Figure 2 A schematic diagram of an elbow method image of a spectral clustering algorithm of cosine distance provided in an embodiment of the present disclosure;

[0088] Figure 3 An image schematic diagram of an elbow method of a mixed Gaussian clustering algorithm provided in an embodiment of the present disclosure;

[0089] Figure 4 A schematic diagram of initial clustering data provided by an embodiment of the present disclosure;

[0090] Figure 5 A schematic diagram of target clustering data provided by an embodiment of the present disclosure;

[0091] Figure 6 A schematic diagram of a process of data production and preprocessing provided by an embodiment of the present disclosure;

[0092] Figure 7 A schematic diagram of a process of obtaining initial clustering data provided by an embodiment of the present disclosure;

[0093] Figure 8 A schematic diagram of a flow chart of a training method for a preset clustering model provided in an embodiment of the present disclosure;

[0094] Fig. 9 A schematic diagram of a process flow for optimizing a preset clustering model provided in an embodiment of the present disclosure;

[0095] Fig.10 A schematic diagram of clustering results of Arabidopsis leaves provided in an embodiment of the present disclosure;

[0096] Fig.11 A schematic diagram of the structure of a device for dividing cell functional areas provided by an embodiment of the present disclosure;

[0097] Fig.12 A schematic diagram of the structure of another device for dividing cell functional areas provided by an embodiment of the present disclosure;

[0098] Fig.13 A schematic diagram of the structure of a training device for a preset clustering model provided in an embodiment of the present disclosure;

[0099] Fig.14 A schematic diagram of the structure of another training device for a preset clustering model provided in an embodiment of the present disclosure;

[0100] Fig.15 A schematic block diagram of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0101] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0102] The following describes the method and device for dividing cell functional areas, electronic equipment, and storage medium according to embodiments of the present disclosure with reference to the accompanying drawings.

[0103] Figure 1 A schematic flow chart of a method for dividing cell functional areas provided in an embodiment of the present disclosure.

[0104] like Figure 1 As shown, the method comprises the following steps:

[0105] Step 101, performing dimension reduction preprocessing on the cell data to be processed to obtain target cell data, wherein the target cell data includes cell expression profile data and position information.

[0106] In the embodiments of the present disclosure, the cell data to be processed includes a preset number of cell data, and the types of the cell data include but are not limited to: spatial transcriptome data of single-cell sequencing, cellbin, etc. The cellbin refers to a unit for extracting and analyzing the characteristic information of each small piece of the tissue slice during the tissue slice process. In order to facilitate the understanding of the embodiments of the present disclosure, cellbin is used as an example for explanation, and the preset number is a custom selected number, which is specifically not limited by the embodiments of the present disclosure.

[0107] Among them, the form of the cell data to be processed includes but is not limited to: matrix data, vector data, etc. Similarly, the form of the target cell data also includes but is not limited to: matrix data, vector data, etc. Specifically, the embodiments of the present disclosure are not limited.

[0108] The dimensionality reduction preprocessing includes but is not limited to: removing low-expression and low-variance genes from the expression spectrum data of each cell data in the cell data to be processed, leaving gene expression spectrum data of a specified dimension, and determining the position information data of each cell data in the cell data to be processed, so that the target cell data can contain cell expression spectrum data and position information.

[0109] Step 102: Determine the reference cluster number corresponding to the target cell data.

[0110] In the embodiment of the present disclosure, the method for determining the reference number of clusters is a preset judgment method selected by a user, and the preset judgment method includes but is not limited to: a mixed Gaussian clustering algorithm, a spectral clustering algorithm of cosine distance, a Kmeans clustering algorithm, an elbow method of a hierarchical clustering algorithm, etc. Specifically, the embodiment of the present disclosure is not limited.

[0111] It should be noted here that the reference cluster number is a numerical value, which is mainly used to determine how many categories the target cell data can be divided into. For example, if the reference cluster number is 10, it means that the target cell data can be divided into 10 categories; if the reference cluster number is 20, it means that the target cell data can be divided into 20 categories.

[0112] Regarding the determination of the reference cluster number, taking the spectral clustering algorithm of cosine distance and the mixed Gaussian clustering algorithm as examples, the embodiment of the present disclosure provides an elbow method image schematic diagram of the spectral clustering algorithm of cosine distance, such as Figure 2 As shown; and a schematic diagram of the elbow method image of a mixed Gaussian clustering algorithm, as shown Figure 3As shown. The determination of the reference cluster number can be achieved by, but not limited to, the following methods: traverse from cluster number 2 to 40, and determine the point corresponding to the corner of the sharpest variance value inflection point in the spectral clustering algorithm of cos distance and the mixed Gaussian clustering algorithm. Find a cluster number between 2 and 9, a cluster number between 10 and 15, a cluster number between 15 and 20, and a cluster number above 20, depending on the number of clusters that the slice tissue to be studied may have. If the cluster numbers of the two algorithms are adjacent to each other, such as 12 and 14, then the midpoint of 13 is taken comprehensively.

[0113] Step 103: Based on the reference cluster number, the target cell data including the cell expression profile data and location information is input into a preset clustering model to obtain initial clustering data.

