Biological system critical state early warning method, device, equipment and storage medium

Abstract

The invention relates to a biological system critical state early warning method and device, equipment and a storage medium. The method comprises the following steps: acquiring gene expression data obtained by sampling a plurality of cells at different moments; respectively constructing a cell specificity directed network of each cell at each moment based on the gene expression data at each moment; extracting a local directed network of each gene from the cell specificity directed network of each cell at each moment; based on the extracted local directed network of each gene, respectively determining local directed network flow entropies of each gene for different cells at each moment; determining a global directed network flow entropy score at each moment according to each local directed network flow entropy at each moment; and determining the occurrence time of the critical state according to the change trend of the global directed network flow entropy score at different moments, and pushing an early warning message based on the occurrence time of the critical state. By adopting the method, the accuracy of critical state early warning can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of biological systems, and particularly to a method, device, computer device, computer-readable storage medium, and computer program product for warning of the critical state of a biological system. Background Art

[0002] Currently, with the progress of multi-omics biological sequencing technology, in related technologies, based on multi-omics biological molecular data, mathematical, statistical, information-theoretic, and machine learning theories and methods are used to study the problem of malignant mutations in biological systems, find biomolecular markers with warning effects, etc., to send warning signals at the critical state before the biological system undergoes a transformation.

[0003] The critical state of a biological system refers to the state in which the system undergoes a drastic change at a specific point, which is called the critical point. When the biological system approaches this critical point, its properties and behaviors will change significantly. For the dynamic development process of complex diseases, detecting its critical point can help warn of the deterioration of the condition, so as to take timely measures to prevent the condition from deteriorating rapidly.

[0004] However, currently, there is a problem of low accuracy in warning of the critical state of a biological system. Summary of the Invention

[0005] Based on this, in view of the above technical problems, it is necessary to provide a method, device, computer device, computer-readable storage medium, and computer program product for warning of the critical state of a biological system, which can improve the accuracy of warning of the critical state of a biological system.

[0006] In a first aspect, the present application provides a method for warning of the critical state of a biological system, including:

[0007] Obtaining gene expression data obtained by sampling multiple cells at different times;

[0008] Based on the gene expression data at each time, respectively constructing a cell-specific directed network for each cell at each time;

[0009] Extracting the local directed network of each gene from the cell-specific directed network of each cell at each time;

[0010] Based on the extracted local directed network of each gene, respectively determining the local directed network flow entropy of each gene for different cells at each time;

[0011] According to the local directed network flow entropy of each gene for different cells at each time, determining the global directed network flow entropy score at each time;

[0012] Determine the occurrence time of the critical state according to the change trend of the global directed network flow entropy score at different times, and push warning messages based on the occurrence time of the critical state.

[0013] In one embodiment, based on the gene expression data at each moment, construct the cell-specific directed network of each cell at each moment, including:

[0014] For each moment, based on the gene expression data at the moment, construct the cell-specific network of each cell at the moment;

[0015] Add directed edges to each cell-specific network to obtain the cell-specific directed network of each cell.

[0016] In one embodiment, based on the gene expression data at the moment, construct the cell-specific network of each cell at the moment, including:

[0017] For each cell, take two genes as a group and perform the following processing to obtain the cell-specific network of the cell:

[0018] Construct a scatter plot respectively according to the expression values of the two genes in different cells, where each point in the scatter plot is a cell;

[0019] Determine the domains of the expression values of the two genes in the cells of the scatter plot respectively;

[0020] Based on the number of cells in the two domains and the number of cells in the intersection region of the two domains, determine the statistical dependence index between the two genes;

[0021] When the statistical dependence index is greater than zero, add an edge between the two genes.

[0022] In one embodiment, add directed edges to each cell-specific network to obtain the cell-specific directed network of each cell, including:

[0023] For each gene in each cell-specific network, evaluate the directivity between the gene and the remaining genes except the gene in the cell-specific network to obtain the directivity evaluation result;

[0024] According to the directivity evaluation result, add directed edges to the cell-specific network to obtain the cell-specific directed network of the cell.

[0025] In one embodiment, based on the local directed network of each gene extracted, determine the local directed network flow entropy of each gene for different cells at each moment, including:

[0026] For each local directed network at each moment, determine the out-degree node probability and in-degree node probability of the central gene of the local directed network;

[0027] Based on the out-degree node probability and the in-degree node probability, determine the local directed network flow entropy of the local directed network at a moment.

[0028] In one embodiment, the gene expression data is obtained based on multiple normal cell samples and target cell samples at different lesion stages. Determining the out-degree node probability of the central gene of the local directed network includes:

[0029] For each first-order out-degree gene of the central gene of the local directed network, determine the correlation weight between the first-order out-degree gene and the central gene under the normal cell samples to obtain the first correlation weight of the first-order out-degree gene, and determine the correlation weight between the first-order out-degree gene and the central gene under the mixed cell samples to obtain the second correlation weight of the first-order out-degree gene. The mixed cell samples include multiple normal cell samples and one target cell sample;

[0030] According to the first correlation weight and the second correlation weight of each first-order out-degree gene, determine the out-degree node probability of the central gene.

[0031] In one embodiment, determining the in-degree node probability of the central gene of the local directed network includes:

[0032] For each first-order in-degree gene of the central gene of the local directed network, determine the correlation weight between the first-order in-degree gene and the central gene under the normal cell samples to obtain the first correlation weight of the first-order in-degree gene, and determine the correlation weight between the first-order in-degree gene and the central gene under the mixed cell samples to obtain the second correlation weight of the first-order in-degree gene;

[0033] According to the first correlation weight and the second correlation weight of each first-order in-degree gene, determine the in-degree node probability of the central gene.

