Function unit quality control method and device, electronic equipment and storage medium

Abstract

The present disclosure provides a functional unit quality control method and device, electronic equipment and storage medium, belonging to the field of biological information technology. The method comprises: acquiring spatial transcriptome data of a target slice, and performing cell type annotation on each cell in the target slice according to the spatial transcriptome data to obtain a target cell type of each cell in the target slice; calling a functional unit prediction model, and predicting a candidate functional unit with a spatial structure of the target slice according to the target cell type of each cell in the target slice and the spatial distance between the cells; determining the communication intensity of the to-be-analyzed cell according to the communication intensity of the cell pair in the spatial transcriptome data of the target slice for each cell pair containing the to-be-analyzed cell in the candidate functional unit; and filtering the to-be-analyzed cell in the candidate functional unit according to the communication intensity of the to-be-analyzed cell to obtain a target functional unit. The embodiments of the present application can reduce the false positive rate of the functional unit prediction result.

Description

Technical Field

[0001] This disclosure relates to the field of bioinformatics technology, specifically to a method and apparatus for quality control of functional units, electronic equipment, and storage medium. Background Technology

[0002] Functional units with spatial structures refer to groups of cells, molecular networks, or tissue structures in biological tissues or complex systems that have a defined spatial location and perform specific functions. For example, when the functional unit is a neurovascular unit (NVU), since the NVU is a functional unit composed of cells of various cell types working closely together, a complete functional analysis of the NVU is crucial for normal cognitive function and nervous system health. Therefore, accurate prediction of functional units can help researchers more deeply analyze the gene expression characteristics of functional units, thereby aiding in understanding the occurrence and development of related diseases.

[0003] While machine learning methods can be used to predict the performance of functional units with spatial structures, these methods have a high false positive rate. Therefore, developing a quality control method for functional units has become a pressing technical problem. Summary of the Invention

[0004] The main objective of this application is to provide a method, apparatus, electronic device, and storage medium for quality control of functional units, aiming to reduce the false positive rate of functional unit prediction results.

[0005] To achieve the above objectives, a first aspect of this application proposes a method for quality control of functional units, the method comprising: Spatial transcriptome data of the target slice is obtained, and cell type annotation is performed on each cell in the target slice based on the spatial transcriptome data of the target slice to obtain the target cell type of each cell in the target slice; The pre-trained functional unit prediction model is invoked to predict candidate functional units with spatial structure in the target slice based on the target cell type of each cell in the target slice and the spatial distance between each cell. For each cell pair containing the cell to be analyzed within the candidate functional unit, the communication strength of the cell to be analyzed is determined based on the communication strength of the cell pair in the spatial transcriptome data of the target slice; the communication strength of the cell pair is calculated by the expression level of the ligand-receptor pair in the cell pair; Based on the communication strength of the cells to be analyzed, the cells to be analyzed in the candidate functional units are filtered out, and cells whose communication strength does not meet the preset index are removed to obtain the target functional units.

[0006] In some embodiments, determining the communication strength of the cells to be analyzed based on the communication strength of the cell pairs in the spatial transcriptome data of the target slice includes: The basic communication strength of the cell pair under the ligand-receptor pair is calculated based on the expression level of the ligand-receptor pair. Based on a preset distance attenuation coefficient and the spatial distance between the cell pairs, the basic communication strength is subjected to distance attenuation processing to obtain the actual communication strength of the cell pairs under the receptor pair. The actual communication strength of the cell pairs under all receptor pairs is summarized to obtain the communication strength of the cell to be analyzed.

[0007] In some embodiments, the communication strength of the cell to be analyzed includes the total communication strength, and the process of summing the actual communication strength of the cell pair across all receptor pairs to obtain the communication strength of the cell to be analyzed includes: The actual communication strength of the cell pair under the receptor pair, representing the transmission direction, is determined as the transmission direction communication strength, and the actual communication strength of the cell pair under the receptor pair, representing the reception direction, is determined as the reception direction communication strength. Traverse each cell pair containing the cell to be analyzed, and obtain the total transmit direction communication strength and / or total receive direction communication strength of the cell pair under all receptor pairs; The total communication strength of the cell to be analyzed is obtained by summing up the communication strength of all transmitting directions and / or all receiving directions of all cell pairs corresponding to the cell to be analyzed.

[0008] In some embodiments, the communication strength of the cell to be analyzed includes significant communication strength, and the summarization of the actual communication strength of the cell pair across all receptor pairs to obtain the communication strength of the cell to be analyzed includes: The expression levels of the receptor pair in each cell of the candidate functional unit are randomly permuted multiple times, and after each random permutation, the random communication strength of the cell pair under the receptor pair is calculated based on the expression levels of the receptor pair after the random permutation. The statistics show that the random communication strength is not less than the target number of times the actual communication strength is not less than the target number of times in multiple random permutations. The significant communication data of the cell pair under the ligand-receptor pair is calculated based on the target number and the total number of permutations of multiple random permutations. If the significant communication data is less than the preset significant threshold, it is determined that there is a significant communication event of the cell pair under the ligand-receptor pair. Iterate through each cell pair containing the cell to be analyzed and obtain the total actual communication strength of the cell pair that is determined to have significant communication events under all receptor pairs; The total actual communication strength of all cell pairs corresponding to the cell to be analyzed is summarized to obtain the significant communication strength of the cell to be analyzed.

[0009] In some embodiments, the traversal includes each cell pair of the cells to be analyzed, comprising: For each cell pair containing the cell to be analyzed, if the spatial distance between the cell pairs is greater than a preset communication distance threshold, the cell pair is filtered out to iterate through the remaining cell pairs.

[0010] In some embodiments, the method for determining the distance attenuation coefficient includes: Determine the type of ligand in the ligand-receptor pair, and determine the distance attenuation coefficient based on the type of ligand.

[0011] In some embodiments, determining the distance attenuation coefficient based on the type of the ligand includes: If the ligand is membrane-bound, the distance attenuation coefficient is determined to be the first attenuation coefficient; If the ligand is secreted, the distance attenuation coefficient is determined to be the second attenuation coefficient, wherein the second attenuation coefficient is greater than the first attenuation coefficient.

[0012] To achieve the above objectives, a second aspect of this application provides a functional unit quality control device, the device comprising: The cell type annotation unit is used to acquire the spatial transcriptome data of the target slice and annotate the cell type of each cell in the target slice based on the spatial transcriptome data of the target slice to obtain the target cell type of each cell in the target slice. The functional unit prediction unit is used to call a pre-trained functional unit prediction model to predict candidate functional units with spatial structure in the target slice based on the target cell type of each cell in the target slice and the spatial distance between each cell. A communication strength determination unit is used to determine the communication strength of the cell to be analyzed for each cell pair containing the cell to be analyzed within the candidate functional unit, based on the communication strength of the cell pair in the spatial transcriptome data of the target slice; the communication strength of the cell pair is calculated by the expression level of the ligand-receptor pair in the cell pair; The cell filtering unit is used to filter the cells to be analyzed in the candidate functional units according to the communication strength of the cells to be analyzed, and remove cells whose communication strength does not meet the preset index to obtain the target functional units.

[0013] To achieve the above objectives, a third aspect of this application provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.

[0014] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0015] To achieve the above objectives, a fifth aspect of this application provides a computer program product comprising a computer program that is read and executed by a processor of a computer device, causing the computer device to perform the method described in the first aspect.

[0016] The functional unit quality control method, apparatus, electronic device, and storage medium proposed in this application first acquire spatial transcriptome data of the target slice and annotate the cell types of the target slice based on this spatial transcriptome data. Then, a pre-trained functional unit prediction model is invoked to predict functional units in the target slice using the target cell types of each cell and the spatial distances between cells. Based on this, communication strength is calculated for the predicted candidate functional units, and cells in the candidate functional units are filtered according to the calculated communication strength. Thus, this application embodiment achieves quality control of candidate functional units, effectively reducing the false positive rate of target functional units. Attached Figure Description

[0017] The accompanying drawings are provided to further understand the technical solutions of this disclosure and constitute a part of the specification. They are used together with the embodiments of this disclosure to explain the technical solutions of this disclosure and do not constitute a limitation on the technical solutions of this disclosure.

[0018] Figure 1 This is a flowchart of an embodiment of the functional unit quality control method provided in this application; Figure 2 This is a flowchart of an embodiment of the functional unit prediction model training method provided in this application; Figure 3 This is a flowchart illustrating an embodiment of determining communication strength provided in this application. Figure 4 This is a flowchart illustrating an embodiment of determining total communication strength provided in this application. Figure 5 This is a flowchart of an embodiment for determining significant communication strength provided in this application. Figure 6 This is a schematic diagram of the cell type annotation results provided in the embodiments of this application; Figure 7 This is a schematic diagram of immunohistochemical fluorescence of the hippocampus provided in an embodiment of this application; Figure 8 This is a schematic diagram of the final determined spatial distribution of the neurovascular unit provided in the embodiments of this application; Figure 9 This is a flowchart of a specific embodiment of the functional unit quality control method provided in this application; Figure 10 This is a schematic diagram illustrating the comparison and verification of cell communication strength provided in an embodiment of this application; Figure 11 This is a schematic diagram of the functional unit quality control device provided in the embodiments of this application; Figure 12 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this disclosure.

[0020] Functional units with spatial structures refer to groups of cells, molecular networks, or tissue structures in biological tissues or complex systems that have a defined spatial location and perform specific functions. For example, when the functional unit is a neurovascular unit (NVU), since the NVU is a functional unit composed of cells of various cell types working closely together, a complete functional analysis of the NVU is crucial for normal cognitive function and nervous system health. Therefore, accurate prediction of functional units can help researchers more deeply analyze the gene expression characteristics of functional units, thereby aiding in understanding the occurrence and development of related diseases.

[0021] Currently, research on functional units with spatial structures mainly relies on traditional anatomical methods and biomedical imaging techniques. However, these methods have certain limitations, making it difficult to conduct a comprehensive and in-depth analysis of the spatial structure and intercellular relationships of functional units with spatial structures. Specifically, existing technical solutions typically include the following methods: 1. Anatomical Methods: Traditional anatomical methods rely on tissue sections, immunohistochemical staining, and microscopic observation to label and observe the structures of functional units with spatial structures. These methods typically require extensive professional experience and may involve subjectivity in tissue section processing and image analysis. Due to their lack of spatial resolution, these methods struggle to reveal the complete structure of functional units with spatial structures and the relationships between cells at the cellular level.

[0022] 2. Immunohistochemical staining and microscopic imaging: Immunohistochemical staining combined with microscopic imaging is a common method for studying functional units with spatial structures. By staining specific cell types (such as neurovascular units including vascular endothelial cells, astrocytes, and neurons), researchers can observe the spatial distribution and interactions of cells. However, this method requires sophisticated tissue sample processing and may be limited by section thickness and labeling accuracy when resolving complex three-dimensional spatial structures.

[0023] 3. Gene Expression Analysis: Gene expression analysis (such as single-cell RNA sequencing) can provide information on cellular composition and gene expression of functional units with spatial structures. However, traditional gene expression techniques often lack spatial resolution, failing to reveal the precise location and interactions of cells within tissues. While the introduction of spatial transcriptomics has compensated for this deficiency, its application in the study of functional units with spatial structures still faces certain technical challenges.

[0024] 4. Functional unit prediction based on machine learning. This method has a certain proportion of false positive results in practical applications. Specifically, these false positives mainly come from: (1) relying solely on spatial distance and cell type composition for prediction without fully considering the functional synergy between cells within the functional unit; (2) lacking a quality assessment and screening mechanism for the prediction results; and (3) potentially misclassifying spatially adjacent but functionally unrelated cells as the same functional unit. These false positives can interfere with downstream analysis and affect the accurate interpretation of the biological characteristics of functional units.

