Spatiotemporal transcriptomics slice alignment method and apparatus

By assigning the same or different weights to the spots in each slice in the spatial transcriptome experiment, and calculating the probability transmission matrix based on the weights and regularization coefficients, the problem of insufficient registration accuracy in the prior art is solved, and spatial transcription is improved Registration accuracy of group slices.

WO2024239284A1PCT designated stage expired Publication Date: 2024-11-28SHENZHEN HUADA GENE INST +1
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
PCT/CN2023/096109
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-24
Publication Date
2024-11-28

AI Technical Summary

Technical Problem

In the existing spatial transcriptome experimental probability registration algorithm (PASTE), each spot point has the same weight, and the proportion of structural information and feature information in the objective function is pre-set and cannot be automatically adjusted, resulting in matching. Insufficient accuracy.

Method used

By obtaining the expression quantity matrix of two adjacent slices, the weight of the same type of spot in each slice is determined separately, and the probability transmission matrix is ​​calculated based on the weight and preset regularization coefficients to obtain the registration score of each spot. This method introduces an automated parameter adjustment mechanism, uses different regularization coefficients to calculate the probability transmission matrix, and selects the matrix with the highest accuracy score as the final result.

Benefits of technology

The accuracy of the probability transfer matrix obtained based on weight calculation is improved, thereby improving the accuracy of registration.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2023096109_28112024_PF_FP_ABST
    Figure CN2023096109_28112024_PF_FP_ABST
Patent Text Reader

Abstract

The present disclosure provides a spatiotemporal transcriptomics slice alignment method and apparatus. The main technical solution comprises: obtaining two adjacent slices, wherein each slice comprises at least one type of expression level matrixes, and each expression level matrix comprises spots of a plurality of pieces of spatiotemporal transcriptomics expression level data; respectively determining weights corresponding to the same type of spots in each slice, wherein the same weight is assigned to the same type of spots, and weights assigned to different types of spots are different (102); and according to the weight corresponding to each type of spots and a preset regularization coefficient, respectively calculating probability transition matrixes to obtain an alignment score of each spot in the expression level matrix corresponding to each category on the two slices (103). Compared with the related art, the same weight is assigned to the same type of spots in each slice, and the weights assigned to different types of spots are different, so that the accuracy of the probability transition matrixes obtained on the basis of weight calculation is improved, and the accuracy of alignment realized on the basis of the probability transition matrixes is thus improved.
Need to check novelty before this filing date? Find Prior Art

Description

Registration method and device for spatiotemporal transcriptome slices Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a method and device for registering spatiotemporal transcriptome slices. Background Art

[0002] Probabilistic Alignment of Spatial Transcriptomics Experiments (PASTE) is an algorithm that aligns the sites (also called spots) of spatiotemporal expression data from adjacent or similar slices. Spots can represent cells or expression detection areas of similar size to cells, which are artificially divided on the spatiotemporal chip sites with a width of n sites multiplied by the height of n sites.

[0003] Each spot point in the PASTE algorithm has the same weight, and in every two pieces of spatiotemporal expression data, the proportion of structural information and feature information in the objective function is pre-set and cannot be automatically adjusted.

[0004] Summary of the Invention

[0005] The present disclosure provides a method and device for registering spatiotemporal transcriptome slices, an electronic device, and a storage medium.

[0006] According to a first aspect of the present disclosure, a method for registering spatiotemporal transcriptome slices is provided, comprising:

[0007] Obtain two adjacent slices, each slice contains at least one type of expression matrix, each expression matrix contains multiple spots of spatiotemporal expression data, and each slice contains at least one type of spot;

[0008] Determine the weights corresponding to the same type of spots in each slice, where spots of the same type are assigned the same weight;

[0009] According to the weight corresponding to each type of spot and the preset regularization coefficient, the probability transfer matrix is ​​calculated respectively to obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

[0010] Optionally, before acquiring two adjacent slices, the method further includes:

[0011] Remove the spots that are not shared by two adjacent slices.

[0012] Optionally, determining the weights corresponding to spots of the same type in each slice includes:

[0013] For the same type of spot in each slice, the corresponding expression matrix is ​​summed along the spot points to obtain the expression matrix corresponding to different types;

[0014] Calculate the distance between the expression matrix of each type of spot on the two slices, and determine the weight corresponding to each type of spot according to the distance.