[0034] In one embodiment, according to the local directed network flow entropies at a moment, determining the global directed network flow entropy score at the moment includes:

[0035] Determine the average value of the local directed network flow entropies at the moment;

[0036] Determine the average value of the local directed network flow entropies as the global directed network flow entropy score at the moment.

[0037] In one embodiment, after obtaining the gene expression data sampled from multiple cells at different moments, the method further includes:

[0038] Perform data preprocessing on the gene expression data at each moment. The data preprocessing includes at least one of data deduplication, handling missing values, data integration, and data normalization.

[0039] In a second aspect, the present application also provides a warning device for the critical state of a biological system, including:

[0040] A data acquisition module, configured to acquire gene expression data obtained by sampling multiple cells at different times;

[0041] A network construction module, configured to respectively construct a cell-specific directed network for each cell at each moment based on the gene expression data at each moment;

[0042] A local network extraction module, configured to extract the local directed network of each gene from the cell-specific directed network of each cell at each moment;

[0043] A network flow entropy determination module, configured to respectively determine the local directed network flow entropy of each gene for different cells at each moment based on the extracted local directed network of each gene; and determine the global directed network flow entropy score at each moment according to the local directed network flow entropy of each gene for different cells at each moment;

[0044] A critical state warning module, configured to determine the occurrence time of the critical state according to the change trend of the global directed network flow entropy score at different times, and push a warning message based on the occurrence time of the critical state.

[0045] In a third aspect, the present application also provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps in any one of the embodiments of the above-mentioned biological system critical state warning method are implemented.

[0046] In a fourth aspect, the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps in any one of the embodiments of the above-mentioned biological system critical state warning method are implemented.

[0047] In a fifth aspect, the present application also provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps in any one of the embodiments of the above-mentioned biological system critical state warning method are implemented.

[0048] The above-mentioned biological system critical state warning method, device, computer equipment, computer-readable storage medium and computer program product, by acquiring gene expression data obtained by sampling multiple cells at different times, constructing a cell-specific directed network for each cell based on the gene expression data. On the one hand, compared with the protein-protein interaction network method, the present application creates a network for a single cell, which can convert data from an unstable gene expression form to a stable gene expression association form at the single-cell level, so that the gene association network can be identified at the single-cell resolution level, overcoming the limitation of relying on the protein-protein interaction network; on the other hand, the present application takes into account the directed associations between molecules and constructs a cell-specific directed network, which can explore the intrinsic relationship between complex diseases and biological molecules, thereby facilitating the accurate analysis of the core mechanism of the evolution of complex diseases in biological systems, thereby facilitating the improvement of the accuracy of critical state warning. The local directed network of each gene is extracted from each cell-specific directed network, and the flow entropy of each local directed network is determined. The global directed network flow entropy score at different times is determined based on the flow entropy of each local directed network, the time of occurrence of the critical state is determined, and an early warning is issued. In this way, the network disturbance caused by the sample is quantified by the global directed network flow entropy score, which reduces the impact of noise and improves the robustness and accuracy of the critical state early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 A diagram of an application environment of a biological system critical state early warning method in one embodiment;

[0051] Figure 2 A schematic diagram of a process of a biological system critical state early warning method in one embodiment;

[0052] Figure 3 A schematic diagram of constructing a cell-specific network in one embodiment;

[0053] Figure 4 A schematic diagram of a process of a biological system critical state early warning method in another embodiment;

[0054] Figure 5 is a schematic diagram of a cell-specific directed network in one embodiment;

[0055] Figure 6 is a schematic diagram of a local directed network in one embodiment;

[0056] Figure 7 Schematic flowchart of the biological system critical state warning method in yet another embodiment;

[0057] Figure 8 Schematic diagram for calculating the probability of nodes in a local directed network in one embodiment;

[0058] Figure 9 Schematic diagram for calculating the single-cell network flow entropy in one embodiment;

[0059] Figure 10 Structural block diagram of the biological system critical state warning device in one embodiment;

[0060] Figure 11 Internal structure diagram of a computer device in one embodiment. Specific implementation manners

[0061] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0062] Complex or chronic diseases of biological systems generally refer to abnormal life activities caused by changes in the balance of biological systems due to pathogenic factors such as genetics or environment. During the development of complex diseases, sudden deterioration may occur. Considering that there is a commonality in complex diseases, that is, there is a short state called the "critical state" between the normal state and the disease deterioration state of the biological system. In the normal state, the biological system is stable and undergoes slow changes. At the critical point, the disease characteristics are not obvious, and it is at the limit point of the normal state, which is often difficult to detect, resulting in the missed best treatment opportunity. However, at this time, the biological system is at the limit point of the normal state. If the system is disturbed by the outside world at this time, it will directly transfer from the normal state period to the disease state. Therefore, detecting the critical warning signal of complex diseases in biological systems is beneficial to taking corresponding measures in a timely manner.

[0063] The biological system critical state warning method provided by the embodiments of the present application can be applied to an application environment as shown in Figure 1 wherein, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or can be placed in the cloud or other network servers.

[0064] Specifically, it can be that the operator uploads the gene expression data obtained by sampling multiple cells at different times through the terminal 102 to the server 104, and then sends a critical state detection message to the server 104 through the terminal 102. The server 104 obtains the gene expression data obtained by sampling multiple cells at different times. Secondly, based on the gene expression data at each time, a cell-specific directed network of each cell at each time is constructed respectively, and the local directed network of each gene is extracted from the cell-specific directed network of each cell at each time. Then, based on the extracted local directed networks of each gene, the local directed network flow entropy of each gene for different cells at each time is determined respectively. According to the local directed network flow entropy of each gene for different cells at each time, the global directed network flow entropy score at each time is determined. Finally, according to the change trend of the global directed network flow entropy score at different times, the occurrence time of the critical state is determined, and an early warning message is pushed based on the occurrence time of the critical state.