[0114] In the disclosed embodiment, the form of the initial clustering data includes but is not limited to: gene expression spectrum data, matrix data, image data, etc., wherein the number of categories of the initial clustering data is the reference cluster number, for example: if the reference cluster number is 10, the number of categories of the initial clustering data is also 10. Figure 4 As shown, Figure 4 A schematic diagram of initial clustering data provided by an embodiment of the present disclosure. The reference cluster number is 3, and it can be clearly observed that Figure 4 It is divided into 3 categories.

[0115] Step 104: classify and correct the initial clustering data to obtain target clustering data, and divide the functional areas based on the target clustering data.

[0116] In the embodiment of the present disclosure, the algorithm used for the classification correction includes but is not limited to: parallelknn function, kd tree algorithm, etc., and the embodiment of the present disclosure does not limit it.

[0117] Among them, classification correction is only performed once, and each point is corrected only based on the category to which other points belong before correction, that is, each cell data in the target cell data is corrected only based on the category of the initial clustering data. Therefore, classification correction is a parallel correction process, and each cell data in the target cell data needs to be corrected based on the category of the initial clustering data at the same time. Figure 5 As shown, Figure 5 A schematic diagram of target clustering data provided in an embodiment of the present disclosure, which is similar to the above steps Figure 4 By comparison, it can be clearly determined Figure 5 The noise is small and the clustering effect is better.

[0118] It should be noted here that the functional area can be determined based on the corrected clustering, i.e., the target clustering data, for example: Figure 5 The three clusters in the model can be divided into three functional areas, and the positions of the functional areas are consistent with the positions of the clusters.

[0119] The method for dividing the cell functional regions provided by the present disclosure obtains target cell data by performing dimension reduction preprocessing on the cell data to be processed, wherein the target cell data includes cell expression spectrum data and location information; determines the reference cluster number corresponding to the target cell data; based on the reference cluster number, inputs the target cell data including the cell expression spectrum data and location information into a preset clustering model to obtain initial cluster data; classifies and corrects the initial cluster data to obtain target cluster data, and divides the functional regions based on the target cluster data. Compared with the related art, the embodiment of the present disclosure, by preprocessing the cell data to be processed and determining the reference cluster number in advance, and classifying and correcting the initial cluster data to divide the functional regions, can eliminate the need to pre-extract histological information or significant spatially specific genes, and has a small amount of calculation, thereby improving the recognition effect and generalization ability of similar cells.

[0120] In one possible implementation of the embodiment of the present disclosure, as a refinement of the above step 101, regarding the dimensionality reduction preprocessing of the cell data to be processed, the embodiment of the present disclosure provides a flow chart of data production and preprocessing, such as Figure 6 As shown, the following steps are included:

[0121] Step 601, performing dimensionality reduction processing on the cell data to be processed by a first preset algorithm to obtain vector data of a first preset dimension.

[0122] In the embodiment of the present disclosure, the first preset dimension is a dimension of custom setting, which is expressed in numerical values. For example, if the first preset dimension is 50, the first preset dimension vector data is 50-dimensional vector data; if the first preset dimension is 30, the first preset dimension vector data is 30-dimensional vector data. The first preset algorithm is a dimensionality reduction algorithm selected by custom selection, such as LDA (Linear Discriminant Analysis), PCA (Principal Component Analysis), etc. Specifically, the embodiment of the present disclosure does not limit the first preset dimension and the first preset algorithm.

[0123] Regarding the dimensionality reduction processing of the cell data to be processed, it can be implemented in the following manner but is not limited to: using Seurat software, performing principal component analysis (PCA) on each cell data in the cell data to be processed, calculating the main characteristic components (PCA components) of the gene expression matrix of each cell data in the cell data to be processed, and selecting the first 50 PCA components to form the first 50-dimensional vectors to obtain 50-dimensional vector data, wherein the first preset dimensional vector data contains the 50-dimensional vector data corresponding to each cell data, that is, all the 50-dimensional vector data together constitute the first preset dimensional vector data. At the same time, since the PCA components of the gene expression matrix of each cell data in the cell data to be processed are calculated, the first preset dimensional vector contains all the information of the gene expression spectrum data in the cell data to be processed.

[0124] Step 602: Merge the coordinate data of the cell data to be processed with the first preset dimensional vector data through a preset merging rule to obtain target cell data, wherein the target cell data is represented by the second preset dimensional vector data.

[0125] In the embodiment of the present disclosure, the coordinate data of the cell data to be processed refers to the Euclidean x, y coordinates of each cell in the cell data to be processed on the slice tissue to be studied, and the preset merging rule is a custom-set rule for merging vector data, such as: splicing by row, etc. Specifically, the embodiment of the present disclosure is not limited.

[0126] The target cell data is obtained by concatenating each cell data in the first preset dimensional vector data with each corresponding cell coordinate data in rows, and is represented by the second preset dimensional vector data. At the same time, since the coordinate data in the cell data to be processed is merged, the second preset dimensional vector contains all the information of the gene expression profile data and position in the cell data to be processed.