[0015] Optionally, before summing the corresponding expression matrices along the spot points for the same type of spots in each slice to obtain expression matrices corresponding to different types, the method further includes:

[0016] The non-hypermutated genes in each slice were filtered and the hypermutated genes in each slice were retained.

[0017] Optionally, before determining the weight corresponding to each type of spot according to the distance, the method further includes:

[0018] Find the distance value greater than the preset distance threshold;

[0019] Calling a preset processing function to compress the distance value greater than the preset distance threshold;

[0020] The compressed distance value and the distance value less than the preset distance threshold are used as the final distance of the expression matrix of the spot on the two slices;

[0021] Determining the weight corresponding to each type of spot according to the distance includes:

[0022] Calculate the inverse of the final distance, and determine the weight corresponding to each type of spot based on the calculation result of the inverse number.

[0023] Optionally, after calculating the probability transfer matrix according to the weight corresponding to each type of spot and the preset regularization coefficient, and obtaining the registration score of each spot in the expression matrix corresponding to each category on the two slices, the method further includes:

[0024] Searching for a probability smaller than a preset threshold in the probability transmission matrix, and setting the probability smaller than the preset threshold to 0;

[0025] Each probability in the probability transfer matrix is ​​divided by the sum of the probability transfer matrices to obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

[0026] Optionally, the calculating the probability transmission matrix according to the weight corresponding to each type of spot and the preset regularization coefficient includes:

[0027] Based on different preset regularization coefficients, the corresponding probability transmission matrix is ​​calculated respectively;

[0028] A probability matrix with an accuracy score higher than a preset score threshold is selected from different probability transfer matrices as the final probability transfer matrix.

[0029] In a second aspect, a spatiotemporal transcriptome slice registration device is also provided, comprising:

[0030] An acquisition unit is used to acquire two adjacent slices, each slice contains at least one type of expression matrix, each expression matrix contains multiple spots of spatiotemporal expression data, and each slice contains at least one type of spot;

[0031] A determination unit is used to determine the weights corresponding to the spots of the same type in each slice, wherein the spots of the same type are assigned the same weight;

[0032] The registration unit is used to calculate the probability transfer matrix according to the weight corresponding to each type of spot and the preset regularization coefficient, and obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

[0033] Optionally, the device further comprises:

[0034] The processing unit is used to remove the spots that are not common to the two adjacent slices before obtaining the two adjacent slices.

[0035] Optionally, the determining unit includes:

[0036] The summation module is used to sum the corresponding expression matrix along the spot points of the same type in each slice to obtain the expression matrix corresponding to different types;

[0037] A calculation module, used to calculate the distance between the expression matrix of each type of spot on the two slices;

[0038] The first determination module is used to determine the weight corresponding to each type of spot according to the distance.

[0039] Optionally, the determining unit further includes:

[0040] The filtering module is used to filter the non-highly variable genes in each slice and retain the high-variable genes in each slice before summing the corresponding expression matrices along the spot points for the same type of spots in each slice to obtain the expression matrices corresponding to different types.

[0041] Optionally, the determining unit further includes:

[0042] A search module, configured to search for a distance value greater than a preset distance threshold before determining the weight corresponding to each type of spot according to the distance;

[0043] A processing module, configured to call a preset processing function to compress the distance value greater than the preset distance threshold;

[0044] A second determining module is configured to use the compressed distance value and the distance value less than the preset distance threshold as the final distance of the expression matrix of the spot on the two slices;

[0045] The first determination module is further configured to calculate the inverse of the final distance, and determine the weight corresponding to each type of spot according to the calculation result of the inverse number.

[0046] Optionally, the device further comprises:

[0047] A search unit is configured to calculate the probability transfer matrix according to the weight corresponding to each type of spot and the preset regularization coefficient in the registration unit, obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices, and then search for the probability in the probability transfer matrix that is less than a preset threshold;

[0048] A setting unit, configured to set the probability of being less than a preset threshold to 0;

[0049] Each probability in the probability transfer matrix is ​​divided by the sum of the probability transfer matrices to obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

[0050] Optionally, the registration unit includes:

[0051] Based on different preset regularization coefficients, the corresponding probability transmission matrix is ​​calculated respectively;

[0052] A probability matrix with an accuracy score higher than a preset score threshold is selected from different probability transfer matrices as the final probability transfer matrix.

[0053] According to a third aspect of the present disclosure, there is provided an electronic device, including:

[0054] at least one processor; and

[0055] a memory communicatively connected to the at least one processor; wherein,

[0056] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect.

