Data processing method for sentinel gene library based on cluster analysis

Abstract

The application belongs to the technical field of data processing, and discloses a sentinel gene library data processing method based on clustering analysis, which comprises the following steps: extracting gene expression data of samples in a sentinel gene library, and constructing a static characteristic space and a dynamic time sequence space; performing preliminary clustering on the static characteristic space to obtain an initial static grouping; switching to the dynamic time sequence space in each static grouping, performing secondary clustering on the samples in the grouping based on time sequence similarity, and refining into a dynamic evolution subgroup; fitting a dynamic trajectory model of gene expression changing with time for each dynamic evolution subgroup, calculating a deviation degree of the sample and a predicted value of the corresponding dynamic trajectory model in the static characteristic space, and marking the sample with a deviation degree exceeding a preset confidence interval as an abnormal sample; and the application further expands the application scene and value of the sentinel gene library in biomedical research.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically, to a method for processing sentinel gene bank data based on cluster analysis. Background Technology

[0002] In biomedical research and clinical applications, sentinel gene banks are core resources for capturing gene expression dynamics and mining disease biomarkers. Their data contains both static information (such as expression levels at a single time point) and dynamic temporal information (such as expression fluctuation trends), and both must be considered during data processing. However, current sentinel gene bank data processing technologies have significant bottlenecks, making it difficult to unlock the full value of the data.

[0003] First, existing technologies tend to focus on static data while neglecting dynamic data, making it difficult to characterize the temporal evolution of gene expression. Most methods rely solely on static data from a single time point or simply piece together time-series information, ignoring dynamic trajectory differences. This leads to the misclassification of samples with similar static characteristics but different evolutionary trends, obscuring the heterogeneity of disease subtypes and key inflection points in disease progression.

[0004] Secondly, cluster analysis often uses a single logic of determining the result through a single clustering, resulting in coarse grouping granularity and difficulty in achieving "static classification and dynamic subdivision". This leads to the masking of heterogeneity within the group and makes it impossible to reveal potential dynamic subgroups.

[0005] Furthermore, the detection of abnormal samples lacks dynamic support, resulting in insufficient accuracy. Traditional methods often rely on fixed thresholds or static statistical indicators, failing to consider dynamic trajectory deviations and lacking model-based confidence interval constraints. This makes them prone to false positives and false negatives, and difficult to effectively identify key abnormal samples related to disease phenotypes or experimental biases.

[0006] Finally, the data processing results are isolated and lack dynamic correlation. Existing methods only provide grouping results or anomaly lists, failing to quantify and visualize trajectory similarities between groups, nor reveal the relationship between anomalous samples and normal subgroups, which is detrimental to researchers' global understanding of gene evolution networks.

[0007] In view of this, the present invention proposes a method for processing sentinel gene bank data based on cluster analysis to solve the above problems. Summary of the Invention

[0008] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a method for processing sentinel gene bank data based on cluster analysis, comprising:

[0009] S1. Extract gene expression data from samples in the sentinel gene bank and construct a static feature space and a dynamic temporal space;

[0010] S2. Perform preliminary clustering on the static feature space to obtain initial static groups; switch to the dynamic time series space within each static group, and perform secondary clustering on the samples within the group based on time series similarity to refine them into dynamic evolution subgroups.

[0011] S3. Fit a dynamic trajectory model of gene expression over time for each dynamic evolution subgroup, calculate the deviation between the sample and the predicted value of the corresponding dynamic trajectory model in the static feature space, and mark the sample with the deviation exceeding the preset confidence interval as an abnormal sample.

[0012] S4. Using dynamic evolution subgroups as core nodes and abnormal samples as independent nodes, calculate the correlation strength between nodes based on trajectory feature similarity, and construct a dynamic map of sentinel genes with correlation strength as edges.

[0013] Specifically, the gene expression data includes transcriptome data and proteome data; transcriptome data refers to the expression level of RNA molecules produced by gene transcription, including mRNA expression level and non-coding RNA expression level; proteome data refers to the expression level of proteins generated by gene transcription products after translation, including protein expression level and protein modification level.