[0025] Furthermore, with the development of spatial transcriptomics technology, the amount of data generated by a single experiment is growing exponentially, covering tens of thousands to hundreds of thousands of cells and thousands of genes. Existing analytical methods face the problem of low computational efficiency when processing large-scale spatial transcriptomics data, making it difficult to complete functional unit prediction and cell communication analysis at the whole tissue slice level within a reasonable time. There is an urgent need to develop efficient parallel computing strategies and optimization algorithms to improve the scalability of the analysis.

[0026] Based on this, embodiments of this application provide a method and apparatus for quality control of functional units, an electronic device and a storage medium, which can reduce the false positive rate of functional unit prediction results.

[0027] The functional unit quality control method provided in this application relates to the field of bioinformatics. This method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the functional unit quality control method, but is not limited to the above forms.

[0028] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0029] The following describes the functional unit quality control method provided in the embodiments of this application.

[0030] Reference Figure 1 In some embodiments, the functional unit quality control method provided in this application includes, but is not limited to, steps S101 to S104.

[0031] Step S101: Obtain spatial transcriptome data of the target slice, and annotate the cell types of each cell in the target slice based on the spatial transcriptome data of the target slice to obtain the target cell type of each cell in the target slice. Step S102: Call the pre-trained functional unit prediction model to predict candidate functional units with spatial structure in the target slice based on the target cell type of each cell in the target slice and the spatial distance between each cell. Step S103: For each cell pair containing the cell to be analyzed within the candidate functional unit, the communication strength of the cell to be analyzed is determined based on the communication strength of the cell pair in the spatial transcriptome data of the target slice; the communication strength of the cell pair is calculated by the expression level of the ligand-receptor pair in the cell pair. Step S104: Based on the communication strength of the cells to be analyzed, filter the cells to be analyzed in the candidate functional units, remove cells whose communication strength does not meet the preset index, and obtain the target functional units.

[0032] In this embodiment, spatial transcriptome data at single-cell resolution and a pre-trained functional unit prediction model are used to predict functional units with spatial structures. Furthermore, by introducing cell communication analysis and statistical significance testing, the accuracy of functional unit prediction is significantly improved, and the false positive rate is reduced. For ease of understanding, the training method of the functional unit prediction model is first explained.

[0033] Reference Figure 2 In some embodiments, the method for constructing the functional unit prediction model (i.e., the training method) may include, but is not limited to, steps S1 to S2.

[0034] Step S1: Obtain at least one set of sample slices containing functional units, wherein each set of sample slices containing functional units includes: Step S1.1, spatial transcriptome data containing the first cell type, is obtained by: acquiring spatial transcriptome data of a first sample slice, and classifying multiple cells in the first sample slice based on the spatial transcriptome data to obtain the first cell type of the multiple cells in the first sample slice; the spatial transcriptome data includes the expression level and spatial distribution information of the transcriptome of multiple cells in the first sample slice; and, Step S1.2, the second cell type and inter-cell distance data of the functional unit related cells are obtained by the following method: obtaining staining data of the second sample slice, determining the second cell type and first inter-cell distance data of the functional unit related cells in the second sample slice based on the staining data, wherein the first inter-cell distance data is used to describe the relative distance between the same or different cells in the second sample slice; wherein the second sample slice is the first sample slice or an adjacent slice of the first sample slice; Step S2: Train the machine learning module based on a dataset containing at least one set of sample slices with functional units to build a functional unit prediction model.

[0035] Steps S1 to S2, as illustrated in the embodiments of this application, introduce spatial transcriptome data to reveal the distance and spatial distribution characteristics between cells at the single-cell level. Combined with other technical features, this provides researchers with spatial coordinates of cells and their interactions, solving the problem of insufficient spatial resolution in related technologies. It clearly displays the fine structure of functional units with spatial structures at the cellular level, thereby enabling precise analysis of the composition and function of these units. Specifically, the embodiments of this application construct a functional unit prediction model based on cell types and inter-cell distance data from multiple cells in a first sample slice and multiple cells in a second sample slice. This model is better suited for predicting the distribution of cell types and spatial distribution of functional units, improving the accuracy of functional unit predictions.

[0036] Steps S1.1 and S1.2 can be performed in any order.

[0037] It should be noted that spatial transcriptomics technology can overcome these limitations, providing more precise data at the cellular and tissue levels. This helps researchers in fields such as neuroscience, oncology, and immunology to deeply analyze the functional units with spatial structures in cells or tissues, identify their gene expression characteristics, and thus provide new perspectives for the study of disease mechanisms.

[0038] In step S1 of some embodiments, at least one set of sample slices containing functional units is used to train the subsequent machine learning module to construct a functional unit prediction model for predicting functions. Specifically, the process of obtaining each set of sample slices containing functional units may include: step S1.1, obtaining spatial transcriptome data containing a first cell type; and step S1.2, obtaining second cell types and inter-cell distance data related to the functional units.

[0039] In step S1.1 of some embodiments, when acquiring spatial transcriptome data containing the first cell type, the first sample slice may refer to a tissue slice to be spatially sequenced. The first sample slice is a biological slice obtained in accordance with relevant legal regulations; biological slices include plant slices and animal slices. The first sample slice refers to a specific tissue slice selected during the research process, which contains the target cell population to be studied (e.g., the first sample slice for testing may contain 50,000-100,000 cells, without limitation), and serves as the basis for subsequent analysis. The spatial transcriptome data of the first sample slice refers to data obtained through spatial transcriptome technology regarding the expression level and spatial distribution of transcriptomes in multiple cells within the first sample slice. Spatial transcriptome technology can simultaneously acquire spatial information about gene expression in tissue or cell samples; this data can reflect the specific location of cells in the tissue and the characteristics of their gene expression. Cell classification refers to the process of classifying multiple cells in the first sample slice into different cell types or cell subpopulations based on their gene expression characteristics, spatial location, and other factors. Cell type refers to a identifier assigned to each cell in the first sample slice to indicate the specific cell type or cell subpopulation to which the cell belongs, such as vascular cells (endothelial cells and pericytes), astrocytes, neurons (inhibitory neurons and excitatory neurons), etc.

[0040] It should be noted that, regarding the specific methods for obtaining spatial transcriptome data from the slices (such as sample slices, target slices, etc.) in this application, to supplement this application, this application may use the patent disclosures regarding methods for obtaining spatial transcriptome data from the following patent applications by reference: European Patent Publication No. EP2697391A1, International Patent Application Publication Nos. WO2024086167A3, and WO2023115536A1. The structure and acquisition methods of all slices mentioned in the embodiments of this application can refer to the first sample slice described above, and will not be repeated here.

[0041] It should be noted that the spatial transcriptomic data of the first sample slice includes spatial location data of multiple cells in the first sample slice (i.e., spatial distribution information of the transcriptome in the first sample slice) and spatial gene expression data (i.e., expression level of the transcriptome in the first sample slice). Specifically, the spatial transcriptomic data in this application is spatial transcriptomic data with single-cell resolution, enabling accurate prediction of functional units with spatial structures, such as spatial transcriptomic data determined using the Stereo-seq method. Spatial location data refers to the specific location information of multiple cells in the tissue within the first sample slice. In spatial transcriptomics studies, researchers can divide tissue slices into multiple small regions, each called a "spot," and each spot represents a small piece of tissue (e.g., a cell in a tissue slice). Cells within this region are analyzed to determine their gene expression patterns. The spatial location of these cells can be described using coordinates, such as (x, y) coordinates in a two-dimensional plane and (x, y, z) coordinates in three-dimensional space. This coordinate information clearly defines the relative or absolute position of each cell in the first sample slice. Spatial gene expression data refers to the specific details of gene expression within multiple cells in the first sample slice. In practical applications, spatial transcriptome data refers to the expression data of all genes in cells or tissues under specific conditions and at specific time points. Spatial gene expression data reflects the transcriptional activity of each gene in each cell of the first sample slice, indicating which genes are expressed and how much. This data can be presented in matrix form, where rows represent genes, columns represent cells (or spots), and elements in the matrix represent the expression level of a particular gene in a particular cell.

[0042] In some embodiments, before classifying multiple cells in the first sample slice based on spatial transcriptome data in step S1, the functional unit prediction model construction method further includes: performing cell segmentation on the first sample slice, wherein the cell segmentation is based on the spatial transcriptome data of the first sample slice, and / or image data obtained by staining the cell nucleus and / or membrane of the first sample slice, such as ssDNA staining images.

[0043] In some embodiments, the process of selecting the first sample slice specifically includes: Spatial transcriptomic data of multiple sample slices were obtained. The spatial transcriptomic data of each sample slice included the expression level and spatial distribution information of the transcriptome of each cell in that sample slice. Data extraction was performed based on spatial transcriptome data to obtain the number of genes and the proportion of mitochondrial genes in each cell of each sample slice. Multiple sample slices were screened based on the number of genes and the proportion of mitochondrial genes in each cell to determine the first sample slice.

[0044] In this embodiment, multiple initial sample slices can be screened first, i.e., quality control is performed on the initial sample slices to remove low-quality cells and slices, and the initial sample slices that meet the quality control standards are used as the first sample slices for constructing a functional unit prediction model for predicting functional units with spatial structures. Specifically, spatial sequencing data and corresponding spatial transcriptome data of multiple spatial sites in each initial sample slice are first obtained. The spatial transcriptome data includes spatial gene expression data of multiple cells in the initial sample slice (i.e., the expression level of the transcriptome of multiple cells in the initial sample slice) and spatial location data (i.e., the spatial distribution information of the transcriptome of multiple cells in the initial sample slice).

[0045] Spatial sequencing data refers to data on transcriptome expression and spatial location obtained by performing spatial transcriptome sequencing technologies (such as Visium, Visium HD, Visium HD3', Stereo-seq, etc.) on each initial sample slice. For example, Visium uses spatial probes on a chip, including spatial barcodes (each location on the chip has a unique spatial barcode sequence), UMIs, and capture sequences (poly T), to capture mRNA in tissues. The spatial probes use the mRNA as a template for extension, and the resulting extension products include the mRNA sequence information, the corresponding UMI information, and the spatial location information of the mRNA (spatial barcode sequence information). The resulting extension products can be decoded by sequencing (such as bridge sequencing, DNB sequencing). Each location (i.e., spatial site) is also called a spot. Generally, a spot has a cluster of spatial probes with the same spatial barcode. UMIs are used to distinguish copies of different mRNA molecules, avoiding the duplicate counting problem caused by polymerase chain reaction (PCR) amplification.

[0046] Furthermore, after determining the spatial transcriptome data of multiple sample slices, a spatial density image of the initial sample slice can be generated based on the spatial sequencing data and the spatial coordinates of each spatial site in the initial sample slice. Specifically, this application can summarize the total UMI count of each spot point in the spatial coordinates of the sample slice based on the spatial sequencing data (specifically, this can be achieved by identifying and counting different UMI sequences at each spot point to obtain the total UMI count corresponding to that spot point. This process can be performed using specialized bioinformatics software and algorithms, without limitation), and generate a spatial density matrix for each sample slice based on the total UMI count and the spatial coordinates of each spatial site. Then, this spatial density matrix is ​​converted into an image to obtain a spatial density image, where the grayscale intensity in the image reflects the number of UMIs.

[0047] It should be noted that, for the spatial density matrix, this embodiment of the application can associate the spatial coordinates of each spot point with the corresponding total UMI count, that is, to map these spatial coordinates one-to-one with the corresponding total UMI count, forming a dataset. Further, based on the spatial layout of the sample slices and the distribution of the spot points, the above dataset is organized into a matrix. The rows and columns of the matrix correspond to the spatial positions on the initial sample slices (e.g., rows correspond to the row coordinates of the sample slices, and columns correspond to the column coordinates of the sample slices), and each element in the matrix represents the total UMI count corresponding to the spot point at that position. This generates the spatial density matrix, and the element values ​​in the matrix reflect the gene expression density at the corresponding position.