[0057] According to a fourth 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.

[0058] According to a fifth 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 as described in the first aspect above.

[0059] The present disclosure provides a registration method for spatiotemporal transcriptome slices, obtaining two adjacent slices, each slice containing at least one type of expression matrix, each expression matrix containing multiple site spots of spatiotemporal group expression data, and each slice containing at least one type of spot; respectively determining the weights corresponding to the same type of spots in each slice, wherein the same weights are assigned to spots of the same type; respectively calculating the probability transfer matrix according to the weights corresponding to each type of spot and the preset regularization coefficient, and obtaining the registration score of each spot in the expression matrix corresponding to each category on the two slices. Compared with the related art, the embodiment of the present disclosure assigns the same weight to spots of the same type in each slice, and different types of spots are assigned different weights, thereby improving the accuracy of the probability transfer matrix obtained based on the weight calculation, and thereby improving the accuracy of the registration achieved based on the probability transfer matrix.

[0060] 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

[0061] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0062] FIG1 is a schematic diagram of a process for registering spatiotemporal transcriptome slices according to an embodiment of the present disclosure;

[0063] FIG2 is a schematic diagram of a process of another spatiotemporal transcriptome slice registration method provided by an embodiment of the present disclosure;

[0064] FIG3 is a flow chart of a method for determining weights corresponding to spots of the same type in each slice provided by an embodiment of the present disclosure;

[0065] FIG4 is a schematic structural diagram of a spatiotemporal transcriptome slice registration apparatus provided by an embodiment of the present disclosure;

[0066] FIG5 is a schematic structural diagram of another spatiotemporal transcriptome slice registration apparatus provided by an embodiment of the present disclosure;

[0067] FIG6 is a schematic block diagram of an example electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0068] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize 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.

[0069] The following describes the spatiotemporal transcriptome slice registration method and apparatus, electronic device, and storage medium of the embodiments of the present disclosure with reference to the accompanying drawings.

[0070] FIG1 is a schematic flow chart of a spatiotemporal transcriptome slice registration method provided in an embodiment of the present disclosure.

[0071] As shown in Figure 1, the method includes the following steps:

[0072] Step 101: obtain two adjacent slices, each slice containing at least one type of expression matrix, each expression matrix containing multiple spots of spatiotemporal expression data, and each slice containing at least one type of spot.

[0073] It should be noted that the method described in the embodiments of the present application is not applicable only to the registration between two adjacent slices. Rather, during a single registration calculation, two slices need to be registered pairwise. For example, if there are five slices that need to be registered, and the current calculation is for the registration between slice 1 and adjacent slice 2, after completing the registration of slices 1 and 2, the registration between slice 2 and adjacent slice 3 is continued, followed by the registration between slice 3 and adjacent slice 4, and finally the registration between slice 4 and adjacent slice 5. The above examples are merely for the convenience of quantitative description, and in actual applications, there is no limit on the number of slices to be registered.

[0074] It should be noted that the expression data described in the examples of this application are clustered or annotated to assist in algorithm implementation; in fact, for the needs of later analysis and presentation, the vast majority of spot-registered data are clustered or annotated. The spots can represent either cells or expression detection regions of approximately cell size, artificially demarcated on the spatiotemporal chip with a width of n sites and a height of n sites.

[0075] Step 102: Determine the weights corresponding to spots of the same type in each slice, where spots of the same type are assigned the same weight.

[0076] In an embodiment of the present application, spots of the same type have the same weight value, which depends on the similarity between the expression of the type on one slice and the expression on another slice, that is, the greater the similarity, the greater the weight value, and the smaller the similarity, the smaller the weight value.

[0077] When determining whether the types are the same, the clustering result or annotation result of the expression data in step 101 can be used to determine. The description of the clustering algorithm is not repeated in this embodiment of the present application. Step 103, based on the weight corresponding to each type of spot and the preset regularization coefficient, the probability transfer matrix is ​​calculated respectively to obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

[0078] To improve the accuracy of the registration algorithm, the disclosed embodiments introduce an automated parameter adjustment mechanism: During the calculation, the algorithm sequentially uses different regularization coefficients with exponential variations to calculate the probability transfer matrix and obtain the corresponding accuracy score. After rotating through all the regularization coefficients specified for the calculation, the probability transfer matrix with the highest accuracy score is selected as the registration score for each spot in the expression matrix corresponding to each category on the two slices.