[0127] It should be noted here that the second dimensional vector data is based on the first preset dimensional vector data, plus 2-dimensional vector data, the 2-dimensional vector data is: the Euclidean x, y coordinate data of each cell in the cell data to be processed on the slice tissue to be studied, and the merging method is: row splicing, the first preset dimensional vector data and the second dimensional vector data are represented in the form of a row vector matrix. For example: if each cell data in the first preset dimensional vector data is 50-dimensional vector data, then the coordinate data of each cell in the corresponding cell data to be processed, that is, the Euclidean x, y coordinates, are used as the 51st and 52nd dimensional vectors, at this time, each cell data in the target cell data can be represented by 52-dimensional vector data; if each cell data in the first preset dimensional vector data is 30-dimensional vector data, then the coordinate data of each cell in the corresponding cell data to be processed, that is, the Euclidean x, y coordinates, are used as the 31st and 32nd dimensional vectors, at this time, each cell data in the target cell data can be represented by 32-dimensional vector data.

[0128] In one possible implementation of the embodiment of the present disclosure, as a refinement of the above step 103, regarding the process of obtaining initial clustering data based on a preset clustering model, the embodiment of the present disclosure provides a flow chart of obtaining initial clustering data, such as Figure 7 As shown, the following steps are included:

[0129] Step 701: Based on the reference cluster number, the distribution parameters corresponding to each cell data in the target cell data in each cluster are determined respectively, and the distribution parameters include mean vector data, covariance matrix data and cluster weights.

[0130] In the disclosed embodiment, the distribution parameters are: taking the reference cluster number as the number of Gaussian distributions, and determining the initial distribution parameters based on the reference cluster number, and then obtaining the initial conditional probability corresponding to each cell data in the target cell data in each cluster based on the initial distribution parameters, and calculating new distribution parameters based on the initial conditional probabilities, and iterating continuously until the model converges to a local optimal solution, thereby obtaining the desired distribution parameters.

[0131] Among them, the initial distribution parameters are determined randomly or by average distribution. For example, if the reference cluster number is 10, the initial cluster weight is 1 / 10, and the initial mean vector data and the initial covariance matrix data are also obtained by averaging the target cell data according to the 1 / 10 weight.

[0132] The distribution parameter expression is required to mean that each cluster conforms to a different Gaussian distribution, that is, a multivariate normal distribution, that is, the data within each cluster conforms to a certain data distribution.

[0133] Step 702: Based on the preset clustering model, according to the mean vector data, covariance matrix data and clustering weights in the distribution parameters, respectively calculate the conditional probability of each cell data in the target cell data in each cluster.

[0134] In the embodiment of the present disclosure, the conditional probability is the probability that the target cell data conforms to each cluster distribution. The calculation of the conditional probability can be implemented in the following manner but is not limited to: the reference cluster number is k, there are N cell data on the tissue slice to be studied, that is, there are N cell data in the target cell data, and the distribution parameters are μ k , σ k , α k , where μ k is the mean vector data, σ k is the covariance matrix data, α k is the clustering weight, then based on the current parameters, calculate the jth data y in N cell data h The probability of belonging to the kth class As shown in formula (1):

[0135]

[0136] in,

[0137] Step 703 , obtaining the maximum value of the conditional probability, and dividing each cell data in the target cell data into the cluster corresponding to the maximum value of the conditional probability, to obtain the initial cluster data.

[0138] In the disclosed embodiment, the maximum value of the conditional probability indicates that the probability that a certain cell data in the target cell data conforms to the cluster distribution is the largest, so the cluster corresponding to each cell data in the target cell data can be determined according to the maximum value of the conditional probability. j =∈argmax i {1,2,...k}γ i .

[0139] Among them, after obtaining the initial clustering data, the cluster label to which each cell data in the target cell data belongs can be obtained. The cluster label is a label customized by the technician and is mainly used to distinguish each cluster. It has no other practical meaning and does not limit the embodiments of the present disclosure.

[0140] In one implementable manner of the disclosed embodiment, as a refinement of the above-mentioned step 104, when the initial clustering data is classified and corrected, it can be implemented in the following manner but not limited to: based on the initial clustering data, the minimum value of the distance between the target correction data and other cell data is obtained to obtain a standard length, wherein the target correction data is the cell data that needs to be corrected in the initial clustering data, and the other cell data is all cell data in the initial clustering data except the target correction data; the standard length is multiplied by the number of clusters to obtain a standard radius; based on the standard radius, a correction range is determined with the target correction data as the center; the target correction data is classified into the cluster with the most other cell data in the correction range to obtain the target cluster data.

[0141] In the disclosed embodiment, the classification correction is performed on the initial clustering data by means of a correction algorithm, which includes but is not limited to: parallelknn function, kd tree algorithm, etc. The classification correction can very effectively reduce the noise of the initial clustering, conforms to the idea that adjacent points are more likely to belong to the same class, and has extremely fast calculation speed.

[0142] Corresponding to the above-mentioned division method of cell functional areas, Figure 8 A flowchart of a training method for a preset clustering model provided in an embodiment of the present disclosure is shown in FIG. Figure 8 As shown, including:

[0143] Step 801, performing dimension reduction preprocessing on the training cell data to be processed to obtain target cell data for training, wherein the target cell data for training includes cell expression profile data and position information.

[0144] In the embodiments of the present disclosure, the form of the training cell data to be processed includes but is not limited to: matrix data, vector data, etc. Similarly, the form of the training target cell data includes but is not limited to: matrix data, vector data, etc. Specifically, the embodiments of the present disclosure do not limit this.