[0048] Furthermore, embodiments of this application can perform cell segmentation on a spatial density image based on the spatial distribution information of the transcriptome of multiple cells in an initial sample slice, thereby determining the cell region of each cell in the spatial density image. Specifically, embodiments of this application can use a preset cell segmentation model (such as the ESPANet model) for cell segmentation, and post-process the initial spatial gene expression data of the initial sample slice using a watershed algorithm to finally generate a cell-gene matrix, which is used to create Seurat RDS objects of cellbins (in the field of spatial transcription technology, a bin can refer to the process of merging spatially adjacent or similar cells into a single unit for analysis).

[0049] The embodiments of this application can also perform cell segmentation based on optical images of initial sample slices, for example, using H&E staining images or mIF images for cell segmentation.

[0050] It should be noted that the watershed algorithm is a morphology-based image segmentation algorithm that treats an image as a terrain with peaks and valleys, segmenting the image by submerging seed regions and progressively expanding them. In cell segmentation, the watershed algorithm can utilize cell boundary information to further refine the segmentation results of the ESPANet model, address issues such as cell adhesion, optimize the segmentation results, and thus more accurately determine the cell region of each cell in the initial sample slice within the spatial density image.

[0051] Furthermore, based on cellular regions, data extraction was performed on the initial spatial gene expression data to obtain the number of genes (nFeatures), the proportion of mitochondrial genes (percent.MT), and the total gene expression level for each cell in the initial sample slice. Here, nFeatures refers to the number of genes in each cell, i.e., the number of different genes detected in each cell. This is obtained by counting the number of non-zero elements in each row of the cell-gene matrix. nCounts represents the total expression level of all genes in each cell, i.e., the total number of UMIs detected in each cell. This is obtained by summing the elements in each row of the cell-gene matrix. percent.mt refers to the proportion of mitochondrial genes in each cell. This is achieved by first identifying which genes are mitochondrial genes, then counting the number of mitochondrial genes in each cell, dividing by the cell's nFeatures value, and finally multiplying by 100 to obtain the percent.mt value. Thus, the quality control standard of this application embodiment may include: at least 50% of the cells in a tissue slice must simultaneously meet the following requirements: (1) the number of genes detected in the cells (nFeature) is greater than 100; and (2) the proportion of mitochondrial genes (percent.MT) in the cells is less than 20%. That is to say, only when at least 50% of the cells in the initial sample slice meet the preset quality control standard can it be considered as qualified, and then the initial sample slice can be used as the first sample slice. The quality control standard can be flexibly set as needed, but it must be ensured that at least 50% of the cells simultaneously meet the quality control standard.

[0052] In the above embodiments, the embodiments of this application can obtain gene expression profiles from these spot points, thereby analyzing idle data. Furthermore, the embodiments of this application can determine cell outlines and delineate cells (i.e., cell segmentation) based on the gene enrichment at the spot points, or any cell segmentation method known to those skilled in the art, such as ssDNA staining or DAPI staining, to achieve single-cell resolution and obtain multiple cellbins. Specifically, the embodiments of this application can generate cell-gene matrices corresponding to cellbins according to the SAW (Stereo-seq Analysis Workflow), and perform quality control on multiple initial sample slices based on this matrix to select the higher-quality initial sample slices for subsequent processing, thereby improving the prediction accuracy of functional units with spatial structures.

[0053] In some embodiments, the step of classifying multiple cells in a first sample slice based on spatial transcriptome data to obtain a first cell type of multiple cells in the first sample slice may specifically include: Using single-cell transcriptome sequencing technology, single-cell transcriptome sequencing data of multiple cells from the source sample in the first sample slice were obtained. Based on the expression level of the transcriptome of multiple cells in the first sample slice and the single-cell transcriptome sequencing data, the common gene expression data of each cell in the first sample slice were determined. Based on common gene expression data and preset cell marker genes, cell annotation was performed on multiple cells in the first sample slice to obtain the first cell type of multiple cells in the first sample slice.

[0054] Single-cell transcriptome sequencing data refers to data determined using single-cell sequencing or single-cell nuclear sequencing (snRNA-seq) methods. This single-cell transcriptome sequencing data serves as a reference dataset for cell annotation. Common gene expression data refers to the intersection of gene expression levels in the transcriptomes of multiple cells in the first sample slice and the gene data from the single-cell transcriptome sequencing data.

[0055] Furthermore, in this embodiment, multiple cells in a first sample slice can be classified based on common gene expression data and preset cell marker genes to obtain a first cell type for each cell in the first sample slice. The preset cell marker genes refer to genes specifically expressed or highly expressed in a specific cell type. These genes can serve as markers for identifying cell types. The first cell type is equivalent to a cell ID in the first sample slice, because each classified cell is assigned an ID when it is circled to represent its identity (and the expression of all genes within this range also belongs to this cell). Further, after cell classification, a column of annotation related to this ID can be added to the spatial expression matrix of the spatial transcriptome data (a matrix used to characterize the expression levels of the transcriptome of multiple cells in the first sample slice), such as cell1: Neuron. Thus, in this embodiment, after classifying each cell using an algorithm, the obtained multiple first cell types can be mapped to a single-cell resolution spatial expression matrix, that is, a label is added to each cell in the first sample slice according to the determined cell ID, without regenerating the matrix.

[0056] Understandably, after determining the first sample slice that meets the quality standards, the embodiments of this application can use the Spatial-ID algorithm to classify the cells in the first sample slice that meets the quality standards. This algorithm combines the common genes of snRNA-seq data and Stereo-seq data (i.e., target spatial gene expression data) to predict cell types.

[0057] In some embodiments, the step of determining common gene expression data for each cell in the first sample slice based on the expression levels of transcriptomes in multiple cells in the first sample slice and single-cell transcriptome sequencing data may specifically include: Based on the expression levels of transcriptomes of multiple cells in the first sample slice, the first gene expression matrix of multiple cells in the first sample slice was determined. The second gene expression matrix of multiple cells in the first sample slice was determined based on single-cell transcriptome sequencing data. Based on the common genes of multiple cells in the first gene expression matrix and the second gene expression matrix, the common gene expression data of each cell in the first sample slice is obtained.

[0058] In this embodiment, the spatial gene expression data (i.e., the expression levels of transcriptomes from multiple cells in the first sample slice) and the gene expression matrix corresponding to single-cell transcriptome sequencing data from the first sample slice are extracted separately. Then, the intersection of the two gene expression matrices is taken as the common gene expression data. The first gene expression matrix is ​​a matrix constructed based on the expression levels of transcriptomes from multiple cells in the first sample slice. The rows of the matrix typically represent genes, and the columns represent cells in the first sample slice. Each element in the matrix represents the expression level of a gene in a specific cell, describing the gene expression situation of cells in the first sample slice. The second gene expression matrix is ​​a matrix constructed based on single-cell transcriptome sequencing data, also with genes as rows and cells as columns. Each element represents the expression level of a gene in a cell. This matrix does not contain spatial location information of cells. The common gene expression data refers to the gene expression information of each cell in the first sample slice that exists in both the first and second gene expression matrices. These genes are reflected in both data sources and can be used for comprehensive analysis of cell gene expression characteristics.

[0059] In some specific embodiments, the embodiments of this application can optimize cell type probabilities by utilizing spatial neighborhood information and gene expression profiles through Graph Convolutional Networks (GCNs) and self-supervised learning. Ultimately, the cell type of each cell determines the identity (i.e., the first cell type) of each cell in the spatial chip based on the highest probability generated by the GCN.

[0060] In step S1.2 of some embodiments, the second cell type refers to the cell type of the cells associated with the functional unit in the second sample slice. The first intercellular distance data describes the relative distance (i.e., spatial distance) between cells of the same or different cell types in the second sample slice, reflecting spatial distribution differences, such as adjacent cells or cells far apart. Staining data refers to data obtained after specific staining treatment of the second sample slice. Staining can mark specific structural or molecular features of cells, aiding in subsequent analysis of information such as intercellular distances. The stained image can be a pathological staining image. The second sample slice can be a first sample slice or an adjacent slice of the first sample slice. An adjacent slice refers to a continuous slice that is close to the first sample slice in the three-dimensional structure of the tissue, without limitation. Therefore, by obtaining staining data, the cell types and spatial distribution of cells required to construct the functional unit can be determined. In the following embodiments, this application uses a neurovascular unit (a functional unit) as an example for detailed explanation.

[0061] It should be noted that the embodiments of this application are described in detail below using a neurovascular unit (a functional unit) as an example. In order to enhance the robustness of the constructed functional unit prediction model in the analysis of neurovascular units, this application can also implement data augmentation strategies (such as image transformation of staining data): under the premise of maintaining cell affiliation and biological structural constraints (such as topological invariance of neuron-vascular connection), geometric diversity is introduced by random rotation (such as rotation ±45°), flipping, translation (such as moving ±15% of the size); and / or, combined with optical adjustment (Gaussian blur to simulate defocus, Poisson noise to simulate low-light imaging, channel perturbation to adapt to staining differences) and bio-specific deformation (local occlusion to avoid key synapses, elastic deformation to protect unit integrity).

[0062] Therefore, embodiments of this application can perform spatial analysis on second sample slices based on immunostaining technology to analyze the spatial distribution of cells of the same or different cell types within functional units with spatial structures (such as astrocytes, vascular endothelial cells, neurons, etc. in neurovascular units). The relative distances between cells of the same or different cell types (such as Euclidean distances between cells) are statistically analyzed and extracted, and these distance feature data are organized for subsequent model construction.

[0063] It should be noted that the staining data can be image data obtained through specific staining techniques, such as immunostaining. The first sample slice is processed to make the cells in the first sample slice appear in different colors or marked, facilitating subsequent observation and analysis of cell distribution and characteristics. The process of determining the distance data between the first cells can utilize image analysis software (such as ImageJ) or bioinformatics tools (such as Phenoptr) to calculate the relative distances between identical or different cells in the second sample slice based on the staining data and the spatial location data of the cells. For example, algorithms such as Euclidean distance can be used to obtain the distance information between each cell and other cells, thereby describing the spatial distribution differences between cells; alternatively, the centers of the cell nuclei in the stained image corresponding to the staining data can be marked first, and then the distances can be statistically obtained by reading the image using the skimage method. For example, since cell type labels can include vascular cells (endothelial cells and pericytes), astrocytes, neurons (inhibitory neurons and excitatory neurons), by analyzing the relative spatial distribution of these cells, the distance between vascular cells, as well as the distance between astrocytes, neurons and vascular cells, can be determined. This helps to deepen the understanding of the spatial structure and functional associations of neurovascular units, and provides key distance feature data for the accurate prediction and analysis of neurovascular units.

[0064] It should be noted that the images corresponding to the staining data can be staining images obtained by staining one or more of the cell nucleus, nuclear membrane, cytoplasm, and cell membrane, such as the cell nucleus and cytoplasm, and the immune repertoire can be stained only for a few specific types of cells.

[0065] It should be noted that the dataset for each group of sample slices containing functional units in this embodiment may further include: spatial transcriptome data containing the first cell type, and fourth cell type and third intercellular distance data of cells related to the functional units in the second sample slice, and another dataset is determined based on the staining data after image transformation of the second sample slice. That is, it is not necessary to provide real biochemical slices to obtain another dataset.

[0066] In step S2 of some embodiments, further, embodiments of this application can train a machine learning module based on a dataset of at least one set of sample slices to construct a functional unit prediction model. Since this functional unit prediction model is constructed based on relevant information of the functional unit indicated by the second sample slice, it can be used to predict the spatial structure of functional units identical to that functional unit. After training the functional unit prediction model, it can be used to accurately predict the composition and function of functional units with spatial structures.

[0067] It should be noted that the machine learning module in this application embodiment includes at least one of the following: supervised learning module, semi-supervised learning module, unsupervised learning module, regression analysis module, reinforcement learning module, self-learning module, feature learning module, sparse dictionary learning module, anomaly detection module, generative adversarial network or association rule module, without limitation.