[0065] Among them, the terminal 102 can be but is not limited to various personal computers, laptop computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. The portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted device can be a virtual reality (VR) device, an augmented reality (AR) device, smart glasses, etc. The server 104 can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0066] In an exemplary embodiment, as Figure 2 shown, a method for warning of the critical state of a biological system is provided. Taking the method applied to the Figure 1 server 104 as an example, it includes the following S100 to S600. Among them:

[0067] S100, obtaining gene expression data obtained by sampling multiple cells at different times.

[0068] Among them, the gene expression data is the abundance data of the gene transcription product mRNA measured in the cells, and includes the expression values of the genes in different cells.

[0069] In practical applications, considering that the changes of cells during the disease development process in biological systems are complex and diverse, critical point detection is performed by obtaining gene expression data sampled from cells at different times. At different times, sampling multiple cells can be adding single cells sampled at different times into a sample containing multiple normal cells, and measuring the gene expression data of the samples sampled at different times through high-throughput sequencing technologies (such as RNA-seq), microarray chips or other methods.

[0070] S200, based on the gene expression data at each time, construct a cell-specific directed network for each cell at each time.

[0071] Among them, the cell-specific directed network is used to analyze the association relationship between diseases and biomolecules.

[0072] In practical applications, for the gene expression data at each time, according to the gene expression data at that time, calculate the correlation coefficient between all gene pairs within each cell. The correlation coefficient can be calculated through the Pearson Correlation Coefficient or the Spearman's Rank Correlation Coefficient, etc., to initially construct an undirected co-expression network. By setting a correlation coefficient threshold, filter out the highly correlated connections, and retain the significant correlation relationships filtered out as network edges. Infer the potential gene regulatory relationships in the undirected co-expression network, and assign directions to the edges in the network to obtain the cell-specific directed network of the cell. Inferring the potential gene regulatory relationships and assigning directions to the edges in the network can be obtained through the inference results of causal inference algorithms. Causal inference algorithms can include but are not limited to GENIE3 (Gene Network Inference with Ensemble of Trees), ARACNe (Algorithm for the Reconstruction of Accurate Cellular Networks), or PIDC (Probabilistic Integration of Differential Co-expression), etc.

[0073] S300, extract the local directed network of each gene from the cell-specific directed networks of each cell at each time.

[0074] In practical applications, for each cell at each moment, with a single gene as the center, the local directed network of each gene is separately extracted from the cell-specific directed network of the cell. The local directed network of each gene contains the genes that interact with this gene and the directed edges between them.

[0075] S400. Based on the local directed networks of each gene extracted, the local directed network flow entropy of each gene for different cells at each moment is determined respectively.

[0076] In practical applications, for the local directed network of each gene of each cell at each moment, the out-degree entropy and in-degree entropy of the gene are determined through the correlation weight of the edges, and the local directed network flow entropy of the local directed network is obtained according to the out-degree entropy and in-degree entropy.

[0077] S500. According to the local directed network flow entropy of each gene for different cells at each moment, the global directed network flow entropy score at each moment is determined.

[0078] In practical applications, for the local directed network flow entropy at each moment, a weight factor is introduced to reflect the importance of each gene in the network, and the global directed network flow entropy score at this moment is obtained by weighted summation. Specifically, the gene expression level weight evaluation criterion and the gene centrality weight evaluation criterion are preset. Exemplarily, considering that highly expressed genes are more important in the network, for the gene expression level weight evaluation criterion, the higher the expression value of the gene, the higher the gene expression level weight of its local directed network flow entropy; considering that genes with more connections are more important, for the gene centrality weight evaluation criterion, the more genes connected in the local directed network of the gene, the higher the gene centrality weight of its local directed network flow entropy. Through the gene expression level weight evaluation criterion and the gene centrality weight evaluation criterion, each local directed network is evaluated to obtain the corresponding gene expression level weight and gene centrality weight, and weight coefficients are assigned to the two weights (the sum of the two weight coefficients is 1) to obtain the comprehensive weight factor. For each moment, according to the local directed network flow entropy at this moment and the corresponding comprehensive weight factor, the global directed network flow entropy score at this moment is obtained by weighted summation.

[0079] S600. According to the change trend of the global directed network flow entropy score at different moments, the occurrence moment of the critical state is determined, and based on the occurrence moment of the critical state, a warning message is pushed.

[0080] In practical applications, the global directed network flow entropy score at each moment reflects the global disturbance caused by the introduction of a single cell compared to a normal cell sample. If the global directed network flow entropy score at a certain moment increases sharply, then this moment is a critical point (the moment when the critical state appears), and the top 5% of the genes in the local directed network flow entropy at this moment are identified as dynamic network biomarkers (DNB). In particular, the global directed network flow entropy characterizes the network fluctuation or collective fluctuation of molecules, rather than the random fluctuation of molecules, and is a key criterion for critical points, thereby providing a reliable critical state early warning signal. It can be based on the global directed network flow entropy score at each moment to determine whether the moment is statistically different from the prior information (such as P-value less than 0.05) to detect the moment when the critical state appears (i.e., the critical point). Specifically, the significance level is pre-set to 0.05, and the significance difference of the global directed network flow entropy score at different moments is evaluated based on the statistical method, and the significance test P-value (P-value) of the global directed network flow entropy score at each moment is obtained. If the significance test P-value is less than the significance level of 0.05, it is determined that the critical state has been entered, and the moment is determined as the critical point. It can also be that the increase amplitude between the global directed network flow entropy scores of two adjacent moments is determined, and when the critical point detection is performed on the global network flow entropy score at each moment, a reference amplitude is set according to the average increase amplitude and standard deviation of the n moments before the moment. If the increase amplitude exceeds a certain multiple of the reference amplitude (such as 1.5 times), it is determined that the critical state has been entered, and the critical point is determined according to the increase amplitude. According to the time of occurrence of the critical state and the dynamic network marker, an early warning message is generated and pushed. The form of pushing the warning message can be pushing the alarm message in the notification bar, displaying the alarm message in the form of a full-screen or half-screen pop-up window, or providing a visual alarm prompt through the flashing of an indicator light, etc. It can be understood that the method of pushing the alarm information can be any one of the aforementioned methods or a combination of any two or more methods, and is not limited here.