[0145] The dimensionality reduction preprocessing includes but is not limited to: removing low-expression and low-variance genes from the expression spectrum data of each cell data in the training cell data to be processed, leaving gene expression spectrum data of a specified dimension, and determining the position information data of each cell data in the training cell data to be processed, so that the training target cell data can include cell expression spectrum data and position information data.

[0146] For details, please refer to the description of step 101 above, so it will not be described in detail here.

[0147] Step 802, determining the number of training reference clusters corresponding to the training target cell data.

[0148] In the embodiment of the present disclosure, the method for determining the number of reference clusters for training is a preset judgment method selected by a user, and the preset judgment method includes but is not limited to: a mixed Gaussian clustering algorithm, a spectral clustering algorithm of cosine distance, a Kmeans clustering algorithm, a hierarchical clustering algorithm, and the like. Specifically, the embodiment of the present disclosure is not limited.

[0149] It should be noted here that the training reference cluster number is a numerical value, which is mainly used to determine how many categories the training target cell data can be divided into. For example, if the training reference cluster number is 10, it means that the training target cell data can be divided into 10 categories; if the training reference cluster number is 20, it means that the training target cell data can be divided into 20 categories.

[0150] For details, please refer to the description of step 102 above, so it will not be described in detail here.

[0151] Step 803: Based on the reference cluster number for training, the target cell data for training including the cell expression profile data and position information is input into a preset clustering model to obtain an initial conditional probability for training.

[0152] In the disclosed embodiment, the number of reference clusters for training is used as the number of Gaussian distributions, and the initial distribution parameters are determined based on the number of reference clusters for training. Then, the initial conditional probability corresponding to each cell data in the training target cell data in each cluster is obtained based on the initial distribution parameters. The initial conditional probability for training is an important data in model training.

[0153] The initial distribution parameters are determined randomly or by average distribution. For example, if the reference cluster number is 10, the initial cluster weight is 1 / 10, and the initial mean vector data and the initial covariance matrix data are also obtained by averaging the target cell data according to the 1 / 10 weight. For the calculation method of the initial conditional probability for training, please refer to formula (1) in step 702 above, so it will not be repeated here.

[0154] Step 804: Based on a preset optimization algorithm, the preset clustering model is optimized according to the initial condition probability for training to obtain a trained preset clustering model.

[0155] In the embodiments of the present disclosure, the optimization algorithm is a user-selected optimization algorithm, wherein the optimization algorithm includes iteration conditions and convergence conditions, for example: calculating new distribution parameters according to the initial condition probability used for training, and continuously iterating until the model converges to a local optimal solution, etc. Specifically, the embodiments of the present disclosure do not limit iteration conditions and convergence conditions.

[0156] In one possible implementation of the embodiment of the present disclosure, performing dimension reduction preprocessing on the to-be-processed cell data for training to obtain target cell data for training includes:

[0157] Performing dimensionality reduction processing on the training cell data to be processed by a first preset algorithm to obtain training first preset dimensional vector data;

[0158] The training coordinate data of the training cell data to be processed and the training first preset dimensional vector data are merged by a preset merging rule to obtain training target cell data, wherein the training target cell data is represented by the training second preset dimensional vector data.

[0159] Specifically, regarding the implementation process of this embodiment, please refer to the description of steps 601-602 above, so they will not be described one by one here.

[0160] In an implementable manner of the embodiment of the present disclosure, inputting the training target cell data into a preset clustering model based on the reference cluster number to obtain the initial conditional probability for training includes:

[0161] Based on the reference number of training clusters, determining initial training distribution parameters corresponding to each cluster, wherein the initial training distribution parameters include initial training mean vector data, initial training covariance matrix data, and initial training cluster weights;

[0162] Deploying the preset clustering model based on the initial distribution parameters for training to obtain an initial preset clustering model;

[0163] Based on the initial preset clustering model, calculation is performed according to the initial training mean vector data, the initial training covariance matrix data and the initial training clustering weights in the initial training distribution parameters to obtain the initial training conditional probability corresponding to each cell data in the training target cell data in each cluster.

[0164] Specifically, regarding the implementation process of this embodiment, please refer to the description of the above steps 701-703 and the above step 803, so they will not be described one by one here.

[0165] In one possible implementation of the embodiment of the present disclosure, as a refinement of the above step 804, regarding the optimization of the preset clustering model, the embodiment of the present disclosure provides a flow chart of the optimization of the preset clustering model, such as Fig. 9 As shown, the following steps are included:

[0166] Step 901, calculate according to the initial condition probability for training and the number of clusters to obtain the training update distribution parameters corresponding to each cluster, wherein the training update distribution parameters include training update mean vector data, training update covariance matrix data and training update cluster weights.

[0167] In the embodiment of the present disclosure, the calculation of the updated distribution parameters for training can be implemented in the following manner, but is not limited to: if the number of reference clusters for training is k, there are N cell data on the tissue slice to be studied, that is, there are N cell data in the target cell data for training, and the initial conditional probability for training is γ jk , the jth cell data among N cell data is y j , then based on the current parameters, calculate the updated mean vector for the training of the j-th cell data in the k-th class Update covariance matrix for training And training to update cluster weights As shown in formula (2), formula (3) and formula (4).