[0068] It should be noted that the embodiments of this application can make predictions based on the transcriptome at single-cell resolution, thus providing greater accuracy. These embodiments can analyze the specific cell types contained within the predicted functional unit's spatial structure, examine pathological changes, enhance the analytical capability for dynamic changes in functional units with spatial structures, and accurately predict their changes under different physiological or pathological states, providing a basis for the development of personalized treatment plans.

[0069] In some embodiments, the step of training a machine learning module based on a dataset containing at least one set of sample slices of functional units to construct a functional unit prediction model may specifically include: Spatial adjacency data of the first sample slice is constructed based on spatial transcriptome data containing the first cell type. The spatial adjacency data is used to describe the adjacency structure between the same or different cells in the first sample slice. Based on the matching results of the second cell type and the first cell type of the related cells of the functional unit, as well as the distance data between the first cells, the unit spatial data related to the functional unit is extracted from the spatial adjacency data; Based on the spatial structure indicated by the unit spatial data, the central cell type, associated cell type, and distance thresholds between central cells, between associated cells, and between central cells and associated cells are determined from the first cell type; the central cell refers to the cell that constitutes the functional unit, and the associated cell refers to the non-central cell that has inter-cell communication with the central cell. The machine learning module is trained based on the central cell type, associated cell types, and distance threshold to construct a functional unit prediction model.

[0070] In this embodiment, the spatial transcriptome data containing the first cell type can be presented in matrix form, where rows represent genes and columns represent cells at spatial locations, thus determining the spatial location of each cell in the first sample slice. An embodiment of this application can set a proximity distance threshold to determine which cells are considered adjacent to each other, constructing spatial adjacency data. This spatial adjacency data can be in matrix form, i.e., a spatial adjacency matrix. The spatial adjacency matrix can be used to describe the adjacency structure between the same or different cells in the first sample slice. Therefore, this embodiment of the application can establish a spatial adjacency matrix based on the first sample slice, and extract the distance features of cells within the neurovascular unit from the immunohistochemical staining of the first sample slice or a neighboring second sample slice. Then, the functional unit prediction model is established by integrating the information from both.

[0071] In this process, the cell types required to construct the functional unit can be determined based on the second cell type associated with the functional unit. Then, the second cell type and the first cell type associated with the functional unit can be matched. Furthermore, based on the relative distances between identical or different cells in the second sample slice, spatial data of the functional unit is extracted from the spatial adjacency data. This spatial data then indicates the spatial structure associated with the spatial distribution of cells associated with the functional unit matched from the first sample slice.

[0072] The central cell type indicates the cell type that plays a crucial role in the functional unit. For example, when the target functional unit is a neurovascular unit, the corresponding central cell type may include endothelial cells and pericytes. Therefore, the central cell type indicates the cells that constitute the functional unit. The associated cell type indicates the cell types that are allowed to exist in the functional unit, such as neurons and astrocytes. Therefore, the associated cell type indicates non-central cells that communicate with the central cell. Matching the first cell type based on the central cell type and associated cell type allows for the selection of central and associated cells relevant to the spatial structure of the functional unit from multiple cells indicated by the first cell type, avoiding the influence of invalid cell types on the prediction of the specific spatial structure of the functional unit.

[0073] The distance threshold between central cell types and associated cell types can refer to the distance threshold between central cells, between associated cells, and between central cells and associated cells; that is, the maximum distance between identical or different cells within the same functional unit. When the distance between a central cell and an associated cell is greater than the distance threshold, they belong to different functional units.

[0074] It should be noted that the process of constructing the functional unit prediction model based on the central cell type, associated cell types, and distance threshold in this application embodiment can be specifically represented as follows: First, construct a set of central cells based on the central cell type, which can be represented as " ",in, This represents the set of central cells selected from multiple cells in the first sample slice, where j represents the cells corresponding to the central cell type in the first sample slice, and cells represents the set of multiple cells in the first sample slice. This indicates the cell type corresponding to cell j. This represents the set of central cell types. Further, the process of identifying cells in the first sample slice that match the second cell type associated with the functional unit can be equivalent to: constructing a distance matrix based on the first inter-cell distance data using the Euclidean distance between points (points refer to the distance between the center points of two cells). ,like ,in, Represents cell pairs in the distance matrix （ The relative distance between cells () () indicates cell pairs of the same or different cell types in the second sample slice. and These represent the spatial location data of the corresponding cells, and ( () represents the candidate distance between each pair of cells.

[0075] Based on this, the spatially structured functional unit prediction model constructed using the distance matrix, central cell type, associated cell type, and distance threshold can be expressed as the following formula: , in, This indicates the selection of a set of cells from multiple cells in the first sample slice that match the central cell type. The distance matrix is... In the matrix The data corresponding to the coordinates, where d represents the distance threshold. This represents a set composed of related cell types. They respectively indicate different associated cell types in the functional unit. Indicates the cells in the first sample slice The corresponding first cell type belongs to a related cell type. Indicates the number is The functional unit can be a single sample slice containing multiple functional units. The formula above describes how, by traversing all cells in the first sample slice and combining the distance matrix, the set of central cells, and the set of associated cell types, all cells contained in a functional unit are identified and integrated to form at least one functional unit. For example, when the functional unit is a neurovascular unit (NVU), the constructed functional unit prediction model is the NVU model.

[0076] The distance threshold refers to the maximum distance between identical or different cells within the same functional unit. Therefore, when the intercellular distance between the associated cell and the central cell is greater than the distance threshold, the cell indicated by the associated cell type is considered to be a cell that does not meet the structural requirements of the functional unit to which the central cell belongs.

[0077] In the above embodiments, this application can establish a distance matrix of cells within a functional unit with spatial structure by combining the Euclidean distance algorithm, and train a functional unit prediction model using machine learning methods by combining the definition of functional units with spatial structure and the spatial positions between cells. For example, the distance differences between vascular endothelial cells, astrocytes, and neurons can reflect the functional state of the blood-brain barrier and the integrity of neurovascular units. The distance features between cells are input into the model, such as: celltype1-celltype2: 30 (30 distance units); celltype1-celltype2: 350 (350 distance units), to predict functional units with spatial structure. The established functional unit prediction model involves traversing the central cell set, identifying all cell types in the distance matrix that satisfy the above distance feature set conditions, and then defining them as the same unit. This functional unit prediction model predicts a spatial functional structure.

[0078] In some embodiments, the present application embodiments may perform functional unit prediction on the third sample slice using a functional unit prediction model, and compare it with the coloring data of the fourth sample slice to determine the accuracy level of the machine learning module, wherein the fourth sample slice is the third sample slice or a neighboring slice of the third sample slice.

[0079] In other words, after establishing a functional unit prediction model (such as a prediction model based on neurovascular units), this application can also use experimental slides stained with immunohistochemistry (i.e., immunofluorescence) (i.e., the fourth sample slide) as a verification standard. The model prediction results (i.e., the spatial distribution of target cells corresponding to the predicted neurovascular units in the third sample slide) are compared with the spatial distribution of the immunohistochemical staining image, and the spatial location of the predicted neurovascular units is matched with the locations of the marked blood vessels and neurons in the immunohistochemical staining image. Furthermore, by calculating the overlap rate (i.e., cell overlap) between the spatial distribution of multiple cells in the predicted neurovascular units and the spatial distribution of cells in the actual labeled functional units, the proportion of intersection between the predicted neurovascular unit region and the immunofluorescence-labeled region is determined, thereby allowing for statistical analysis of the prediction accuracy of the functional unit prediction model. The experimental verification results show that the overlap between the predicted neurovascular units and the immunofluorescence staining image is approximately 75%, which also reflects the model's accuracy in spatial structure prediction. After accurately determining the cells corresponding to the functional units, the interrelationships between the cells in the functional units can be clarified by combining the cell type characteristics and their spatial locations.

[0080] It should be noted that after determining the cell corresponding to the functional unit, the embodiments of this application can also visualize the prediction results in order to intuitively show the spatial organization structure and functional characteristics of the functional unit with spatial structure.

[0081] It should be noted that, during the training process of the functional unit prediction model, this embodiment can use the neurovascular unit in the human hippocampus sample as an example to predict and analyze the neurovascular unit. The original process of the distance feature extraction model in this embodiment can be divided into four parts: data preprocessing, constructing a convolutional neural network (CNN) model, training the model, and feature extraction. Specifically, the image data is first preprocessed to accurately label different cell types (such as Neuron, Astro, etc.) and neurovascular units, that is, to clarify the unit and category to which the cell belongs, and to normalize the image pixel values ​​to the range of 0-1. Next, a CNN model is constructed, specifically through the following sequence: a first convolutional layer (using 16 3x3 kernels, stride 1, padding 1, and ReLU activation function to extract initial image features), a first pooling layer (using 2x2 max pooling with a stride 2 to downsample the feature map and reduce feature dimensionality), a second convolutional layer (using 32 3x3 kernels, stride 1, padding 1, and ReLU activation function to further extract more complex features), a second pooling layer (again using 2x2 max pooling with a stride 2 for downsampling), and a fully connected layer (flattening the pooled features and connecting them to a fully connected layer; the number of neurons can be set to 128, and the activation function is ReLU). During training, the preprocessed data can be divided into training and validation sets. Multiple rounds of training are performed on the training set, with loss calculated and model parameters updated in each round. Simultaneously, model performance is monitored on the validation set to prevent overfitting. After the model is trained, test images are input into the model, and feature vectors are obtained from the output layer. These feature vectors contain information such as distance features between different cell types, which can be used for subsequent analysis.

[0082] The training process of the functional unit prediction model is explained below in conjunction with the training labels.

[0083] In some embodiments, before training the functional unit prediction model, a distance threshold is first determined based on manually annotated immunohistochemical images (i.e., second sample slice images). d (or distance threshold vector). Specifically, manually labeled immunohistochemical sample slices (i.e., second sample slices) can be obtained, wherein the manual labeling results are used to indicate whether the cells in the sample slice belong to the same neurovascular unit. For any cell pair in the immunohistochemical sample slice, the spatial distance between the two cells in the cell pair is calculated. And determine the reference label for the cell pair based on the results of manual annotation. In this case, when two cells are manually labeled as belonging to the same neurovascular unit, the reference label is used. The label is positive; when two cells are not labeled as belonging to the same neurovascular unit, the label is referenced. These are negative sample labels.

[0084] In order to learn the distance threshold d Distance discrimination functions can be constructed: , in, This represents the Sigmoid function, where τ is the smoothing coefficient (optionally, to accommodate differences in interaction distances between different cell types, this parameter can be configured as an independent threshold for specific cell type combinations (such as neuron-endothelial cells, glial cells-endothelial cells), rather than a single global value). This represents the probability that cell pairs belong to the same functional unit. When Less than hour, Larger; when Greater than hour, Smaller. Furthermore, it can be based on reference labels. With predicted probability Construct the loss function: , By analyzing the loss function with respect to Find the derivative and update using gradient descent, quasi-Newton methods, or other optimization methods. This continues until the loss function converges, thus obtaining a distance threshold suitable for immunohistochemical sample slices. d .

[0085] Determining the distance threshold d Then, based on the spatial distance between cells in the sample slice, cell type, and the learned distance discrimination relationship, the functional unit labels of each cell pair in the sample slice can be assigned values. Specifically, the cell pairs in the sample slice are traversed, and for any cell pair, if the spatial distance between the two cells in the pair is not greater than a distance threshold... d If two cells have different cell types, and the sample cell type of one cell belongs to the central cell type set while the sample cell type of the other cell belongs to the non-central cell type set (i.e., the associated cell type set), then the functional unit label corresponding to that cell pair is determined as a positive sample label; otherwise, the functional unit label corresponding to that cell pair is determined as a negative sample label.