[0081] In the above-mentioned critical state early warning method of biological system, by acquiring gene expression data obtained by sampling multiple cells at different times, the cell-specific directed network of each cell is constructed based on the gene expression data. On the one hand, compared with the protein-protein interaction network method, the present application creates a network for each single cell, which can convert the data from unstable gene expression form to stable gene expression association form at the single cell level, so that the gene association network can be identified at the single cell resolution level, overcoming the limitation of relying on protein-protein interaction network; on the other hand, the present application considers the directed association between molecules, constructs a cell-specific directed network, and can mine the intrinsic relationship between complex diseases and biological molecules, which is conducive to accurately analyzing the core mechanism of complex disease evolution of biological systems, thereby helping to improve the accuracy of critical state early warning. Extract the local directed network of each gene from each cell-specific directed network, determine the flow entropy of each local directed network, determine the global directed network flow entropy score at different times according to the flow entropy of each local directed network, determine the time of occurrence of the critical state, and issue an early warning. In this way, the network disturbance caused by the sample is quantified by the global directed network flow entropy score, which reduces the influence of noise and improves the robustness and accuracy of the critical state early warning.

[0082] To analyze the correlation between the disease of the biological system and the change of the biological analysis over time, in an exemplary embodiment, S200 includes S220 to S240. Among them:

[0083] S220, for each moment, based on the gene expression data at the moment, construct a cell-specific network for each cell at the moment.

[0084] In practical applications, the correlation coefficients between all gene pairs in each cell can be calculated using the Pearson Correlation Coefficient based on the gene expression data at each moment, and a correlation coefficient threshold can be set. Network edges can be added between gene pairs that are greater than the correlation coefficient threshold to construct a cell-specific network for each cell.

[0085] S240, adding directed edges to each cell-specific network to obtain a cell-specific directed network for each cell.

[0086] In practical applications, for each cell-specific network, the regulatory relationship between genes with network edges is determined by the GENIE3 algorithm, and directions are assigned to the network edges between genes to obtain a cell-specific directed network.

[0087] In this embodiment, by constructing cell-specific directed networks for cells at different times, it is beneficial to analyze the correlation between diseases and biomolecules, so as to improve the accuracy of detecting critical states and the accuracy of early warning.

[0088] In order to improve the accuracy of critical state early warning, in an exemplary embodiment, S220 includes S222 to S228. Among them:

[0089] In this embodiment, for the gene expression data at one time, it is obtained by adding a single cell sample to a sample containing n normal cells. The added single cell sample is sampled at this time, and a gene expression matrix composed of N(n + 1) cells and M genes is formed. .

[0090] For each cell , taking two genes as a group, perform the following processing to obtain the cell 's cell-specific network :

[0091] S222, respectively construct a scatter plot according to the expression values of the two genes in different cells. Among them, each point in the scatter plot is a cell.

[0092] S224, respectively determine the domains of the expression values of the two genes in the cells of the scatter plot.

[0093] S226, based on the number of cells in the two domains and the number of cells in the intersection region of the two domains, determine the statistical dependence index between the two genes.

[0094] S228, when the statistical dependence index is greater than zero, add an edge between the two genes.

[0095] Among them, the statistical dependence index characterizes the degree of non-linear association between genes and is used to construct the cell-specific network.

[0096] In practical applications, as Figure 3 shown, for the cell , taking any two genes as a group to construct a scatter plot, a total of M(M - 1) / 2 scatter plots are generated.

[0097] Exemplarily, for a pair of genes ( ), construct a scatter plot in a Cartesian coordinate system, where axis and axis respectively correspond to the expression values of gene and gene in N cells. Each point in the scatter plot represents a cell. For the cell , the coordinate of the cell is represented as , which represents in the cell the expression value therein; the coordinates of the cell are represented as , which represents in the cell the expression value.

[0098] After constructing the scatter plot of two genes, as Figure 3 shown, in the scatter plot of the gene pair ( ), for the cell , it can be the domain size determined according to preset parameters and close to the cell . Specifically, the preset parameters are used to limit the number of cells in the domain, and the preset parameters can be 0.1N. Through the given preset parameters, the domains of the expression values of the two genes in the cell are respectively determined, that is, the domain of and the domain of = 0.1N; for the domain of = 0.1N.

[0099] After obtaining the two domains, count the number of cells in the intersection region of the two domains. According to the number of cells in the two domains and the number of cells in the intersection region of the two domains, determine the statistical dependence index between the two genes through the following formula:

[0100]

[0101] If the statistical dependence index is greater than zero, then add an edge between and ; otherwise, do not add an edge.

[0102] In this embodiment, by constructing a cell-specific network for each cell, the data can be converted from an unstable gene expression form to a stable gene expression association form at the single-cell level. At the same time, the gene association network can be identified at the single-cell resolution level, overcoming the limitations of relying on the protein-protein interaction network in the prior art.