[0168]

[0169]

[0170]

[0171] Step 902: Update the initial clustering model based on the training updated distribution parameters to obtain an updated preset clustering model.

[0172] In the embodiment of the present disclosure, when the initial clustering model is updated, it is necessary to calculate the training update probability density according to the training update distribution parameter to update the initial clustering model. The calculation of the training update probability density can be implemented in the following manner but is not limited to: if the number of training reference clusters is k, there are N cell data in the training target cell data, and the initial conditional probability for training is γ jk , the jth cell data among N cell data is y j , the training update mean vector for the kth class is The updated covariance matrix for training is And the training update cluster weights are Based on the current parameters, the updated probability density for training is calculated As shown in formula (5).

[0173]

[0174] Step 903, based on the updated preset clustering model, calculation is performed according to the training updated mean vector data, the training updated covariance matrix data and the training updated clustering weights in the training updated distribution parameters to obtain the training updated conditional probability corresponding to each cell data in the training target cell data in each cluster.

[0175] In the embodiment of the present disclosure, regarding the calculation of the updated conditional probability for training, reference may be made to the description in the above step 702, and therefore, it will not be described in detail here.

[0176] Step 904: based on the preset iteration condition and the preset convergence condition, according to the training update conditional probability and the training update distribution parameter, converge the updated preset clustering model to obtain the training expected distribution parameter.

[0177] In the disclosed embodiment, the preset iteration condition is a custom-set iteration condition, for example: the number of reference clusters for training is taken as the number of Gaussian distributions, and the initial distribution parameters for training are determined according to the number of reference clusters for training, and then the initial distribution parameters for training are obtained according to the initial distribution parameters for training, and the updated distribution parameters for training are calculated according to the initial distribution parameters for training, and the iteration is continued until the model converges to a local optimal solution, thereby obtaining the expected distribution parameters for training.

[0178] The preset convergence condition is a convergence condition set by a user, for example: during training, after 10 consecutive iterations, the training update distribution parameter remains substantially unchanged or changes less than a preset threshold, and the preset threshold is a sufficiently small value set by a user (0.001, 0.002, etc.). Specifically, the embodiments of the present disclosure do not limit the preset convergence condition.

[0179] The expected distribution parameters for training are distribution parameters used to obtain a trained preset clustering model after training under preset iteration conditions and preset convergence conditions.

[0180] Step 905: Update the updated preset clustering model based on the training expected distribution parameters to obtain a trained preset clustering model.

[0181] In the embodiment of the present disclosure, regarding the updating of the updated preset clustering model, reference may be made to the description in the above step 903, so it will not be described in detail here.

[0182] In one possible implementation of the embodiment of the present disclosure, in order to verify the effect and accuracy of the method described in the present disclosure, the present disclosure provides a schematic diagram of clustering results of Arabidopsis leaves, such as Fig.10 As shown, the reference cluster number is 3, and the labels 0, 1, and 2 represent different categories and functional areas respectively. At the same time, through verification with the existing technology, Fig.10 The functional regions in the method can be biologically corresponded to the specific anatomical structure of Arabidopsis leaves. Therefore, it can be determined that the effect of the method described in the present disclosure is good and the accuracy is high.

[0183] In summary, the embodiments of the present disclosure can achieve the following effects:

[0184] 1. The disclosed embodiments improve the generalization ability by clustering cell data, spatial omics data, etc., and have good performance in animal and plant tissues. They are sensitive to clustering irregular blocks and have good clustering effects. When clustering spatial omics data of plant seed germination, the accuracy is high.

[0185] 2. The disclosed embodiment combines the clustering algorithm and the classification algorithm, has strong noise resistance and high accuracy, and the clustering results of cell data of different scales are stable.

[0186] 3. The embodiment of the present disclosure performs clustering through a preset clustering model and automatically performs correction, which occupies little memory and has a short calculation time.

[0187] 4. The disclosed embodiment pre-processes the cell data to be processed and determines the reference cluster number in advance, and classifies and corrects the initial cluster data to divide the functional areas. This eliminates the need to pre-extract histological information or significant spatially specific genes, and has a small amount of calculation, thereby improving the recognition effect and generalization ability of similar cells.

[0188] Corresponding to the above-mentioned method for dividing cell function regions and the method for training a preset clustering model, the present invention also proposes a device for dividing cell function regions and a device for training a preset clustering model. Since the device embodiment of the present invention corresponds to the above-mentioned method embodiment, the details not disclosed in the device embodiment can be referred to the above-mentioned method embodiment, and will not be repeated in the present invention.

[0189] Fig.11 A schematic diagram of a device for dividing cell functional regions provided in an embodiment of the present disclosure is shown in FIG. Fig.11 As shown, including:

[0190] A processing unit 111 is used to perform dimension reduction preprocessing on the cell data to be processed to obtain target cell data, wherein the target cell data includes cell expression profile data and position information;

[0191] A determination unit 112, configured to determine a reference cluster number corresponding to the target cell data;

[0192] An input unit 113, configured to input the target cell data including the cell expression profile data and location information into a preset clustering model based on the reference cluster number to obtain initial clustering data;

[0193] A correction unit 114 is used to perform classification correction on the initial clustering data to obtain target clustering data;

[0194] The division unit 115 is used to perform functional area division based on the target clustering data.