[0086] Based on the cell type of each cell in the immunohistochemical sample slice, the spatial distance between each cell, and the results of manual annotation or distance thresholds. dThe functional unit labels are used to train a functional unit prediction model. This model learns the correspondence between cell type combinations, spatial distances and their relationships, and the probability of neurovascular unit formation. The functional unit prediction model can be a machine learning classifier such as Support Vector Machine, Random Forest, Gradient Boosting Tree, or XGBoost; there are no limitations on its application. Finally, the trained functional unit prediction model can be deployed to unknown spatial transcriptome data, using a standardized processing workflow to quickly and initially predict all cells within a whole slice, outputting candidate functional units and a detailed list of their constituent cells.

[0087] The following describes the actual inference process after the functional unit prediction model has been trained.

[0088] In step S101 of some embodiments, spatial transcriptome data of the target slice is acquired. The target slice can refer to a slice derived from the biological tissue to be analyzed (e.g., human hippocampus, liver, kidney, or tumor tissue). The target slice needs to undergo spatial transcriptome sequencing library construction and sequencing processes to obtain spatial transcriptome data usable for downstream analysis. It is understood that the methods for acquiring spatial transcriptome data of the target slice, the methods for annotating cell types (i.e., cell types) of each cell in the target tangent, and the methods for quality control of the target slice and the cells in the target slice are similar to those described in the aforementioned functional unit prediction model training method, and will not be repeated here.

[0089] In step S102 of some embodiments, a pre-trained functional unit prediction model is invoked, i.e., a machine learning classifier or discriminant function that has been trained through the aforementioned steps and has converged to its optimal parameter state. The core function of this functional unit prediction model is to automatically output a binary discrimination result representing whether multiple cells can constitute the same functional unit based on the input spatial distance features and cell type combination features. Thus, when the target cell type of each cell in the target slice and the spatial distance between each cell are input into the functional unit prediction model, candidate functional units can be obtained. A candidate functional unit can refer to a cell group output by the functional unit prediction model after preliminary prediction that spatially meets the distance requirement from the central cell and whose cell type composition satisfies the requirements of the corresponding cell type set.

[0090] In specific implementation, the reasoning process of the functional unit prediction model includes: First, selecting cells belonging to a preset central cell set (e.g., for a neurovascular unit, this central cell set includes endothelial cells and pericytes) from the target cell types obtained in step S101 as spatial anchors for candidate functional units. Then, traversing all other cells in the target slice except for the spatial anchor, if a cell's target cell type belongs to the non-central cell cell type set corresponding to the candidate functional unit (e.g., astrocytes and neurons), and the distance between the cell and the nearest spatial anchor is not greater than the optimal distance threshold learned by the model, the functional unit prediction model outputs a positive judgment, classifying the cell as a candidate functional unit anchored by the spatial anchor; otherwise, it is not classified. By traversing all cells in the entire target slice except for the spatial anchor and repeating the above judgment process, the complete candidate functional units of the target slice are finally obtained. Each candidate functional unit contains one or more central cells, and a list of non-central cells that meet the spatial and cell type conditions determined based on the central cells.

[0091] In step S103 of some embodiments, cell pairs corresponding to each cell to be analyzed within a candidate functional unit are determined, and the expression levels of ligand-receptor pairs (i.e., ligand-receptor pairs) in the cell pair are determined. The cells to be analyzed can refer to all cells within the candidate functional unit. A cell pair includes the cell to be analyzed and any other cell within the candidate functional unit; each cell pair is treated as an independent unit for calculating communication strength. A ligand-receptor pair can refer to a known combination of molecular interactions consisting of secreted signaling molecules (i.e., ligands) and their specific cell surface or transmembrane receptors, which enables the directional transmission of signals from sending cells to receiving cells. Ligand-receptor pairs can be derived from publicly available standard ligand-receptor pair interaction databases (e.g., the CellChat database or the CellPhoneDB database), which contain known ligand-receptor pairs and their interactions. In practical implementation, the database first needs to be preprocessed: the gene names of all ligand-receptor pairs included in the database are precisely mapped to the gene names in the target slice spatial transcriptome data to screen out a set of valid ligand-receptor pairs that are simultaneously expressed in the target slice spatial transcriptome data and have reliable biological interactions. The expression level of the ligand-receptor pair refers to the expression level of the ligand gene in the sending cell of the selected ligand-receptor pair and the expression level of the receptor gene in the receiving cell of the cell pair. These expression levels can be directly read from the gene expression matrix of the target slice spatial transcriptome data. After determining the expression levels of the ligand-receptor pair in the cell pair, the communication strength of the cell pair can be calculated based on these expression levels, and the communication strength of the cell to be analyzed can be determined based on the communication strength of the cell pair. Communication strength can be a numerical indicator that quantifies the strength of functional interactions between the cell to be analyzed and other cells within the candidate functional unit, and can serve as a quantitative basis for subsequent quality screening.

[0092] Reference Figure 3 In some embodiments, step S103, "determining the communication strength of the cells to be analyzed based on the communication strength of cell pairs in the spatial transcriptome data of the target slice", may include, but is not limited to, steps S301 to S303.

[0093] Step S301: Calculate the basic communication strength of cell pairs under ligand-receptor pairs based on the expression levels of the ligand-receptor pairs. Step S302: Based on the preset distance attenuation coefficient and the spatial distance between cell pairs, the basic communication strength is subjected to distance attenuation processing to obtain the actual communication strength of the cell pair under the ligand-receptor pair. Step S303: The actual communication strength of cell pairs under all ligand-receptor pairs is summarized to obtain the communication strength of the cell to be analyzed.

[0094] In step S301 of some embodiments, the basic communication strength can refer to the initial theoretical signal strength determined solely by the ligand expression level of the sending cell and the receptor expression level of the receiving cell, without considering spatial distance attenuation effects. It reflects the molecular potential for signal transduction between the cell pair and the receptor pair under ideal zero-distance conditions. In specific implementations, for a given cell pair (i,j) and a given receptor pair (L,R), the expression level of ligand L in cell i is read from the gene expression matrix of the target slice spatial transcriptome data. and the expression level of receptor R in cell j Then, the basic communication strength is calculated according to the following formula 3. .

[0095] (Formula 3), In step S302 of some embodiments, a preset distance attenuation coefficient can be used. (If set to 100 micrometers, this is not a limitation) and the spatial distance between cell pairs (i,j) affects the basic communication strength. Distance attenuation processing is performed to obtain the actual communication strength between cell pairs (i,j). Specifically, distance attenuation can be performed according to the method shown in Formula 4 below.

[0096] (Formula 4), in, This represents the spatial distance between cell pairs (i,j). Equation 4 shows that when the spatial distance between cell pairs (i,j) is much smaller than the distance decay coefficient... At that time, the attenuation factor Approaching 1, the actual communication strength is close to the baseline communication strength, meaning the signal reaches the receiving cell with almost no loss. This occurs when the spatial distance between cell pairs (i,j) is much greater than the distance attenuation coefficient. At that time, the attenuation factor Approaching zero, the actual communication strength drops sharply, meaning the signal attenuates significantly during propagation and becomes difficult to reach the receiving cell effectively. It is understandable that, in addition to the exponential attenuation shown in Formula 4, linear attenuation, Gaussian attenuation, or the sigmoid function can also be used; there is no specific limitation on the attenuation function used for distance attenuation.

[0097] In some embodiments, distance attenuation coefficient The methods for determining this include: Determine the type of ligand in the ligand-receptor pair, and determine the distance attenuation coefficient based on the type of ligand.

[0098] In the embodiments of this application, the type of ligand L in the ligand-receptor pair (L,R) can refer to the category classified according to the biochemical characteristics and mode of action of the ligand molecule, such as being divided into two main categories: membrane-bound ligands and secretory ligands. Membrane-bound ligands are typically anchored to the cell membrane surface of the sending cell, and their signal transduction depends on direct intercellular contact; therefore, their effective action distance is extremely short. Secretory ligands, on the other hand, are released into the extracellular space in soluble molecular form and can exert biological effects over a longer distance through diffusion; therefore, their effective action distance is longer. The distance attenuation coefficient can be determined based on the type of ligand L. Distance attenuation coefficient Distance attenuation coefficient is a numerical parameter used to quantify the rate at which the signal intensity of a ligand molecule decays with increasing spatial distance during diffusion and transport in the interstitial space. The larger the value, the greater the effective signal strength the ligand can maintain at a greater distance. Distance attenuation coefficient. The smaller the value, the more limited the effective range of the ligand is to the vicinity of the sending cell. Therefore, for membrane-bound ligands, a smaller distance attenuation coefficient can be set. For example, it can be set as the first attenuation factor (e.g., 20 micrometers). For secreted ligands, a larger distance attenuation factor can be set. For example, it can be set as the second attenuation coefficient (e.g., 100 micrometers). The specific values ​​of the first and second attenuation coefficients are not limited.

[0099] In step S303 of some embodiments, a cell pair may simultaneously express multiple different receptor-ligand pairs, meaning that the sending and receiving cells in the cell pair may exchange information via multiple different molecular signaling pathways. Therefore, for each receptor-ligand pair in the cell pair, the calculation processes of steps S301 and S302 can be performed independently to obtain multiple actual communication strengths. These multiple actual communication strengths are then aggregated, such as by summation or averaging, to obtain the communication strength of the cell to be analyzed (e.g., cell i), thereby summing the dispersed communication signals of the cell to be analyzed into a unified comprehensive value. This comprehensive value (i.e., communication strength) can be used to quantitatively describe the overall functional interaction activity of the cell to be analyzed within its respective candidate functional unit.

[0100] Reference Figure 4 In some embodiments, the communication strength of the cell to be analyzed includes the total communication strength, and step S303 may include, but is not limited to, steps S401 to S403.

[0101] Step S401: Determine the actual communication strength of the cell pair under the ligand-receptor pair, which represents the transmission direction, as the transmission direction communication strength, and determine the actual communication strength of the cell pair under the ligand-receptor pair, which represents the reception direction, as the reception direction communication strength. Step S402: Traverse each cell pair containing the cell to be analyzed, and obtain the total transmit direction communication strength and / or total receive direction communication strength of the cell pair under all receptor pairs. Step S403: Summarize the total communication strength of all cell pairs corresponding to the cell to be analyzed in all transmission directions and / or all reception directions to obtain the total communication strength of the cell to be analyzed.

[0102] In step S401 of some embodiments, the transmitting direction can refer to the signal transmission direction with the cell to be analyzed as the signal transmitter and the other cell in the cell pair as the signal receiver. In the transmitting direction, the expression level of the ligand in the cell to be analyzed and the expression level of the receptor in the other cell jointly determine the calculation of the communication strength. Correspondingly, the receiving direction can refer to the signal transmission direction with the other cell in the cell pair as the signal transmitter and the cell to be analyzed as the signal receiver. In the receiving direction, the expression level of the ligand in the other cell and the expression level of the receptor in the cell to be analyzed jointly determine the calculation of the communication strength. It is understood that for the same ligand-receptor pair of the same cell pair, since the positions of the ligand-expressing cells and receptor-expressing cells involved in the transmitting and receiving directions are interchanged, the actual communication strength values ​​in the two directions are not necessarily equal. Assuming the cell to be analyzed is cell i, then... This indicates the expression level of ligand L in the sending cells. This represents the expression level of receptor R in the receiving cell. Thus, the actual communication strength calculated using formulas 3 and 4 is the communication strength in the transmitting direction. Similarly, the expression level of receptor R in cell i to be analyzed can be determined from the gene expression matrix of the target slice spatial transcriptome data. and the expression level of ligand L in cell j Based on expression levels and expression level By using the same calculation method as formulas 3 and 4, the communication strength in the receiving direction can be calculated. The distinction between the transmitting and receiving directions accurately reflects the directionality of signal transmission between cells.

[0103] In step S402 of some embodiments, the cell to be analyzed can form multiple cell pairs with other cells in the candidate functional units, and in each cell pair, multiple different receptor pairs may be expressed simultaneously. Therefore, for each cell pair, based on the method of step S401, the transmit direction communication strength and / or receive direction communication strength of the cell pair under each receptor pair is determined. Based on all receptor pairs present in the cell pair, the total transmit direction communication strength and / or total receive direction communication strength of the cell pair can be obtained.