[0014] Specifically, the method for constructing the static feature space and the dynamic temporal space includes:

[0015] Based on gene expression data, the gene expression values ​​of the sample at a preset fixed time point are used as feature inputs. After standardization and feature dimensionality reduction, the static feature vector of the sample is obtained, forming a static feature space that characterizes the molecular state of the sample at a static time section.

[0016] Meanwhile, in the dynamic time series space, the gene expression data of the samples at different continuous time points are arranged in chronological order and organized into a multidimensional time series. The processed time series is mapped into a low-dimensional trajectory vector by principal component analysis, which serves as the coordinate point of the sample in the dynamic time series space and is embedded to form the dynamic time series space.

[0017] Specifically, the method for preliminary clustering of the static feature space includes:

[0018] The static feature vectors of all samples in the static feature space are standardized to make the expression levels of different genes comparable. Based on the standardized static feature vectors, an unsupervised clustering method is used to group the samples to obtain the initial static groups of samples in the static feature space.

[0019] Specifically, the method for performing secondary clustering on samples within a group includes:

[0020] For samples belonging to the same initial static group, gene expression data at different consecutive time points are extracted to form time series trajectories; the similarity between time series samples is measured by the cosine similarity method, and a similarity matrix of samples within the group is constructed based on the similarity measurement results;

[0021] Based on the similarity matrix, a time series clustering method is used to perform secondary clustering of the samples, refining samples in the same initial static group into different dynamic evolution subgroups. Each dynamic evolution subgroup corresponds to a class of samples with similar gene expression evolution trajectories in the time dimension.

[0022] Specifically, the method for obtaining the dynamic trajectory model includes:

[0023] For each dynamic evolution subgroup, gene expression data at continuous time points are extracted; the gene expression data are preprocessed; a preset function space is selected as the fitting form of the dynamic trajectory model, and the samples in the dynamic evolution subgroup are fitted in the function space by minimizing the fitting error. Finally, a dynamic trajectory model that can characterize the trend of gene expression over time is generated for each dynamic evolution subgroup.

[0024] Specifically, the method for calculating the deviation between the sample and the corresponding dynamic trajectory model prediction value includes:

[0025] At any given time point, the gene expression value of the sample in the static feature space is extracted and compared with the predicted value of the dynamic trajectory model to which the sample belongs at the corresponding time point. The difference between the two is calculated as the residual. The absolute value of the residual is normalized according to the standard deviation of the residual within the dynamic evolution subgroup to which the sample belongs at the corresponding time point to obtain the normalized deviation value. The normalized deviation values ​​of all time points are averaged to obtain the overall deviation of the sample over the entire time series, which is then used as the deviation of the sample from the predicted value of the corresponding dynamic trajectory model.

[0026] Specifically, the method for obtaining the abnormal samples includes:

[0027] Statistical analysis was performed on the residuals of all samples within the dynamic evolution subgroup. The residual covariance matrix at that time point was obtained by calculating the covariance of the residuals in each dimension. The gene expression values ​​of the dynamic trajectory model and the samples were parametrically fitted to obtain the summary parameter vector of the dynamic trajectory model and the estimated parameter vector of the samples. Then, the cross-spatial consistency deviation was calculated.

[0028] When the deviation between a sample and the predicted value of the corresponding trajectory model, as well as the deviation from the cross-spatial consistency, both exceed the corresponding preset confidence interval, the sample is determined to be an abnormal sample.

[0029] Specifically, the method for obtaining the association strength between the nodes includes:

[0030] Each dynamic evolution subgroup is used as a core node, and each abnormal sample is used as an independent node. For each node, dynamic trajectory features are extracted in the time series. Between any two nodes, cosine similarity is used to calculate the similarity of dynamic trajectory features, thereby obtaining the correlation strength between each node.