[0079] The present disclosure provides a registration method for spatiotemporal transcriptome slices, obtaining two adjacent slices, each slice containing at least one type of expression matrix, each expression matrix containing multiple site spots of spatiotemporal group expression data, and each slice containing at least one type of spot; respectively determining the weights corresponding to the same type of spot in each slice, wherein the same type of spot is assigned the same weight, and different types of spots are assigned different weights; respectively calculating the probability transfer matrix according to the weight corresponding to each type of spot and the preset regularization coefficient, and obtaining the registration score of each spot in the expression matrix corresponding to each category on the two slices. Compared with the related art, the embodiment of the present disclosure assigns the same weight to the same type of spot in each slice, and different types of spots are assigned different weights, thereby improving the accuracy of the probability transfer matrix obtained based on the weight calculation, and thereby improving the accuracy of the registration achieved based on the probability transfer matrix.

[0080] The present application embodiment also provides a flowchart of another spatiotemporal transcriptome slice registration method.

[0081] As shown in FIG2 , the method further includes:

[0082] Step 201: remove the unique spots in two adjacent slices.

[0083] In the embodiment of the present application, an operation of removing the unique spot types is added. The spots that appear only on one of the two pieces of expression data are removed to prevent them from becoming noise and interfering with the calculation of the probability transfer matrix.

[0084] Step 202: Obtain two adjacent slices, each slice containing at least one type of expression matrix, each expression matrix containing multiple spots of spatiotemporal expression data, and each slice containing at least one type of spot;

[0085] For step 202 , please refer to the detailed description of step 101 .

[0086] Step 203: Determine the weights corresponding to the same type of spots in each slice, where the same type of spots are assigned the same weight, and different types of spots are assigned different weights.

[0087] When determining the weights corresponding to the same type of spots in each slice, the following methods can be used, but are not limited to them, as shown in Figure 3, including:

[0088] Step 2031 , filtering the non-hypermutated genes in each slice and retaining the hypermutated genes in each slice.

[0089] For the spot gene expression matrix, we select and retain highly variable genes and remove all non-variable genes. This aims to reduce the impact of non-variable genes on the calculation of the similarity of the spot class across different slices, resulting in more accurate calculation results.

[0090] To facilitate understanding of highly variable genes, for example, a slice captures a total of 10,000 different genes. The gene expression matrix records how many genes are captured in each spot. Therefore, the number of rows in the gene expression matrix is ​​equal to the number of spots, and the number of columns is equal to 10,000. To screen for highly variable genes, the following three steps are performed in order: (1) standardization, (2) calculation of the variance of each gene, and (3) sorting and screening. First, we introduce standardization using z-score standardization as an example: for each row in the gene expression matrix, calculate the mean and standard deviation of all values ​​on the row, subtract the mean from each value, and then divide by the standard deviation; after completing this calculation for each row, step (1) is completed. Note that the value of the matrix is ​​updated; then, for these 10,000 genes, calculate the variance of all data on the column of each gene as the variance of the gene, and step (2) is completed; finally, for these 10,000 genes, sort them from large to small according to the variance calculated in (2), and take the top ranked genes to obtain the highly variable genes. Among these 10,000 genes, those that are not highly variable genes are non-hypervariable genes. Specifically, several numbers in this process include but are not limited to 10, 20, 50, 100, 200, 500, etc., and the specific number is not limited in this embodiment of the application.

[0091] Step 2032: For spots of the same type in each slice, the corresponding expression matrix is ​​summed along the spot points to obtain expression matrices corresponding to different types;

[0092] For each type of spot on each slice, its expression matrix is ​​summed along the spot points to obtain the expression matrix of different genes in the entire spot class.

[0093] For each type of spot on each slice, the obtained expression matrix is ​​log-normalized to control the scale effect of the data, that is, to reduce the problem that the expression matrices of each spot class are not on the same scale due to the uneven sequencing depth of different capture sites on the chip and the uneven number of spots in different spot groups.

[0094] For example, suppose there are 4 spot points, and there are two categories of spot points: category 1 and category 2 (two points in one category), and 3 genes a, b, and c: count the two spots in category 1, and see how many a genes are expressed in total; count the two spots in category 1, and see how many b genes are expressed in total; count the two spots in category 1, and see how many c genes are expressed in total; count the two spots in category 2, and see how many a genes are expressed in total; count the two spots in category 2, and see how many b genes are expressed in total; count the two spots in category 2, and see how many c genes are expressed in total.