[0104] In step S403 of some embodiments, the total communication strength of all transmitting directions and / or all receiving directions of all cell pairs corresponding to the cell to be analyzed is aggregated to obtain the total communication strength of the cell to be analyzed. The aggregation process may refer to merging the communication strength values ​​related to the same cell to be analyzed, distributed across multiple cell pairs and multiple receptor pairs, according to preset mathematical rules, in order to condense the signal interaction information of the cell to be analyzed in multidimensional space into a comprehensive numerical value with biological interpretability.

[0105] Total communication strength refers to a single quantitative indicator obtained after aggregation processing, used to quantitatively describe the overall functional interaction activity of the cell under analysis within its candidate functional unit in all possible directions (including the direction in which the cell under analysis sends signals to other cells, and the direction in which the cell under analysis receives signals from other cells). The higher the value of the total communication strength, the denser the signal interaction between the cell under analysis and other cells within the candidate functional unit, and the deeper the functional embedding.

[0106] In practice, various methods can be used for data aggregation and processing: The first feasible approach is to sum the total communication strength in all transmission directions and the total communication strength in all reception directions for all cell pairs corresponding to cell i under all ligand pairs.

[0107] The second feasible approach is to independently summarize the communication strength in all transmitting and receiving directions of the cell i to be analyzed, and output these two independently summarized values ​​as the total communication strength in the transmitting and receiving directions of the cell i to be analyzed. In the second implementation, the total communication strength in the transmitting and receiving directions can be used as independent analysis indicators, or they can be further combined into a unified total communication strength (e.g., the sum of the two or their weighted average).

[0108] A third possible implementation is to limit the aggregation processing to a single direction, either the transmitting or receiving direction. This means aggregating only the communication strength in the transmitting direction while ignoring the communication strength in the receiving direction, or vice versa. The specific direction chosen depends on the specific focus of the downstream analysis. For example, when studying the signal output capability of the cells being analyzed, the transmitting direction communication strength is aggregated. When studying the signal sensing capability of the cells being analyzed, the receiving direction communication strength is aggregated. Regarding the specific algorithm implementation for aggregation processing, besides using summation, averaging (e.g., normalizing by dividing by the number of cell pairs or the total number of ligand-receptor pairs), weighted summation (assigning different weight coefficients based on the biological importance of different ligand pairs or the spatial distance between cell pairs), etc., can also be used as needed. No limitation is imposed on this approach.

[0109] This application's embodiments clearly distinguish the transmitting and receiving directions of signal transmission within the same cell pair and the same receptor pair, avoiding information loss caused by indiscriminately mixing and summarizing bidirectional signals, thus providing a data foundation for distinguishing the functional roles of cells. Secondly, by flexibly and controllably summarizing the communication strengths of all transmitting and receiving directions obtained through the traversal (including various optional methods such as unified summation, separate summarization, and single-direction summarization), it is possible to output total communication strengths in different dimensions according to the analytical focus of different biological problems. This can be used to comprehensively evaluate the overall functional embedding degree of the cell under analysis within its functional unit, and also to specifically evaluate the signal emission or signal sensing capabilities of the cell under analysis, greatly improving the application flexibility and scenario adaptability of the method.

[0110] Reference Figure 5 In some embodiments, the communication strength of the cells to be analyzed includes significant communication strength, and step S303 may include, but is not limited to, steps S501 to S505.

[0111] Step S501: The expression levels of the ligand-receptor pair in each cell of the candidate functional unit are randomly permuted multiple times, and after each random permutation, the random communication strength of the cell pair under the ligand-receptor pair is calculated based on the expression levels of the ligand-receptor pair after the random permutation. Step S502: Count the number of times that the random communication strength is not less than the actual communication strength in multiple random permutations. Step S503: Calculate the significant communication data of the cell pair under the ligand-receptor pair based on the target number and the total number of permutations of multiple random permutations. If the significant communication data is less than the preset significant threshold, it is determined that there is a significant communication event of the cell pair under the ligand-receptor pair. Step S504: Traverse each cell pair containing the cell to be analyzed and obtain the total actual communication strength of the cell pair that is determined to have significant communication events under all receptor-transfer pairs. Step S505: Summarize all actual communication strengths of all cell pairs corresponding to the cell to be analyzed to obtain the significant communication strength of the cell to be analyzed.

[0112] In step S501 of some embodiments, random permutation is a statistical simulation operation based on the principle of resampling. Specifically, it involves randomly shuffling and redistributing the expression levels of all cells within a candidate functional unit on a certain ligand gene, and simultaneously, independently and randomly shuffling and redistributing the expression levels of all cells within the same candidate functional unit on the corresponding receptor gene. This disrupts the potential biological correlation between ligand and receptor expression levels in the original data (i.e., the gene expression matrix of the target slice spatial transcriptome data), constructing a virtual data state where the ligand-receptor expression association is completely randomized. This random permutation operation can be independently repeated N times, where N is a pre-set positive integer (N=1000) to ensure sufficient statistical stability of the constructed random distribution. After each random permutation operation, the random communication strength of the cell pairs is determined. Random communication strength can refer to the communication strength value calculated after each random permutation, based on the randomly shuffled ligand and receptor expression levels, using the same calculation method as the actual communication strength (i.e., Formula 4). Random communication strength represents the level of communication strength that can be observed due to random factors, assuming the null hypothesis (i.e., there is no real ligand-receptor mediated signal transduction between cell pairs) holds. By performing N independent random permutations, a null hypothesis distribution consisting of N random communication strength values ​​can be constructed for each cell pair and each ligand-receptor pair.

[0113] In step S502 of some embodiments, the number of times the random communication strength is greater than or equal to the actual communication strength in multiple random permutations is counted. The target number reflects the relative position of the actual communication strength in the random null hypothesis distribution. If the actual communication strength is much higher than the random communication strength obtained by the vast majority of random permutations, the target number will approach zero. That is, the value of the target number is an integer between 0 and N. The smaller the target number, the more extreme the tail position of the actual communication strength in the random distribution, meaning that the actual communication strength is more difficult to explain by random factors.

[0114] In step S503 of some embodiments, the significant communication data of the cell pair under a certain ligand-receptor pair is calculated based on the target number and the total number of random permutations N. The significant communication data can refer to a quantitative indicator used to assess whether the actual communication strength is statistically significant relative to a random distribution, such as using the P-value. The P-value characterizes the probability that the random communication strength reaches or exceeds the actual communication strength when the null hypothesis is true. The smaller the P-value, the more difficult it is for the actual communication strength to be explained by random factors, and the more statistically significant the communication of the cell pair under that ligand-receptor pair is. Specifically, in the embodiments of this application, it is defined as follows: Formula 5.

[0115] P-value = (n_random + 1) / (N + 1) (Formula 5). Here, n_random represents the number of times the random communication strength is greater than or equal to the actual communication strength (i.e., the target number). Furthermore, various tests (Benjamini-Hochberg FDR correction) can be applied to the P-value to obtain the corrected P-value (padj).

[0116] The corrected P-value is compared with a preset significance threshold, and the presence of a significant communication event between the cell pair and the ligand-receptor pair is determined based on the comparison result. The significance threshold can be a pre-set critical probability value used to determine whether a communication event is statistically significant, such as 0.05, without limitation. When the corrected P-value is less than the significance threshold (e.g., padj < 0.05), it indicates that the probability of observing the current actual communication strength or stronger communication strength under the random null hypothesis is less than 5%. Therefore, there is sufficient statistical reason to reject the null hypothesis, and the cell pair is determined to have a real signal transduction relationship between the ligand-receptor pair, i.e., a significant communication event exists. Conversely, it indicates that no significant communication event exists. A significant communication event can be defined as a biological event that, after statistical testing, is determined to be significant and reflects the existence of a real ligand-receptor mediated signal transduction between the cell pair.

[0117] In step S504 of some embodiments, the cell to be analyzed can form multiple cell pairs with other cells in the candidate functional units, and in each cell pair, multiple different receptor pairs may be expressed simultaneously. Therefore, for each cell pair, based on the methods of steps S501 to S503, the actual communication strength (e.g., referred to as significant actual communication strength) of the cell pair when it is determined as a significant communication event under each receptor pair is determined. Based on all receptor pairs present in the cell pair, the total actual communication strength of the cell pair can be obtained. That is, the total actual communication strength refers to the actual communication strength of only the cell pair determined as a significant communication event and all receptor pairs, excluding all non-significant communication events that failed the significance test.

[0118] In step S505 of some embodiments, the total actual communication intensities of all cell pairs corresponding to all cells to be analyzed are aggregated to obtain the significant communication intensity of the cell to be analyzed. That is, the aggregation process here can refer to the process of merging the significant actual communication intensities related to the same cell to be analyzed, distributed across multiple cell pairs and multiple receptor pairs, according to a preset rule. Significant communication intensity can be used as a single quantitative indicator to quantitatively describe the overall activity of the cell to be analyzed within its candidate functional unit, where functional interactions occur solely through statistically significant signaling pathways. Significant communication intensity has a higher signal-to-noise ratio and biological specificity than total communication intensity. In specific implementations, the aggregation process can be implemented by averaging (e.g., dividing all significant actual communication intensities by the number of significant communication events to obtain the average significant communication intensity), weighted summing (assigning differentiated weights based on the biological importance of different receptor pairs or the spatial distance between different cell pairs), or finding the maximum value, etc., without limitation. Furthermore, similar to the method of determining the total communication strength by distinguishing directions, before summarizing the significant actual communication strength, it can be further decomposed directionally: significant communication events of the cell being analyzed as a sender and significant communication events of the cell being analyzed as a receiver are summarized independently, obtaining the significant actual communication strength of the cell being analyzed in the sending direction and the significant actual communication strength in the receiving direction, thus preserving the flow information of cell signal transmission in the dimension of significant communication. Through the above summarization process, the final output significant communication strength of the cell being analyzed will serve as the core quantitative basis for subsequent functional unit quality control and cell filtering.

[0119] This application embodiment, by independently and repeatedly randomly shuffling the expression levels of ligands and receptors, completely disrupts the expression coupling relationship between ligands and receptors while preserving the marginal distribution characteristics of the original data. This constructs a null hypothesis distribution that accurately reflects the level of random background noise, providing a rigorous scientific benchmark for the statistical testing of communication events and avoiding model biases that may arise from using parameterized hypothetical distributions (such as normal distributions). Secondly, by calculating empirical p-values ​​and introducing a Benjamini-Hochberg false discovery rate multiple test correction mechanism, this application embodiment can strictly control the false positive rate in large-scale parallel testing scenarios (involving tens of thousands of cell-ligand-receptor pair combinations), effectively distinguishing real biological signals from random background noise, and ensuring that the identified significant communication events have high statistical reliability and repeatability.

[0120] Understandably, in addition to using random permutations and Benjamini-Hochberg error discovery rate multiple test correction methods to assess the statistical significance of communication events, Bootstrap resampling, Bayesian inference, or machine learning classifiers (such as random forests and SVMs) can also be used for permutation tests, and there are no restrictions on this.

[0121] In some embodiments, "traversing each cell pair containing the cell to be analyzed" in steps S402 and S504 may include the following steps: For each cell pair containing the cell to be analyzed, if the spatial distance between the cell pairs is greater than a preset communication distance threshold, the cell pairs are filtered out in order to traverse the remaining cell pairs.