[0103] To further analyze the core mechanism of the disease evolution of the biological system, in an exemplary embodiment, as Figure 4 shown, S240 includes S242 to S244. Among them:

[0104] S242, for each gene of each cell-specific network, evaluate the directivity between the gene and the remaining genes except the gene in the cell-specific network, and obtain the directivity evaluation result.

[0105] S244. According to the directivity evaluation result, add a directed edge to the cell-specific network to obtain the cell-specific directed network of the cell.

[0106] Among them, the directivity evaluation result may include the directivity between genes.

[0107] In this embodiment, the MarkRank algorithm is applied to evaluate the directivity between genomes in the cell-specific network by calculating the mutual information of gene pairs. Specifically, following the above embodiment, for the cell , use the following formula to calculate and evaluate the directivity of gene and other genes:

[0108]

[0109] Among them, is the expression value of gene in cell , is the expression value of gene in cell , is the gene set containing gene and having an edge with gene , that is, for any .

[0110] is the number of genes contained in the gene set .

[0111] Judge the directivity between genes by updating the weight of the edge (the statistical dependence index between genes) to obtain the directivity evaluation result. The weight of the edge is defined as:

[0112]

[0113] According to the directivity evaluation result, add a directed edge between genes in the cell-specific network to obtain the cell-specific directed network of the cell, which can be:

[0114] If , it is determined that gene has a positive impact on gene , add a directed edge ( , ) between gene and gene , and the flow direction is , ), and the flow direction is → . In this way, construct a cell-specific directed network for each cell ,like Figure 5 shown.

[0115] In this embodiment, by considering the directed associations between molecules and establishing a single-cell cell-specific directed network, it is possible to explore the intrinsic relationship between complex diseases and biological molecules, which is conducive to identifying potential biomarkers, thereby facilitating accurate analysis of the core mechanisms of complex disease evolution in biological systems, thereby facilitating improving the accuracy of critical state warnings.

[0116] In an exemplary embodiment, S400 includes S420 to S440. Among them:

[0117] S420, for each local directed network at each moment, determining the out-degree node probability and the in-degree node probability of the central gene of the local directed network.

[0118] Among them, the out-degree node probability and the in-degree node probability are used to measure the directionality of information flow in local directed networks to reveal the directionality and strength of regulatory relationships.

[0119] In practical applications, for each cell, each local directed network extracted from the cell-specific directed network ,in is the number of genes, A local directed network. Figure 6 As shown, each local directed network consists of a central gene and First-order out-degree gene , First-order in-degree gene constitute.

[0120] For a locally directed network , including the central gene and first-order neighbor genes, among which out-degree genes and In-degree genes ( = The out-degree node probability of the central gene can be calculated based on the expression values ​​of the central gene and its first-order out-degree genes in the cell, as well as the number of first-order out-degree genes. The in-degree node probability of the central gene can be determined based on the expression values ​​of the central gene and its first-order in-degree genes in the cell, as well as the number of first-order in-degree genes.

[0121] S440, determining the local directed network flow entropy of the local directed network at the time instant based on the out-degree node probability and the in-degree node probability.

[0122] In practical applications, for each local directed network at each moment, the out-degree network flow entropy and in-degree network flow entropy of the local directed network are respectively determined according to the out-degree node probability and in-degree node probability of the local directed network, and the local directed network flow entropy of the local directed network is determined according to the out-degree network flow entropy and in-degree network flow entropy.

[0123] In this embodiment, by determining the out-degree node probability and in-degree node probability of the central gene of the local directed network, the local directed network flow entropy is determined to measure the uncertainty of information transmission in the network, which is beneficial to identifying the key regulatory relationships that change significantly under disease states, thereby improving the accuracy of detecting critical points.

[0124] The disease development of biological systems goes through different pathological stages. For a normal cell, the transformation from a normal cell to a cancer cell will go through a long multi-stage process, such as from inflammation, hyperplasia, dysplasia, carcinoma in situ to invasive carcinoma stage. To improve the accuracy of critical state early warning, in an exemplary embodiment, gene expression data is obtained based on multiple normal cell samples and target cell samples at different pathological stages, such as Figure 7 shown, determining the out-degree node probability of the central gene of the local directed network includes S422 to S424. Among them:

[0125] S422, for each first-order out-degree gene of the central gene of the local directed network, determine the correlation weight between the first-order out-degree gene and the central gene under normal cell samples to obtain the first correlation weight of the first-order out-degree gene, and determine the correlation weight between the first-order out-degree gene and the central gene under mixed cell samples to obtain the second correlation weight of the first-order out-degree gene. The mixed cell samples include multiple normal cell samples and one target cell sample.

[0126] Among them, the correlation weight is used to characterize the correlation between the central gene and the first-order out-degree gene. The target cell sample is a sample obtained by sampling cells at different moments in the biological system.

[0127] In practical applications, as Figure 8 shown, the correlation weight between the central gene and the first-order out-degree gene under n normal cells is determined by the following formula , and the correlation weight under mixed cell n + 1 :

[0128]

[0129]

[0130] In the formula, are respectively the expression values of gene in cell under.

[0131] S424. Determine the out-degree node probability of the central gene according to the first correlation weight and the second correlation weight of each first-order out-degree gene.

[0132] In practical applications, the out-degree node probability of the central gene is determined by the following formula:

[0133]

[0134] where, =

[0135] In this embodiment, by calculating the correlation weights between the central gene and its first-order out-degree genes under different cells, and determining the out-degree node probability of the central gene according to the correlation weights, it is beneficial to capture the transmission direction of information in the local directed network to reveal the directionality and strength of gene regulatory relationships.