[0195] The cell functional area division device provided by the present disclosure obtains target cell data by performing dimension reduction preprocessing on the cell data to be processed, and the target cell data includes cell expression spectrum data and location information; calls a preset judgment method to determine the reference cluster number corresponding to the target cell data; based on the reference cluster number, the target cell data including the cell expression spectrum data and location information is input into a preset clustering model to obtain initial clustering data; classifies and corrects the initial clustering data to obtain target clustering data, and divides the functional areas based on the target clustering data. Compared with the related art, the embodiment of the present disclosure, by preprocessing the cell data to be processed and determining the reference cluster number in advance, and classifying and correcting the initial clustering data to divide the functional areas, can eliminate the need to pre-extract histological information or significant spatially specific genes, and has a small amount of calculation, thereby improving the recognition effect and generalization ability of similar cells.

[0196] Furthermore, in a possible implementation of the embodiment of the present disclosure, as Fig.12 As shown, the processing unit 111 includes:

[0197] A processing module 1111 is used to perform dimensionality reduction processing on the cell data to be processed by a first preset algorithm to obtain vector data of a first preset dimension;

[0198] The merging module 1112 is used to merge the coordinate data of the cell data to be processed with the first preset dimensional vector data according to a preset merging rule to obtain target cell data, wherein the target cell data is represented by the second preset dimensional vector data.

[0199] Furthermore, in a possible implementation of the embodiment of the present disclosure, as Fig.12As shown, the input unit 113 includes:

[0200] A determination module 1131 is used to determine the distribution parameters corresponding to each cell data in the target cell data in each cluster based on the reference cluster number, wherein the distribution parameters include mean vector data, covariance matrix data and cluster weights;

[0201] A calculation module 1132, configured to calculate the conditional probability corresponding to each cell data in the target cell data in each cluster based on the preset clustering model and the mean vector data, the covariance matrix data and the clustering weight in the distribution parameters;

[0202] The division module 1133 is used to obtain the maximum value of the conditional probability, and divide each cell data in the target cell data into the cluster corresponding to the maximum value of the conditional probability to obtain the initial cluster data.

[0203] Furthermore, in a possible implementation of the embodiment of the present disclosure, as Fig.12 As shown, the correction unit 114 includes:

[0204] An acquisition module 1141 is used to acquire the minimum value of the distance between the target correction data and other cell data based on the initial clustering data to obtain a standard length, wherein the target correction data is the cell data that needs to be corrected in the initial clustering data, and the other cell data is all cell data in the initial clustering data except the target correction data;

[0205] A calculation module 1142, configured to calculate a standard radius by multiplying the standard length by the number of clusters;

[0206] A determination module 1143 is used to determine a correction range based on the standard radius and centered on the target correction data;

[0207] The classification module 1144 is used to classify the target correction data into the cluster with the most other cell data in the correction range to obtain the target cluster data.

[0208] Fig.13 A schematic diagram of a training device for a preset clustering model provided in an embodiment of the present disclosure is shown in FIG. Fig.13 As shown, including:

[0209] The processing unit 131 is used to perform dimension reduction preprocessing on the training cell data to be processed to obtain target cell data for training, wherein the target cell data for training includes cell expression profile data and position information;

[0210] A determination unit 132, configured to determine the number of training reference clusters corresponding to the training target cell data;

[0211] An input unit 133 is used to input the training target cell data including the cell expression profile data and position information into a preset clustering model based on the training reference cluster number to obtain an initial conditional probability for training;

[0212] The optimization unit 134 is used to optimize the preset clustering model based on the preset optimization algorithm and the initial condition probability for training to obtain a trained preset clustering model.

[0213] Furthermore, in a possible implementation of the embodiment of the present disclosure, as Fig.14 As shown, the processing unit 131 includes:

[0214] A processing module 1311 is used to perform dimensionality reduction processing on the training cell data to be processed by a first preset algorithm to obtain training first preset dimensional vector data;

[0215] The merging module 1312 is used to merge the training coordinate data of the training cell data to be processed with the training first preset dimensional vector data according to a preset merging rule to obtain training target cell data, wherein the training target cell data is represented by the training second preset dimensional vector data.

[0216] Furthermore, in a possible implementation of the embodiment of the present disclosure, as Fig.14 As shown, the input unit 133 includes:

[0217] A determination module 1331 is used to determine the initial training distribution parameters corresponding to each cluster based on the reference number of training clusters, wherein the initial training distribution parameters include initial training mean vector data, initial training covariance matrix data, and initial training cluster weights;

[0218] A deployment module 1332, configured to deploy the preset clustering model based on the initial distribution parameters for training to obtain an initial preset clustering model;

[0219] The calculation module 1333 is used to calculate based on the initial preset clustering model, according to the initial training mean vector data, the initial training covariance matrix data and the initial training clustering weights in the initial training distribution parameters, to obtain the initial training conditional probability corresponding to each cell data in the training target cell data in each cluster.