[0122] In this embodiment, before traversing all cell pairs corresponding to the cell to be analyzed, all cell pairs can be screened. Specifically, for each cell pair corresponding to the cell to be analyzed, the spatial distance between the cell to be analyzed and the other cell in the cell pair can first be obtained. Then, this spatial distance is compared with a preset communication distance threshold. If the spatial distance is greater than the communication distance threshold, it indicates that the two cells in the cell pair are too far apart in space, exceeding the physiological range allowed for ligand molecules to diffuse through the interstitial space and effectively activate the target cell receptor. Even if there is a ligand-receptor molecular expression match between the two, their signal transduction will be difficult to produce an effective biological effect in the actual tissue environment due to excessive attenuation. Therefore, the cell pair is filtered out, that is, no subsequent calculations related to communication strength are performed based on the cell pair. If the spatial distance is less than or equal to the communication distance threshold, it indicates that the two cells in the cell pair are within the spatial range that allows for effective signal transduction physiologically, and it is retained as a valid cell pair. Finally, the communication strength of the cell to be analyzed is calculated only based on all the retained valid cell pairs. Among them, the communication distance threshold is a pre-set numerical boundary with spatial scale meaning, and there is no specific limitation on the value of the address book distance threshold.

[0123] The method for screening cell pairs in this application eliminates those cell pairs that are too far apart in space and have no biological significance even if the communication strength is calculated. This ensures that all cell pairs used for the final communication strength summary are valid data with potential interaction rationality on a spatial scale, thereby significantly improving the biological signal-to-noise ratio of the communication strength index at the data level.

[0124] In step S104 of some embodiments, the cells to be analyzed within the candidate functional units are filtered based on the communication strength (total communication strength and / or significant communication strength) of the cells to be analyzed obtained in step S103. Specifically, if the communication strength of a cell to be analyzed is higher than or equal to a preset communication strength threshold (the value is not limited), it is determined that there is a sufficiently strong functional cooperation signal between the cell to be analyzed and other cells in the candidate functional unit, and the cell to be analyzed is retained. Conversely, if the communication strength of the cell to be analyzed is lower than the communication strength threshold, it is determined to be a weakly communicating cell or an isolated cell, and it is removed from the current candidate functional units. Through the above filtering operation, false positive cells that are initially classified into candidate functional units by the model due to spatial proximity but lack substantial molecular interactions with other cells in the unit in terms of biological function can be eliminated. After performing the above filtering operation on each cell to be analyzed within the candidate functional unit, the target functional unit can be obtained. The target functional unit refers to the functional unit that is finally confirmed after dual verification of preliminary spatial distance prediction and cell communication strength quality assessment. All cells within it meet the preset standards in both spatial proximity and functional synergy, thus having a high degree of biological reliability.

[0125] The final output of this application embodiment may include the target functional unit and the following data: Cell-level communication strength matrix: This describes the total communication strength between cell pairs, where each element represents the sum of the actual communication strengths of all receptor-transfer pairs from cell i to cell j. Similarly, a significant communication strength matrix can also be output, where each element represents the sum of the significant actual communication strengths of all receptor-transfer pairs from cell i to cell j.

[0126] Raw communication data: Calculation results used to describe all cell pairs and ligand-receptor pairs, including basic communication strength, actual communication strength after distance decay, distance d(i,j), P-value, etc., which have not yet been filtered by significance.

[0127] Significant communication data: used to describe significant communication events that reach the significance threshold after random permutation and FDR correction, including significant actual communication strength and corresponding cell pairs and receptor pairs.

[0128] Receptor-ligand pair statistics: Count the number of occurrences of each receptor-ligand pair in the target functional unit, the number of significant communication events, the total communication strength (or average communication strength), and the number of cell pairs involved.

[0129] Functional unit statistics: Statistics on the number of cells, cell type composition, total communication strength, significant communication strength, number of significant communication events, and number of cells retained or removed in each candidate or target functional unit.

[0130] It is understandable that cell communication analysis involves a large number of cell pairs and receptor-ligand pair combinations, resulting in a huge computational burden. To address this issue, a parallel computing strategy can be employed in the embodiments of this application, which involves dividing candidate functional units into batches and using multi-core parallel processing to simultaneously handle the communication analysis of multiple candidate functional units.

[0131] The functional unit quality control method provided in this application significantly improves the spatial resolution and cell analysis accuracy of functional units through single-cell resolution spatial transcriptomics technology, enabling precise revelation of cell spatial location and subtle changes. Compared with traditional methods, this application reduces reliance on samples, avoids complex immunohistochemical staining and microscopic operations, and establishes an accurate functional unit prediction model by combining Euclidean distance algorithms and machine learning methods. Furthermore, this application is the first to integrate cell communication analysis into the functional unit prediction process, innovatively introducing spatial distance as a regulatory factor for communication strength. Distance attenuation accurately reflects the actual communication efficiency between cells, overcoming the shortcomings of traditional methods that ignore spatial information. Further, this application employs a statistical method based on random permutation to calculate the P-value for each cell communication event and performs multiple validation corrections, effectively distinguishing between real biological signals and random background noise, significantly improving the reliability and specificity of the analysis. Furthermore, this application innovatively uses cell communication strength as a quantitative indicator for functional unit quality control. By screening high-communication cells and removing isolated cells, it ensures that the predicted functional units are not only spatially structurally complete but also biologically significant in functional synergy. The parallel processing strategy significantly improves computational efficiency, enabling the method provided in this application to process large-scale whole-tissue slice data and complete the analysis within a reasonable time. In summary, this application enhances the ability to analyze the dynamic changes of functional units, accurately predicting their changes under different physiological or pathological states, providing effective technical support for the early diagnosis and personalized treatment planning of neurological diseases (such as stroke and Alzheimer's disease). Through automated machine learning analysis methods, the prediction process is more efficient and accurate, lowering the technical threshold and cost, greatly facilitating scientific research and clinical applications, and possessing broad application prospects.

[0132] In a specific embodiment, taking the neurovascular unit in a human brain hippocampus sample as an example, the functional unit quality control method provided in this application includes the following steps: 1. Data preparation and quality control: First, the total UMI counts for each DNB point's corresponding spatial coordinates are aggregated to generate a spatial density matrix, which is then converted into an image where grayscale intensity reflects the number of UMIs. Next, cell segmentation is performed using an ESPANet-based model, followed by post-processing using the watershed algorithm to generate a cell-gene matrix (i.e., gene expression matrix), used to create Seurat RDS objects for cellbins. Subsequently, the nFeatures, nCounts, and percent.mt levels of cells in each chip are calculated. The quality control standards are as follows: the average number of genes per cell (i.e., nFeatures) should be greater than 100, and the average proportion of mitochondrial genes per cell (i.e., percent.MT) should be less than 20%. Chips (i.e. slices) meeting these standards are considered of acceptable quality and can be used for predictive analysis of neurovascular units.

[0133] 2. Cell type annotation: Cells in the qualified microarray are annotated. Specifically, the Spatial-ID algorithm can be used, which combines common genes from snRNA-seq and Stereo-seq data for cell type prediction. Cell type probabilities are optimized using spatial neighborhood information and gene expression profiles through a Graph Convolutional Network (GCN) and self-supervised learning. Ultimately, the cell type of each cell in the spatial microarray is determined based on the highest probability generated by the GCN. Figure 6 The diagram shown illustrates the results of cell type annotation for chip one and chip two.

[0134] 3. Statistical analysis of cell distance characteristics: Based on the immunohistochemical staining results of adjacent slices from the spatial microarray, the spatial distances between different cell types were identified and statistically analyzed. These cell types include vascular cells (endothelial cells and pericytes), astrocytes, and neurons (inhibitory neurons and excitatory neurons). By analyzing the relative spatial distribution of these cells, the spatial distances between vascular cells, as well as the spatial distances between astrocytes, neurons, and vascular cells, can be specifically identified. This process contributes to a deeper understanding of the spatial structure and functional relationships of neurovascular units, providing crucial distance characteristic data for the accurate prediction and analysis of neurovascular units. Figure 7 The image shown is an immunohistochemical fluorescence image of the hippocampus region provided in an embodiment of this application. Figure 7 The image shows the spatial distribution of neurovascular units in immunohistochemically stained sections of chip one and chip two. Figure 7 The circled area in the middle represents a functional unit of a neurovascular unit, artificially labeled. Astro represents an astrocyte, Neuron represents a neuron, and DAPI represents a fluorescent dye in the cell nucleus.

[0135] 4. Establish a distance model for neurovascular units: Using the Euclidean distance algorithm, a spatial relationship model between cells within a neurovascular unit can be established, quantifying the relative distances between cells and their spatial distribution characteristics. This process first clarifies the interrelationships between cells based on the definition of a neurovascular unit, combined with the cell type characteristics and their spatial location. For example, the distance differences between vascular endothelial cells, astrocytes, and neurons can reflect the functional state of the blood-brain barrier and the integrity of the neurovascular unit. Building upon this, machine learning methods can be further used to construct predictive models. 1) Construct the distance matrix: Construct a distance matrix D using the Euclidean distance between the points, such as .

[0136] 2) Define the central cell set (vascular cells): C={j∣j∈cells, celltype(j)∈{Central cell}}.

[0137] 3) Establish a predictive model for functional units with spatial structure: , Here, Astro represents astrocytes, Endo represents endothelial cells, and Neuron represents neurons.

[0138] 5. Cell communication analysis and neurovascular unit quality assessment: After preliminary prediction of neurovascular units based on trained functional unit predictions, cell communication analysis can be further performed on the predicted neurovascular units to assess their integrity and functional status. 1) Preparation of ligand-receptor database: Preparation of ligand-receptor database: The CellChat database can be used as a reference for ligand-receptor interactions. The database was preprocessed, with gene names matched and mapped to spatial transcriptome data. Ultimately, approximately 773 high-confidence ligand-receptor pairs within neurovascular units were extracted from the database for subsequent analysis.

[0139] 2) Communication strength calculation based on spatial distance: For each predicted neurovascular unit, the Euclidean distance between all cell pairs within the neurovascular unit can be calculated. A maximum interaction distance of 100 micrometers is set; cell pairs exceeding this distance are disregarded for communication. Then, the basic communication strength of the cell pairs can be calculated using a product method, and the effect of spatial distance is considered using an exponential decay function to calculate the actual communication strength. Here, the decay coefficient is set to 100 micrometers, representing the effective diffusion distance of the ligand.

[0140] 3) Statistical test for the significance of communication: To assess the statistical significance of each cell communication event, a permutation test can be performed on each ligand-receptor pair. The specific procedure is as follows: 1000 random permutations are performed, each time randomly sampling from the ligand and receptor expression distributions, and the random communication strength is calculated. The position of the observed communication strength (i.e., the actual communication strength) within the random distribution is calculated to obtain an empirical p-value. The p-value is then corrected using Benjamini-Hochberg FDR to obtain the corrected p-value (padj). Furthermore, significant cell communication events are screened based on a significance threshold (padj < 0.05). In the analysis of chip one, 59065 potential ligand-receptor communication pairs were detected. After statistical significance testing and multiple testing correction, 17553 significant cell communication events (approximately 29.7%) were identified, effectively distinguishing real biological signals from random background noise, providing a reliable quantitative basis for subsequent functional unit quality control.

[0141] 4) Functional unit quality control: Based on cell communication analysis results, a communication strength threshold of 0.2 was set, and the predicted neurovascular units were quality-screened according to this threshold. In chip one, approximately 2300 neurovascular units, containing 17650 cells, were predicted before quality control. After quality control, approximately 6385 cells were retained, distributed across approximately 1446 neurovascular units. The 11016 removed cells were mainly isolated or peripheral cells with weak communication to the central cells of the units. After quality control, the average communication strength of cells within the neurovascular units was significantly improved, and the proportion of significant communication events increased.

[0142] Furthermore, the neurovascular units can be divided into N batches, each containing 100 units, and processed in parallel on a server equipped with a multi-core CPU. Using 32 cores for parallel computation, a complete cell communication analysis (including permutation tests) on chip one takes approximately 0.5 hours. If single-core processing were used, it would be estimated to take approximately 15 hours. The parallelization strategy improves computational efficiency by approximately 30 times, enabling the method of this embodiment to be applied to large-scale whole-tissue slice analysis.