[0136] In an exemplary embodiment, determining the in-degree node probability of the central gene of the local directed network includes S426 to S428. Among them:

[0137] S426. For each first-order in-degree gene of the central gene of the local directed network, determine the correlation weight between the first-order in-degree gene and the central gene under the normal cell sample to obtain the first correlation weight of the first-order in-degree gene, and determine the correlation weight between the first-order in-degree gene and the central gene under the mixed cell sample to obtain the second correlation weight of the first-order in-degree gene.

[0138] where the correlation weight is used to characterize the correlation between the central gene and the first-order in-degree gene.

[0139] In practical applications, the correlation weight between the central gene and the first-order in-degree gene under n normal cells is determined by the following formula , and the correlation weight under the mixed cell n + 1 :

[0140] =

[0141] =

[0142] In the formula, are the expression values of the gene under the cell respectively.

[0143] S428. Determine the in-degree node probability of the central gene according to the first correlation weight and the second correlation weight of each first-order in-degree gene.

[0144] In practical applications, the in-degree node probability of the central gene is determined by the following formula:

[0145] ,

[0146] =

[0147] In this embodiment, by calculating the correlation weight between the central gene and its first-order out-degree genes under different cells, and determining the out-degree node probability of the central gene according to the correlation weight, it is beneficial to capture the transmission direction of information in the local directed network to reveal the directionality and strength of gene regulatory relationships.

[0148] In an exemplary embodiment, the local directed network flow entropy of each gene for different cells is determined at each moment, including:

[0149] For each local directed network, based on the out-degree node probability of the central gene of the local directed network, the out-degree network flow entropy of the local directed network is determined.

[0150] Based on the in-degree node probability of the central gene of the local directed network, the in-degree network flow entropy of the local directed network is determined.

[0151] Based on the out-degree network flow entropy and the in-degree network flow entropy, the local directed network flow entropy is determined.

[0152] In practical applications, following the above embodiment, for the moment , as Figure 9 shown, in the local directed network , the central gene 's directed network flow entropy is defined as follows:

[0153]

[0154] Among them, the out-degree network flow entropy is calculated as follows:

[0155] ,

[0156] With,

[0157]

[0158] Similarly, the in-degree network flow entropy is calculated as follows:

[0159] ,

[0160] With

[0161] =

[0162] wherein is the total number of all cells included at different time points T. and are the variances of the central gene and the neighborhood gene under n + 1 mixed cells and under n normal cells, respectively.

[0163] In this embodiment, by calculating the out-degree and in-degree network flow entropy respectively, the dynamic changes of genes in the network are further analyzed, so as to more accurately capture the flow pattern of information in the network, reveal the directionality and intensity of the regulatory relationship, and contribute to improving the accuracy of detecting the critical state.

[0164] To quantify network perturbation, in an exemplary embodiment, S500 further includes S520 to S540. Wherein:

[0165] S520, determining the average value of each local directed network flow entropy at a moment.

[0166] S540, determining the average value of each local directed network flow entropy as the global directed network flow entropy score at a moment.

[0167] In practical applications, at the moment the global directed network flow entropy is used to capture the early warning signal related to the critical transition. The average value of each local directed network flow entropy at a moment is determined by the following formula, and the average value of each local directed network flow entropy is determined as the global directed network flow entropy score at a moment:

[0168]

[0169] In this embodiment, the network perturbation caused by a single target cell relative to the reference sample is quantified by the global directed network flow entropy, thereby effectively reducing high noise and contributing to improving the accuracy of critical state warning.

[0170] To improve the accuracy of gene expression data analysis, in an exemplary embodiment, after obtaining the gene expression data sampled from multiple cells at different times, the method further includes:

[0171] Performing data preprocessing on the gene expression data at each moment, where the data preprocessing includes at least one of data deduplication, handling missing values, data integration, and data normalization.

[0172] In practical applications, data deduplication of gene expression data can be to identify duplicate gene entries based on gene identifiers and delete the duplicates. Processing missing values in gene expression data can be to detect null values in the gene expression data and delete the null values. If there are different probes corresponding to a gene, the average value of the intensities of the different probes corresponding to the gene is taken as the expression value of the gene. Data integration is performed on the gene expression data at different times, and data normalization is performed on the gene expression data. Data normalization methods include, but are not limited to, total number normalization and RPKM (Reads Per Kilobase of transcript per Million mapped reads), etc.

[0173] In this embodiment, by performing data preprocessing on gene expression data, it is beneficial to improve the accuracy and effectiveness of critical state warning based on gene expression data.

[0174] To make a clearer description of the biological system critical state warning method provided in this application, the following will be described with a specific embodiment. The specific embodiment includes the following steps:

[0175] S1. Obtain gene expression data obtained by sampling multiple cells at different times, and perform data preprocessing on the gene expression data at each time. The data preprocessing includes at least one of data deduplication, processing missing values, data integration, and data normalization.

[0176] S2. For each time, based on the gene expression data at that time, for each cell, taking two genes as a group, perform the following processing to obtain the cell-specific network of the cell. Scatter plots are constructed respectively according to the expression values of the two genes in different cells. Each point in the scatter plot is a cell. The domains of the expression values of the two genes in the cells of the scatter plot are determined respectively. Based on the number of cells in the two domains and the number of cells in the intersection region of the two domains, the statistical dependence index between the two genes is determined. If the statistical dependence index is greater than zero, an edge is added between the two genes.

[0177] S3. For each gene in each cell-specific network, evaluate the directionality between the gene and the remaining genes in the cell-specific network except the gene to obtain a directionality evaluation result. According to the directionality evaluation result, directed edges are added to the cell-specific network to obtain the cell-specific directed network of the cell.