[0095] When performing normalization processing, logarithmic normalization is used to control the scale effect of the expression of the spot class. Other methods of eliminating / controlling scale effects may also be used, such as minimum-maximum normalization, etc. The specific embodiments of the present application do not limit this.

[0096] Step 2033 , calculating the distance between the expression matrix of each type of spot on the two slices, and searching for a distance value greater than a preset distance threshold.

[0097] Calculate the distance between the expression matrices of each type of spot on the two slices and assign this distance value to the distance value of the corresponding spot on the two slices. This distance includes but is not limited to KL divergence distance, Euclidean distance, cosine similarity, Jensen-Shannon distance (JS distance), or other distance measures.

[0098] For example, the category information obtained by clustering and annotating different slices is based on the same reference system: for example, the first slice has two types of spots, a and b, and the second slice has three types of spots, a, b, and c. Then, we need to count the distance between a in the first slice and a in the second slice, and count the distance between b in the first slice and b in the second slice, without counting the distance between c.

[0099] Step 2034, calling the preset processing function to compress the distance value greater than the preset distance threshold; the compressed distance value and the distance value less than the preset distance threshold are used as the final distance of the expression matrix of the spot on the two slices.

[0100] The distance values ​​of each spot class are processed by a preset processing function (such as the logistic function). This function compresses the larger distance values ​​to a certain extent and maintains the smaller distance values, so that the next step of weight calculation can better highlight the spot classes with small distance values, that is, the distance value differences between the spot classes with larger weights, so that the spot classes that are more important for registration can get more refined weight distribution; the logistic function is as follows:

[0101] It should be noted that, in addition to the logistic function, the preset processing function can also be other functions whose derivatives decrease in the range of [R, +∞) (where R represents any real number less than positive infinity, and +∞ represents positive infinity), such as a logarithmic function; specifically, the embodiments of the present disclosure do not limit this.

[0102] Step 2035: Calculate the inverse of the final distance, and determine the weight corresponding to each type of spot based on the calculation result of the inverse.

[0103] Calculate the inverse of the distance value processed in step 2034 to obtain the weight, shift the weight to the minimum value of 0, and then divide each weight by the sum of all weights. The process is as follows: weight = -1*dis logistic weight=weight-min(weight) weight=weight / Σweight

[0104] In step 204, the probability transfer matrix is ​​calculated according to the weight corresponding to each type of spot and the preset regularization coefficient, and the registration score of each spot in the expression matrix corresponding to each category on the two slices is obtained.

[0105] Based on different preset regularization coefficients, the corresponding probability transfer matrices are calculated respectively, and a probability matrix with an accuracy score higher than a preset score threshold is selected from the different probability transfer matrices as the final probability transfer matrix.

[0106] For every two slices, first, calculate M AB , C1, C2 matrix: M AB The matrix stores the similarity of the expression of each two spots between the two slices. The C1 matrix stores the distance between all points in the first slice of the two slices used for calculation, and the C2 matrix stores the distance between all points in the second slice of the two slices used for calculation.

[0107] For every two slices, first, calculate M AB Matrix, C1 matrix, C2 matrix, a vector, b vector: M AB The matrix stores the similarity of expression levels of each two spots between the two slices. The C1 matrix stores the distance between all points in the first slice of the two slices used for calculation. The C2 matrix stores the distance between all points in the second slice of the two slices used for calculation. The a and b vectors respectively store the weights corresponding to different cells on the two slices.

[0108] Then, the Fused Gromov-Wasserstein version of the OT algorithm used by the PASTE algorithm will iteratively calculate the probability transfer matrix based on the above matrices and vectors, as well as a fixed regularization coefficient alpha (which is essentially the weight ratio of the structural information term and the expression information term in the cost function). The above matrices, vectors, and coefficients are collectively referred to as input parameters. In the PASTE algorithm, the user needs to specify an alpha value, and then the OT algorithm will perform calculations based on this fixed value. After multiple iterations, the data stored in the matrix tends to stabilize. The probability transfer matrix calculation for these two slices is completed. Then the calculation of the next pair of slices begins.