[0143] 5. Neurovascular Unit Prediction and Analysis: After cell communication analysis and quality control screening, high-quality neurovascular unit prediction results were obtained. These results were then visualized, intuitively presenting the spatial organizational structure, cellular composition, communication network, and functional characteristics of neurovascular units, providing data support for mechanistic research and clinical applications in neurovascular diseases. Figure 8 As shown, the spatial distribution map of the finally determined neurovascular units is presented.

[0144] In a specific embodiment, such as Figure 9 The diagram illustrates the overall workflow for neurovascular unit (NVU) cell communication analysis and quality screening based on spatial transcriptome data, as provided in this application embodiment. Figure 9 The entire data processing chain, from raw data input to high-quality functional unit output, is clearly presented from left to right. Its overall architecture includes the input data column on the left, the algorithm processing column in the middle, and the result output column on the right.

[0145] The left-hand input panel displays the input data source for this embodiment of the application, namely, single-cell resolution spatial transcriptome data. This data contains three core elements: first, a spatial coordinate matrix of cells, recording the precise physical location (e.g., x, y coordinates) of each cell in a tissue slice on a two-dimensional plane; second, a gene expression matrix, recording the quantitative information of transcripts for thousands of genes in each cell; and third, a cell type label, which is a biological identifier assigned to each cell by mapping the spatial transcriptome data to a single-cell RNA sequencing reference map. These three elements together constitute the data foundation for subsequent analysis steps.

[0146] The intermediate algorithm processing section is the core part of this application's embodiments. Using a distance threshold of 100 μm, it details how spatial distance information, ligand-receptor databases, and ligand-receptor expression levels are integrated to calculate communication strength and assess the quality of each candidate neurovascular unit. Specifically, the input data for the intermediate algorithm processing section is candidate neurovascular units predicted by the functional unit prediction model. Then, spatial distance screening (100 μm threshold) and statistical significance screening (through permutation tests and FDR correction) are performed. Further, the distance matrix, ligand expression (L-Exp), and receptor expression (R-Exp) are organically integrated to jointly calculate the spatially resolved ligand-receptor communication strength (LR-intensity).

[0147] The right-hand output panel displays the final results after communication analysis and quality screening, namely the spatial distribution of the selected high-quality neurovascular units. In this output, each high-quality neurovascular unit underwent triple quality assurance through spatial distance pre-screening (cell-to-cell distance ≤ 100 μm), ligand-receptor communication strength calculation, and statistical significance testing (permutation test + FDR correction). Its internal cells not only meet the spatial proximity requirement, but more importantly, significant communication events demonstrate functional synergy between them. In the right-hand output panel, different colors represent different cell types.

[0148] To verify the effectiveness of the communication analysis and quality screening in the embodiments of this application, a scheme that only performs functional unit prediction based on the functional unit prediction model without performing communication analysis and quality screening (i.e., skipping steps S103 and S104 and directly using the model's output in step S102 as the final prediction result) is used as a comparison scheme. The functional unit quality control method provided in the embodiments of this application is compared and verified with the comparison scheme as follows: 1. Accuracy Comparison: On the same hippocampal sample chip dataset, the comparative scheme predicted 2324 neurovascular units, while the present application embodiment, through integrated cell communication analysis and statistical significance testing, ultimately identified 1446 high-quality neurovascular units. The present application embodiment successfully removed 878 false positive units (approximately 37.8% of the prediction results from the comparative scheme). These removed units were mainly spatially adjacent cell combinations lacking effective intercellular communication. The neurovascular units retained after quality control were more spatially compact and rational, and the average communication strength within the units was significantly improved (p<0.001, Wilcoxon test), with a significantly increased proportion of significant communication events. The biological reliability and functional relevance of the prediction results of the present application embodiment are significantly improved, effectively solving the problem of high false positive rates in the comparative scheme.

[0149] 2. Computation Time: For 68,715 cells in chip one, the prediction time for the comparative scheme was approximately 2.5 hours, while the complete cell communication analysis (using 32 cores in parallel) in this embodiment took approximately 0.5 hours. This embodiment reduces the analysis time, but considers the significant improvement in accuracy and the substantial reduction in the false positive rate.

[0150] 3. Scalability: The parallelization strategy of this application embodiment enables the method to efficiently process large-scale whole tissue slice data, greatly improving the computational efficiency and providing technical support for the systematic study of the communication network of functional units within an organization.

[0151] Furthermore, such as Figure 10 The image shown is a comparison and verification diagram of cell communication strength between embodiment (V2) and comparative scheme (V1) of this application. Figure 10It is understood that the cells within the target functional unit identified in the embodiments of this application have stronger ligand-receptor signaling interactions.

[0152] Reference Figure 11 This application also provides a functional unit quality control device, which includes: Cell type annotation unit 1101 is used to acquire spatial transcriptome data of the target slice and annotate the cell types of each cell in the target slice based on the spatial transcriptome data of the target slice to obtain the target cell type of each cell in the target slice. The functional unit prediction unit 1102 is used to call the pre-trained functional unit prediction model to predict candidate functional units with spatial structure in the target slice based on the target cell type of each cell in the target slice and the spatial distance between each cell. The communication strength determination unit 1103 is used to determine the communication strength of the cell to be analyzed for each cell pair containing the cell to be analyzed within the candidate functional unit, based on the communication strength of the cell pair in the spatial transcriptome data of the target slice; the communication strength of the cell pair is calculated by the expression level of the ligand-receptor pair in the cell pair. The cell filtering unit 1104 is used to filter the cells to be analyzed in the candidate functional units according to the communication strength of the cells to be analyzed, and remove the cells whose communication strength does not meet the preset index to obtain the target functional units.

[0153] It is evident that the content of the above-described functional unit quality control method embodiments is applicable to the embodiments of this functional unit quality control device. The specific functions implemented by this functional unit quality control device embodiment are the same as those of the above-described functional unit quality control method embodiments, and the beneficial effects achieved are also the same as those achieved by the above-described functional unit quality control method embodiments.

[0154] Reference Figure 12 , Figure 12 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 1201 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 1202 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1202 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1202 and is called and executed by the processor 1201 to execute the functional unit quality control method of the embodiments of this application. The input / output interface 1203 is used to implement information input and output; The communication interface 1204 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 1205 transmits information between various components of the device (e.g., processor 1201, memory 1202, input / output interface 1203, and communication interface 1204); The processor 1201, memory 1202, input / output interface 1203 and communication interface 1204 are connected to each other within the device via bus 1205.

[0155] This application also provides a computer program product, which includes a computer program. A processor of a computer device reads and executes the computer program, causing the computer device to perform the functional unit quality control method described above.

[0156] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described single-cell functional unit quality control method.

[0157] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0158] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0159] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0160] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0161] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0162] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0163] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0164] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0165] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0166] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0167] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0168] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for quality control of functional units, characterized in that, The method includes: Spatial transcriptome data of the target slice is obtained, and cell type annotation is performed on each cell in the target slice based on the spatial transcriptome data of the target slice to obtain the target cell type of each cell in the target slice; The pre-trained functional unit prediction model is invoked to predict candidate functional units with spatial structure in the target slice based on the target cell type of each cell in the target slice and the spatial distance between each cell. For each cell pair containing the cell to be analyzed within the candidate functional unit, the communication strength of the cell to be analyzed is determined based on the communication strength of the cell pair in the spatial transcriptome data of the target slice; the communication strength of the cell pair is calculated by the expression level of the ligand-receptor pair in the cell pair; Based on the communication strength of the cells to be analyzed, the cells to be analyzed in the candidate functional units are filtered out, and cells whose communication strength does not meet the preset index are removed to obtain the target functional units.

2. The method according to claim 1, characterized in that, The step of determining the communication strength of the cells to be analyzed based on the communication strength of the cell pairs in the spatial transcriptome data of the target slice includes: The basic communication strength of the cell pair under the ligand-receptor pair is calculated based on the expression level of the ligand-receptor pair. Based on a preset distance attenuation coefficient and the spatial distance between the cell pairs, the basic communication strength is subjected to distance attenuation processing to obtain the actual communication strength of the cell pairs under the receptor pair. The actual communication strength of the cell pairs under all receptor pairs is summarized to obtain the communication strength of the cell to be analyzed.

3. The method according to claim 2, characterized in that, The communication strength of the cell to be analyzed includes the total communication strength. The actual communication strength of the cell pair across all receptor pairs is summarized to obtain the communication strength of the cell to be analyzed, including: The actual communication strength of the cell pair under the receptor pair, representing the transmission direction, is determined as the transmission direction communication strength, and the actual communication strength of the cell pair under the receptor pair, representing the reception direction, is determined as the reception direction communication strength. Traverse each cell pair containing the cell to be analyzed, and obtain the total transmit direction communication strength and / or total receive direction communication strength of the cell pair under all receptor pairs; The total communication strength of the cell to be analyzed is obtained by summing up the communication strength of all transmitting directions and / or all receiving directions of all cell pairs corresponding to the cell to be analyzed.

4. The method according to claim 2, characterized in that, The communication strength of the cells to be analyzed includes significant communication strength. The process of summarizing the actual communication strength of the cell pairs across all receptor-transfer pairs to obtain the communication strength of the cells to be analyzed includes: The expression levels of the receptor pair in each cell of the candidate functional unit are randomly permuted multiple times, and after each random permutation, the random communication strength of the cell pair under the receptor pair is calculated based on the expression levels of the receptor pair after the random permutation. The statistics show that the random communication strength is not less than the target number of times the actual communication strength is not less than the target number of times in multiple random permutations. The significant communication data of the cell pair under the ligand-receptor pair is calculated based on the target number and the total number of permutations of multiple random permutations. If the significant communication data is less than the preset significant threshold, it is determined that there is a significant communication event of the cell pair under the ligand-receptor pair. Iterate through each cell pair containing the cell to be analyzed and obtain the total actual communication strength of the cell pair that is determined to have significant communication events under all receptor pairs; The total actual communication strength of all cell pairs corresponding to the cell to be analyzed is summarized to obtain the significant communication strength of the cell to be analyzed.

5. The method according to claim 3 or 4, characterized in that, The traversal includes each cell pair of the cells to be analyzed, including: For each cell pair containing the cell to be analyzed, if the spatial distance between the cell pairs is greater than a preset communication distance threshold, the cell pair is filtered out to iterate through the remaining cell pairs.

6. The method according to claim 3, characterized in that, The method for determining the distance attenuation coefficient includes: Determine the type of ligand in the ligand-receptor pair, and determine the distance attenuation coefficient based on the type of ligand.

7. The method according to claim 6, characterized in that, Determining the distance attenuation coefficient based on the type of the ligand includes: If the ligand is membrane-bound, the distance attenuation coefficient is determined to be the first attenuation coefficient; If the ligand is secreted, the distance attenuation coefficient is determined to be the second attenuation coefficient, wherein the second attenuation coefficient is greater than the first attenuation coefficient.

8. A functional unit quality control device, characterized in that, The device includes: The cell type annotation unit is used to acquire the spatial transcriptome data of the target slice and annotate the cell type of each cell in the target slice based on the spatial transcriptome data of the target slice to obtain the target cell type of each cell in the target slice. The functional unit prediction unit is used to call a pre-trained functional unit prediction model to predict candidate functional units with spatial structure in the target slice based on the target cell type of each cell in the target slice and the spatial distance between each cell. A communication strength determination unit is used to determine the communication strength of the cell to be analyzed for each cell pair containing the cell to be analyzed within the candidate functional unit, based on the communication strength of the cell pair in the spatial transcriptome data of the target slice; the communication strength of the cell pair is calculated by the expression level of the ligand-receptor pair in the cell pair; The cell filtering unit is used to filter the cells to be analyzed in the candidate functional units according to the communication strength of the cells to be analyzed, and remove cells whose communication strength does not meet the preset index to obtain the target functional units.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 7.

Images