[0220] Furthermore, in a possible implementation of the embodiment of the present disclosure, as Fig.14As shown, the optimization unit 134 includes:

[0221] A first calculation module 1341 is used to calculate according to the training initial condition probability and the number of clusters to obtain the training update distribution parameters corresponding to each cluster, wherein the training update distribution parameters include training update mean vector data, training update covariance matrix data and training update cluster weights;

[0222] A first updating module 1342, configured to update the initial clustering model based on the training update distribution parameter to obtain an updated preset clustering model;

[0223] A second calculation module 1343 is used to calculate, based on the updated preset clustering model, according to the training updated mean vector data, the training updated covariance matrix data and the training updated clustering weights in the training updated distribution parameters, to obtain the training updated conditional probability corresponding to each cell data in the training target cell data in each cluster;

[0224] A convergence module 1344 is used to converge the updated preset clustering model based on a preset iteration condition and a preset convergence condition, according to the training update conditional probability and the training update distribution parameter, to obtain the training expected distribution parameter;

[0225] The second updating module 1345 is used to update the updated preset clustering model based on the training expected distribution parameters to obtain a trained preset clustering model.

[0226] It should be noted that the above explanation of the method embodiment is also applicable to the device of the embodiment of the present disclosure, and the principle is the same, which is no longer limited in the embodiment of the present disclosure.

[0227] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0228] Fig.15 A schematic block diagram of an example electronic device 1500 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0229] like Fig.15 As shown, the device 1500 includes a computing unit 1501, which can perform various appropriate actions and processes according to a computer program stored in a ROM (Read-Only Memory) 1502 or a computer program loaded from a storage unit 1508 to a RAM (Random Access Memory) 1503. In the RAM 1503, various programs and data required for the operation of the device 1500 can also be stored. The computing unit 1501, the ROM 1502, and the RAM 1503 are connected to each other via a bus 1504. An I / O (Input / Output) interface 1505 is also connected to the bus 1504.

[0230] A number of components in the device 1500 are connected to the I / O interface 1505, including: an input unit 1506, such as a keyboard, a mouse, etc.; an output unit 1507, such as various types of displays, speakers, etc.; a storage unit 1508, such as a disk, an optical disk, etc.; and a communication unit 1509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 1509 allows the device 1500 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0231] The computing unit 1501 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 1501 include, but are not limited to, CPU (Central Processing Unit), GPU (Graphic Processing Units), various dedicated AI (Artificial Intelligence) computing chips, various computing units running machine learning model algorithms, DSP (Digital Signal Processor), and any appropriate processor, controller, microcontroller, etc. The computing unit 1501 performs the various methods and processes described above, such as the division method of the cell function area. For example, in some embodiments, the division method of the cell function area may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 1508. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 1500 via ROM 1502 and / or communication unit 1509. When the computer program is loaded into RAM 1503 and executed by the computing unit 1501, one or more steps of the method described above may be performed. Alternatively, in other embodiments, the computing unit 1501 may be configured to execute the aforementioned method for dividing cell functional areas by any other appropriate means (eg, by means of firmware).

[0232] Various embodiments of the systems and techniques described above herein may be implemented in digital electronic circuit systems, integrated circuit systems, FPGAs (Field Programmable Gate Arrays), ASICs (Application-Specific Integrated Circuits), ASSPs (Application Specific Standard Products), SOCs (System On Chips), CPLDs (Complex Programmable Logic Devices), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs that may be executed and / or interpreted on a programmable system including at least one programmable processor that may be a special purpose or general purpose programmable processor that may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0233] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0234] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a RAM, a ROM, an EPROM (Electrically Programmable Read-Only-Memory) or a flash memory, an optical fiber, a CD-ROM (Compact Dis sc Read-Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0235] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (Cathode-Ray Tube) or LCD (Liquid Crystal Display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball), through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0236] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: LAN (Local Area Network), WAN (Wide Area Network), the Internet, and blockchain networks.

[0237] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server may also be a server of a distributed system, or a server combined with a blockchain.

[0238] It should be noted that artificial intelligence is a discipline that studies how computers can simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.), and includes both hardware-level and software-level technologies. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, and big data processing; artificial intelligence software technologies mainly include computer vision technology, speech recognition technology, natural language processing technology, as well as machine learning / deep learning, big data processing technology, knowledge graph technology, and other major directions.

[0239] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0240] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for dividing cell functional areas, characterized in that: include: Performing dimension reduction preprocessing on the cell data to be processed to obtain target cell data, wherein the target cell data includes cell expression profile data and position information; Determining the reference cluster number corresponding to the target cell data; Based on the reference cluster number, the target cell data including the cell expression profile data and location information is input into a preset clustering model to obtain initial clustering data; The initial clustering data is classified and corrected to obtain target clustering data, and functional areas are divided based on the target clustering data.

2. The method according to claim 1, characterized in that The dimensionality reduction preprocessing of the cell data to be processed to obtain the target cell data comprises: Performing dimensionality reduction processing on the cell data to be processed by a first preset algorithm to obtain vector data of a first preset dimension; The coordinate data of the cell data to be processed is merged with the first preset dimensional vector data according to a preset merging rule to obtain target cell data, wherein the target cell data is represented by the second preset dimensional vector data.