[0109] When calculating the probability transfer matrix, the following methods can be used, but are not limited to them. For example, the calculation problem of the probability transfer matrix is ​​formulated as an optimization problem in mathematical form, including the optimization objective function and constraints. The objective function is:

[0110] where

[0111] The constraints are:

[0112] In the above input parameters, M AB The matrix corresponds to M in the formula AB , the C1 matrix corresponds to C1 in the formula, the C2 matrix corresponds to C2 in the formula, the a vector corresponds to h in the formula, and the b vector corresponds to g in the formula.

[0113] After obtaining the above mathematical form, we use the Fused Gromov Wasserstein conditional gradient algorithm to solve it. The solution process is as follows:

[0114] Where, Eq. (7) is:

[0115] Eq. (6) is:

[0116] in,

[0117] Among the above input parameters, M AB The matrix corresponds to M in the formula AB , C1 matrix corresponds to C1 in the formula, C2 matrix corresponds to C2 in the formula, and a vector corresponds to μ in the formula X , b vector corresponds to μ in the formula Y .

[0118] In summary, to improve algorithm accuracy, this embodiment of the application introduces an automated parameter adjustment mechanism: during the calculation, the algorithm sequentially uses exponentially varying regularization coefficients to calculate the probability transfer matrix and obtain its corresponding accuracy score. After rotating through all the regularization coefficients specified for the calculation, the probability transfer matrix with the highest accuracy score is selected as the calculation result.

[0119] Step 205: Find the probability smaller than a preset threshold in the probability transmission matrix, and set the probability smaller than the preset threshold to 0.

[0120] This step is also called the post-processing process of the probability transmission matrix, which sets the element values ​​(probability) below the preset threshold to 0.

[0121] The preset threshold is calculated as follows: thresh=min(pi[pi>0])+(max(pi)–min(pi[pi>0])) / 10

[0122] In practical applications, when post-processing the probability transfer matrix, the calculation of the preset threshold is not limited to the above calculation; it can also be a percentage of the maximum probability value as the threshold, etc. In addition to the threshold comparison method, other methods can also be used, such as clustering the paired spots represented by each spot in the probability transfer matrix results based on their coordinates, retaining the cluster with a large number of spots and removing the cluster with a small number of spots, etc.

[0123] Step 206 : Divide each probability in the probability transfer matrix by the sum of the probability transfer matrices to obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

[0124] Corresponding to the above-mentioned spatiotemporal transcriptome slice registration method, the present invention also provides a spatiotemporal transcriptome slice registration device. Since the device embodiment of the present invention corresponds to the above-mentioned method embodiment, any details not disclosed in the device embodiment can be referred to the above-mentioned method embodiment and will not be further described in this invention.

[0125] FIG4 is a schematic diagram of the structure of a spatiotemporal transcriptome slice registration device provided by an embodiment of the present disclosure, as shown in FIG4 , comprising:

[0126] An acquisition unit 41 is configured to acquire two adjacent slices, each slice comprising at least one type of expression matrix, each expression matrix comprising multiple spots of spatiotemporal expression data, and each slice comprising at least one type of spot;

[0127] a determination unit 42 for determining weights corresponding to spots of the same type in each slice, wherein spots of the same type are assigned the same weight, and spots of different types are assigned different weights;

[0128] The registration unit 43 is used to calculate the probability transfer matrix according to the weight corresponding to each type of spot and the preset regularization coefficient, and obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

[0129] The present disclosure provides a registration device for spatiotemporal transcriptome slices, which obtains two adjacent slices, each slice contains at least one type of expression matrix, each expression matrix contains multiple site spots of spatiotemporal group expression data, and each slice contains at least one type of spot; the weights corresponding to the same type of spots in each slice are determined respectively, wherein the same type of spots are assigned the same weight, and different types of spots are assigned different weights; according to the weights corresponding to each type of spot and the preset regularization coefficient, the probability transfer matrix is ​​calculated respectively to obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices. Compared with the related art, the embodiment of the present disclosure assigns the same weight to the same type of spots in each slice, and different types of spots are assigned different weights, which improves the accuracy of the probability transfer matrix obtained based on the weight calculation, and thereby improves the accuracy of the registration achieved based on the probability transfer matrix.

[0130] In some aspects of this embodiment, as shown in FIG5 , the apparatus further includes:

[0131] The processing unit 44 is configured to remove spots that are not common to the two adjacent slices before acquiring the two adjacent slices.