[0178] S4. For each first-order out-degree gene of the central gene of the local directed network, determine the correlation weight between the first-order out-degree gene and the central gene under normal cell samples to obtain the first correlation weight of the first-order out-degree gene, and determine the correlation weight between the first-order out-degree gene and the central gene under the mixed cell samples to obtain the second correlation weight of the first-order out-degree gene. The mixed cell samples include multiple normal cell samples and one target cell sample. Determine the out-degree node probability of the central gene according to the first correlation weight and the second correlation weight of each first-order out-degree gene.

[0179] S5. For each first-order in-degree gene of the central gene of the local directed network, determine the correlation weight between the first-order in-degree gene and the central gene under normal cell samples to obtain the first correlation weight of the first-order in-degree gene, and determine the correlation weight between the first-order in-degree gene and the central gene under the mixed cell samples to obtain the second correlation weight of the first-order in-degree gene. Determine the in-degree node probability of the central gene according to the first correlation weight and the second correlation weight of each first-order in-degree gene.

[0180] S6. Based on the out-degree node probability and the in-degree node probability, determine the local directed network flow entropy of the local directed network at the moment.

[0181] S7. Determine the average value of the local directed network flow entropies at the moment, and determine the average value of the local directed network flow entropies as the global directed network flow entropy score at the moment.

[0182] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least a part of other steps or steps or stages in other steps.

[0183] In an exemplary embodiment, as Figure 10 shown, a biological system critical state warning device 600 is provided, including: a data acquisition module 610, a network construction module 620, a local network extraction module 630, a network flow entropy determination module 640, and a critical state warning module 650, where:

[0184] The data acquisition module 610 is configured to acquire gene expression data obtained by sampling multiple cells at different moments;

[0185] A network construction module 620, configured to construct a cell-specific directed network for each cell at each moment based on the gene expression data at each moment.

[0186] A local network extraction module 630, configured to extract the local directed network of each gene from the cell-specific directed network of each cell at each moment.

[0187] A network flow entropy determination module 640, configured to respectively determine the local directed network flow entropy of each gene for different cells at each moment based on the extracted local directed network of each gene; and determine the global directed network flow entropy score at each moment according to the local directed network flow entropy of each gene for different cells at each moment.

[0188] A critical state warning module 650, configured to determine the occurrence moment of the critical state according to the change trend of the global directed network flow entropy score at different moments, and push a warning message based on the occurrence moment of the critical state.

[0189] In an exemplary embodiment, the network construction module 620 is further configured to, for each moment, construct a cell-specific network for each cell based on the gene expression data at the moment; and add directed edges to each cell-specific network to obtain the cell-specific directed network of each cell.

[0190] In an exemplary embodiment, the network construction module 620 is further configured to, for each cell, take two genes as a group and perform the following processing to obtain the cell-specific network of the cell: respectively construct a scatter plot according to the expression values of the two genes in different cells, where each point in the scatter plot is a cell; respectively determine the domains of the expression values of the two genes in the cells of the scatter plot; determine the statistical dependence index between the two genes based on the number of cells in the two domains and the number of cells in the intersection region of the two domains; and add an edge between the two genes when the statistical dependence index is greater than zero.

[0191] In an exemplary embodiment, the network construction module 620 is further configured to evaluate the directivity between each gene of each cell-specific network and the remaining genes except the gene in the cell-specific network to obtain a directivity evaluation result; and add directed edges to the cell-specific network according to the directivity evaluation result to obtain the cell-specific directed network of the cell.

[0192] In an exemplary embodiment, the network flow entropy determination module 640 is further configured to, for each local directed network at each moment, determine the out-degree node probability and in-degree node probability of the central gene of the local directed network.

[0193] Based on the out-degree node probability and the in-degree node probability, determine the local directed network flow entropy of the local directed network at a moment.

[0194] In an exemplary embodiment, the network flow entropy determination module 640 is further configured to, for each first-order out-degree gene of the central gene of the local directed network, determine the correlation weight between the first-order out-degree gene and the central gene in a normal cell sample to obtain the first correlation weight of the first-order out-degree gene, and determine the correlation weight between the first-order out-degree gene and the central gene in a mixed cell sample to obtain the second correlation weight of the first-order out-degree gene, where the mixed cell sample includes a plurality of normal cell samples and one target cell sample; and determine the out-degree node probability of the central gene according to the first correlation weight and the second correlation weight of each first-order out-degree gene.

[0195] In an exemplary embodiment, the network flow entropy determination module 640 is further configured to, for each first-order in-degree gene of the central gene of the local directed network, determine the correlation weight between the first-order in-degree gene and the central gene in a normal cell sample to obtain the first correlation weight of the first-order in-degree gene, and determine the correlation weight between the first-order in-degree gene and the central gene in a mixed cell sample to obtain the second correlation weight of the first-order in-degree gene; and determine the in-degree node probability of the central gene according to the first correlation weight and the second correlation weight of each first-order in-degree gene.

[0196] In an exemplary embodiment, the network flow entropy determination module 640 is further configured to, for each local directed network, determine the out-degree network flow entropy of the local directed network based on the out-degree node probability of the central gene of the local directed network; determine the in-degree network flow entropy of the local directed network based on the in-degree node probability of the central gene of the local directed network; and determine the local directed network flow entropy based on the out-degree network flow entropy and the in-degree network flow entropy.

[0197] In an exemplary embodiment, the network flow entropy determination module 640 is further configured to determine the average value of the local directed network flow entropies at a moment; and determine the average value of the local directed network flow entropies as the global directed network flow entropy score at the moment.

[0198] In an exemplary embodiment, the biological system critical state warning device 600 further includes a data preprocessing module 660, configured to perform data preprocessing on the gene expression data at each moment, where the data preprocessing includes at least one of data deduplication, missing value processing, data integration, and data normalization.