3. The method according to claim 2, characterized in that Based on the reference cluster number, the cell expression profile data and location information are input into a preset clustering model to obtain initial clustering data, including: Based on the reference cluster number, respectively determine the distribution parameters corresponding to each cell data in the target cell data in each cluster, wherein the distribution parameters include mean vector data, covariance matrix data, and cluster weights; Based on the preset clustering model, according to the mean vector data, covariance matrix data and clustering weights in the distribution parameters, respectively calculating the conditional probability of each cell data in the target cell data in each cluster; The maximum value of the conditional probability is obtained, and each cell data in the target cell data is divided into clusters corresponding to the maximum value of the conditional probability to obtain the initial cluster data.

4. The method according to claim 3, characterized in that The classifying and correcting the initial clustering data to obtain target clustering data includes: Based on the initial clustering data, obtaining the minimum value of the distance between the target correction data and other cell data to obtain the standard length, wherein the target correction data is the cell data that needs to be corrected in the initial clustering data, and the other cell data is all cell data in the initial clustering data except the target correction data; The standard radius is calculated by multiplying the standard length by the number of clusters; Based on the standard radius, determining a correction range with the target correction data as the center; The target correction data is classified into the cluster with the most other cell data in the correction range to obtain the target cluster data.

5. A training method for a preset clustering model, characterized in that: include: Performing dimensionality reduction preprocessing on the training cell data to be processed to obtain target cell data for training, wherein the target cell data for training includes cell expression profile data and position information; Determining the number of training reference clusters corresponding to the training target cell data; Based on the training reference cluster number, inputting the training target cell data including the cell expression profile data and position information into a preset clustering model to obtain an initial conditional probability for training; Based on a preset optimization algorithm, the preset clustering model is optimized according to the initial condition probability for training to obtain a trained preset clustering model.

6. The method according to claim 5, characterized in that The preprocessing of the cell data to be processed for training with dimension reduction to obtain the target cell data for training includes: Performing dimensionality reduction processing on the training cell data to be processed by a first preset algorithm to obtain training first preset dimensional vector data; The training coordinate data of the training cell data to be processed and the training first preset dimensional vector data are merged by a preset merging rule to obtain training target cell data, wherein the training target cell data is represented by the training second preset dimensional vector data.

7. The method according to claim 6, characterized in that The step of inputting the training target cell data into a preset clustering model based on the reference cluster number to obtain the initial conditional probability for training includes: Based on the reference number of training clusters, determining initial training distribution parameters corresponding to each cluster, wherein the initial training distribution parameters include initial training mean vector data, initial training covariance matrix data, and initial training cluster weights; Deploying the preset clustering model based on the initial distribution parameters for training to obtain an initial preset clustering model; Based on the initial preset clustering model, calculation is performed according to the initial training mean vector data, the initial training covariance matrix data and the initial training clustering weights in the initial training distribution parameters to obtain the initial training conditional probability corresponding to each cell data in the training target cell data in each cluster.

8. The method according to claim 7, characterized in that The step of optimizing the preset clustering model based on the preset optimization algorithm and the initial condition probability for training to obtain the trained preset clustering model includes: Calculating according to the training initial condition probability and the number of clusters to obtain training update distribution parameters corresponding to each cluster, the training update distribution parameters including training update mean vector data, training update covariance matrix data and training update cluster weights; Updating the initial clustering model based on the training updated distribution parameters to obtain an updated preset clustering model; Based on the updated preset clustering model, calculation is performed according to the training updated mean vector data, the training updated covariance matrix data and the training updated clustering weights in the training updated distribution parameters to obtain the training updated conditional probability corresponding to each cell data in the training target cell data in each cluster; Based on a preset iteration condition and a preset convergence condition, according to the training update conditional probability and the training update distribution parameter, the updated preset clustering model is converged to obtain the training expected distribution parameter; The updated preset clustering model is updated based on the training expected distribution parameters to obtain a trained preset clustering model.

9. A device for dividing cell functional areas, characterized in that: include: A processing unit, used for performing dimension reduction preprocessing on the cell data to be processed to obtain target cell data, wherein the target cell data includes cell expression profile data and position information; A determination unit, used to determine the reference cluster number corresponding to the target cell data; An input unit, configured to input the target cell data including the cell expression profile data and position information into a preset clustering model based on the reference cluster number to obtain initial clustering data; A correction unit, used for classifying and correcting the initial clustering data to obtain target clustering data; A division unit is used to perform functional area division based on the target clustering data.

10. A training device for a preset clustering model, characterized in that: include: A processing unit, used to perform dimensionality reduction preprocessing on the to-be-processed cell data for training to obtain target cell data for training, wherein the target cell data for training includes cell expression profile data and position information; A determination unit, used to determine the number of reference clusters for training corresponding to the target cell data; An input unit, configured to input the target cell data for training including the cell expression profile data and position information into a preset clustering model based on the reference cluster number for training to obtain an initial conditional probability for training; The optimization unit is used to optimize the preset clustering model based on the preset optimization algorithm and the initial condition probability for training to obtain a trained preset clustering model.

11. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 4 or the method of any one of claims 5 to 8.

12. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-4 or the method according to any one of claims 5-8.

13. A computer program product, characterized in that The method comprises a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 4 or the method according to any one of claims 5 to 8.

Images