[0132] In some aspects of this embodiment, as shown in FIG5 , the determining unit 42 includes:

[0133] The summing module 421 is used to sum the corresponding expression matrix along the spot points of the same type in each slice to obtain the expression matrix corresponding to different types;

[0134] A calculation module 422 is used to calculate the distance between the expression matrix of each type of spot on the two slices;

[0135] The first determining module 423 is configured to determine a weight corresponding to each type of spot according to the distance.

[0136] In some aspects of this embodiment, as shown in FIG5 , the determining unit 42 further includes:

[0137] The filtering module 424 is used to filter the non-highly variable genes in each slice and retain the high-variable genes in each slice before adding the corresponding expression matrices along the spot points for the same type of spots in each slice to obtain the expression matrices corresponding to different types.

[0138] In some aspects of this embodiment, as shown in FIG5 , the determining unit 42 further includes:

[0139] A search module 425 is configured to search for a distance value greater than a preset distance threshold before determining the weight corresponding to each type of spot according to the distance;

[0140] The processing module 426 is configured to call a preset processing function to compress the distance value greater than the preset distance threshold;

[0141] The second determining module 427 is configured to use the compressed distance value and the distance value less than the preset distance threshold as the final distance of the expression matrix of the spot on the two slices;

[0142] The first determination module 423 is further configured to calculate the inverse of the final distance, and determine the weight corresponding to each type of spot according to the calculation result of the inverse.

[0143] In some aspects of this embodiment, as shown in FIG5 , the apparatus further includes:

[0144] A search unit 45 is configured to calculate the probability transfer matrix based on the weight corresponding to each type of spot and the preset regularization coefficient in the registration unit 44, obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices, and then search for the probability in the probability transfer matrix that is less than a preset threshold;

[0145] A setting unit 46, configured to set the probability of being less than a preset threshold to 0;

[0146] Each probability in the probability transfer matrix is ​​divided by the sum of the probability transfer matrices to obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

[0147] In some aspects of this embodiment, as shown in FIG5 , the registration unit 44 includes:

[0148] Based on different preset regularization coefficients, the corresponding probability transmission matrix is ​​calculated respectively;

[0149] A probability matrix with an accuracy score higher than a preset score threshold is selected from different probability transfer matrices as the final probability transfer matrix.

[0150] It should be noted that the above explanation of the method embodiment is also applicable to the device of this embodiment, and the principles are the same, which is not limited in this embodiment.

[0151] 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.

[0152] FIG6 shows a schematic block diagram of an example electronic device 500 that can be used to implement an embodiment of the present disclosure. 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 assistants, 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.

[0153] As shown in FIG6 , the device 500 includes a computing unit 501 that can perform various appropriate actions and processes based on a computer program stored in a ROM (Read-Only Memory) 502 or a computer program loaded from a storage unit 508 into a RAM (Random Access Memory) 503. Various programs and data required for the operation of the device 500 can also be stored in the RAM 503. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An I / O (Input / Output) interface 505 is also connected to the bus 504.

[0154] Various components in device 500 are connected to I / O interface 505, including: an input unit 506, such as a keyboard, mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; a storage unit 508, such as a magnetic disk, optical disk, etc.; and a communication unit 509, such as a network card, modem, wireless communication transceiver, etc. The communication unit 509 allows device 500 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0155] The computing unit 501 can be a variety of general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a CPU (Central Processing Unit), a GPU (Graphic Processing Units), various dedicated AI (Artificial Intelligence) computing chips, various computing units that run machine learning model algorithms, a DSP (Digital Signal Processor), and any appropriate processor, controller, microcontroller, etc. The computing unit 501 performs the various methods and processes described above, such as the spatiotemporal transcriptome segment registration method. For example, in some embodiments, the spatiotemporal transcriptome segment registration method can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as a storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 500 via the ROM 502 and / or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the computing unit 501, one or more steps of the method described above can be performed. Alternatively, in other embodiments, the computing unit 501 may be configured to execute the aforementioned spatiotemporal transcriptome slice registration method in any other appropriate manner (eg, by means of firmware).

[0156] Various embodiments of the systems and techniques described herein can 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 can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that can 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.

[0157] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0158] 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 apparatus. 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 apparatus, 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 wires, a portable computer disk, a hard disk, RAM, ROM, EPROM (Electrically Programmable Read-Only-Memory) or flash memory, optical fiber, CD-ROM (Compact Disc Read-Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0159] 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 pointing device (e.g., a mouse or 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).