[0199] Each module in the above biological system critical state warning device 600 can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or stored in a memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0200] In an exemplary embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 11 shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a biological system critical state warning method.

[0201] Those skilled in the art can understand that Figure 11 the structure shown in

[0202] is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0203] In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps in any one of the above embodiments of the biological system critical state warning method.

[0204] In an embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, it implements the steps in any one of the above embodiments of the biological system critical state warning method.

[0205] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data that have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of the relevant data need to comply with the relevant regulations.

[0206] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0207] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this application.

[0208] The above-described embodiments merely represent several implementation manners of this application. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of this application. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several modifications and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of this application shall be subject to the appended claims.

Claims

1. A method for warning of the critical state of a biological system, characterized in that, The method includes: Obtaining gene expression data sampled from multiple cells at different times; Based on the gene expression data at each time, respectively constructing a cell-specific directed network for each cell at each time; Extracting the local directed network of each gene from the cell-specific directed networks of each cell at each time; Based on the extracted local directed networks of each gene, respectively determining the local directed network flow entropy of each gene for different cells at each time; According to the local directed network flow entropy of each gene for different cells at each time, determining the global directed network flow entropy score at each time; According to the change trend of the global directed network flow entropy score at different times, determining the occurrence time of the critical state, and based on the occurrence time of the critical state, pushing a warning message.

2. The method according to claim 1, wherein The constructing, based on the gene expression data at each time, respectively constructing a cell-specific directed network for each cell at each time, includes: For each time, based on the gene expression data at that time, respectively constructing a cell-specific network for each cell at that time; Adding directed edges to each of the cell-specific networks to obtain the cell-specific directed networks of each of the cells.

3. The method according to claim 2, characterized in that, The constructing, based on the gene expression data at that time, respectively constructing a cell-specific network for each cell at that time, includes: For each cell, taking two genes as a group, performing the following processing to obtain the cell-specific network of the cell: Respectively constructing a scatter plot according to the expression values of the two genes in different cells, where each point in the scatter plot is a cell; Respectively determining the domains of the expression values of the two genes in the cells in the scatter plot; Based on the number of cells in the two domains and the number of cells in the intersection region of the two domains, determining the statistical dependence index between the two genes; In the case where the statistical dependence index is greater than zero, adding an edge between the two genes.

4. The method according to claim 3, characterized in that The adding directed edges to each of the cell-specific networks to obtain the cell-specific directed networks of each of the cells, includes: For each gene in each of the cell-specific networks, evaluating the directionality between the gene and the remaining genes in the cell-specific network other than the gene to obtain a directionality evaluation result; According to the directionality evaluation result, adding directed edges to the cell-specific network to obtain the cell-specific directed network of the cell.

5. The method according to claim 4, characterized in that The determining, based on the extracted local directed networks of each gene, respectively determining the local directed network flow entropy of each gene for different cells at each time, includes: For each of the local directed networks at each time, determining the out-degree node probability and in-degree node probability of the central gene of the local directed network; Based on the out-degree node probability and in-degree node probability, determining the local directed network flow entropy of the local directed network at that time.

6. The method according to claim 5, wherein The gene expression data is obtained based on multiple normal cell samples and target cell samples at different lesion stages. Determining the out-degree node probability of the central gene of the local directed network includes: For each first-order out-degree gene of the central gene of the local directed network, determine the correlation weight between the first-order out-degree gene and the central gene under the normal cell samples to obtain the first correlation weight of the first-order out-degree gene, and determine the correlation weight between the first-order out-degree gene and the central gene under the mixed cell samples to obtain the second correlation weight of the first-order out-degree gene, where the mixed cell samples include a plurality of the normal cell samples and one target cell sample; Determine the out-degree node probability of the central gene according to the first correlation weight and the second correlation weight of each first-order out-degree gene.

7. The method according to claim 6, wherein Determine the in-degree node probability of the central gene of the local directed network, including: For each first-order in-degree gene of the central gene of the local directed network, determine the correlation weight between the first-order in-degree gene and the central gene under the normal cell samples to obtain the first correlation weight of the first-order in-degree gene, and determine the correlation weight between the first-order in-degree gene and the central gene under the mixed cell samples to obtain the second correlation weight of the first-order in-degree gene; Determine the in-degree node probability of the central gene according to the first correlation weight and the second correlation weight of each first-order in-degree gene.

8. The method according to claim 5, characterized in that, The determining the global directed network flow entropy score at the moment according to the local directed network flow entropies at the moment includes: Determine the average value of the local directed network flow entropies at the moment; Determine the average value of the local directed network flow entropies as the global directed network flow entropy score at the moment.

9. The method according to any one of claims 1 to 8, characterized in that, After obtaining the gene expression data obtained by sampling multiple cells at different times, the method further includes: Perform data preprocessing on the gene expression data at each moment, where the data preprocessing includes at least one of data deduplication, missing value processing, data integration, and data normalization.

10. A warning device for the critical state of a biological system, characterized in that, The device includes: A data acquisition module, configured to acquire gene expression data obtained by sampling multiple cells at different times; A network construction module, configured to respectively construct cell-specific directed networks of each cell at each moment based on the gene expression data at each moment; A local network extraction module, configured to extract local directed networks of each gene from the cell-specific directed networks of each cell at each moment; A network flow entropy determination module, configured to respectively determine the local directed network flow entropies of each gene for different cells at each moment based on the extracted local directed networks of each gene; and determine the global directed network flow entropy score at each moment according to the local directed network flow entropies of each gene for different cells at each moment; A critical state warning module, configured to determine the occurrence moment of the critical state according to the change trend of the global directed network flow entropy score at different times, and push a warning message based on the occurrence moment of the critical state.

Images