[0160] The systems and techniques described herein can be implemented in a computing system that includes back-end 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 front-end components (e.g., a user computer with a graphical user interface or 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 back-end components, middleware components, or front-end components. The components of the system can 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.

[0161] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact via a communication network. This client-server relationship is established by computer programs running on the respective computers, establishing a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host, a host product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosts and VPS services ("Virtual Private Servers" or simply "VPS"). The server may also be a server in a distributed system or a server integrated with blockchain.

[0162] It's important to note that artificial intelligence (AI) is the study of how computers can simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, and planning). This encompasses both hardware and software technologies. AI hardware technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, and big data processing. AI software technologies primarily encompass computer vision, speech recognition, natural language processing, machine learning / deep learning, big data processing, and knowledge graphs.

[0163] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0164] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for registering spatiotemporal transcriptome slices, characterized in that: include: Obtain two adjacent slices, each slice contains at least one type of expression matrix, each expression matrix contains multiple site spots of spatiotemporal group expression data, and each slice contains at least one type of spot; Determine the weights corresponding to the same type of spots in each slice, where the same type of spots are assigned the same weight; According to the weight corresponding to each type of spot and the preset regularization coefficient, the probability transfer matrix is ​​calculated respectively to obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

2. The method according to claim 1, characterized in that Before acquiring two adjacent slices, the method further includes: Remove the spots that are not common to two adjacent slices.

3. The method according to claim 1, characterized in that Determining the weights corresponding to the same type of spots in each slice includes: For the same type of spot in each slice, the corresponding expression matrix is ​​added along the spot points to obtain the expression matrix corresponding to different types; The distance between the expression matrices of each type of spot on the two slices is calculated, and the weight corresponding to each type of spot is determined according to the distance.

4. The method according to claim 3, characterized in that: Before respectively adding the corresponding expression matrixes along the spot points for the same type of spots in each slice to obtain expression matrixes corresponding to different types, the method further includes: The non-highly variable genes in each slice were filtered and the high-variable genes in each slice were retained.

5. The method according to claim 3, characterized in that: Before determining the weight corresponding to each type of spot according to the distance, the method further includes: Find the distance value greater than the preset distance threshold; Calling a preset processing function to compress the distance value greater than the preset distance threshold; The compressed distance value and the distance value less than the preset distance threshold are used as the final distance of the expression matrix of the spot on the two slices; Determining the weight corresponding to each type of spot according to the distance includes: The inverse of the final distance is calculated, and the weight corresponding to each type of spot is determined according to the calculation result of the inverse number.

6. The method according to any one of claims 1 to 5, characterized in that After respectively calculating the probability transfer matrix according to the weight corresponding to each type of spot and the preset regularization coefficient to obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices, the method further includes: Finding a probability in the probability transmission matrix that is less than a preset threshold, and setting the probability that is less than the preset threshold to 0; Each probability in the probability transfer matrix is ​​divided by the sum of the probability transfer matrices to obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

7. The method according to claim 6, characterized in that The calculating of the probability transmission matrix according to the weight corresponding to each type of spot and the preset regularization coefficient includes: Based on different preset regularization coefficients, the corresponding probability transmission matrices are calculated respectively; A probability matrix with an accuracy score higher than a preset score threshold is selected from different probability transfer matrices as the final probability transfer matrix.

8. A device for registering spatiotemporal transcriptome slices, characterized in that: include: An acquisition unit, used for acquiring two adjacent slices, each slice comprising at least one type of expression matrix, each expression matrix comprising multiple spots of spatiotemporal group expression data, and each slice comprising at least one type of spot; A determination unit, used to determine the weights corresponding to the same type of spots in each slice, wherein the same type of spots are assigned the same weight; The registration unit is used to calculate the probability transfer matrix according to the weight corresponding to each type of spot and the preset regularization coefficient, and obtain the registration score of each spot in the expression matrix corresponding to each category on the two slices.

9. 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 according to any one of claims 1 to 7.

10. 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 1-7.

Citation Information

Patent Citations

  • Method and device for performing image registration based on slices

    CN107194959A

  • Biological tissue sequence slice microscopic image registration method and apparatus

    CN107545567A

  • Point cloud registration precision evaluation method and device and electronic device.

    CN113436238A

  • Point cloud registration method and device, equipment and storage medium

    CN114387319A

  • System and method for region detection in tissue sections using image registration

    US20210035297A1