[0031] Specifically, the method for obtaining the sentinel gene dynamic map includes:

[0032] The calculated association strength is used as the weight of the connecting edge between nodes in the sentinel gene dynamic graph. When the association strength between two nodes is greater than or equal to the preset association strength threshold, a connecting edge is established between the two nodes, and the weight of the connecting edge is set to the association strength value between the two nodes. If the association strength between the independent node corresponding to the abnormal sample and any core node is less than the preset association strength threshold, it is kept isolated. All nodes and their connecting edges are combined into a weighted undirected graph to obtain the sentinel gene dynamic graph.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] This invention constructs a static feature space and a dynamic temporal space simultaneously, and combines two-level clustering to achieve "static grouping-dynamic subdivision". It can accurately characterize the evolutionary trajectory of gene expression on the basis of global hierarchical structure, effectively avoiding the misclassification problem of statically similar but dynamically different samples in existing methods, thereby revealing the key subtypes and inflection points of disease progression.

[0035] Secondly, by fitting a dynamic trajectory model within a dynamic evolution subgroup and calculating a dual deviation based on model residuals and cross-spatial consistency, a statistical discrimination mechanism for abnormal samples is established. This mechanism is more objective and robust than the traditional empirical threshold method, and can significantly reduce false positives and false negatives, improve the accuracy of abnormal sample detection, and facilitate researchers in capturing important information related to disease phenotypes, drug responses, or experimental biases.

[0036] Furthermore, a dynamic atlas construction method is introduced, with dynamic evolutionary subgroups as core nodes and anomalous samples as independent nodes. The node associations are quantified through trajectory feature similarity to form a dynamic atlas of sentinel genes with structured expression capabilities. This atlas can not only intuitively display typical evolutionary patterns, but also reveal the branching paths and attribution tendencies of anomalous samples, providing an intuitive tool for a global understanding of gene networks.

[0037] In summary, this invention overcomes the limitations of traditional methods, providing more refined, dynamic, and visualized data processing tools for sentinel gene bank research, thereby enhancing the value of disease mechanism exploration and clinical application. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the data processing method for sentinel gene bank based on cluster analysis according to the present invention;

[0039] Figure 2 This is a schematic diagram of the structure of the sentinel gene bank data processing system based on cluster analysis of the present invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Example 1

[0042] Please see Figure 1 As shown, this embodiment further illustrates the sentinel gene bank data processing method based on cluster analysis proposed in this invention, including:

[0043] In the field of biomedical research and clinical applications, sentinel gene banks serve as core data resources for capturing dynamic changes in gene expression, mining disease diagnostic biomarkers, and elucidating drug mechanisms of action. The accuracy and effectiveness of their data processing directly determine the reliability of subsequent research conclusions. With the rapid development of high-throughput sequencing technology, the number of samples accumulated in sentinel gene banks continues to grow. Gene expression data not only includes static information reflecting the basic biological characteristics of samples (such as gene expression levels at specific time points and differential expression folds), but also covers dynamic temporal information reflecting changes in gene expression with physiological states, disease progression, drug interventions, and other factors (such as the fluctuation trend of gene expression levels at different time points and inflection point characteristics). This places a dual requirement on data processing methods: "balancing the integrity of static features with the regularity of dynamic temporal patterns."

[0044] However, current data processing technologies for sentinel gene banks still have many shortcomings:

[0045] First, existing technologies generally exhibit a bias towards static data while neglecting dynamic data, failing to characterize the temporal evolution of gene expression. Most methods still rely on static data from a single time point, or simply concatenate time-series data from multiple time points into a static vector, or even ignore temporal information altogether. This leads to the misclassification of samples with similar static characteristics but different dynamic pathways. For example, in tumor research, some samples initially show consistent sentinel gene expression, but exhibit increasing or decreasing trends as the disease progresses. Current technologies cannot distinguish between these, masking disease subtype heterogeneity and missing crucial inflection points in disease progression.

[0046] Secondly, cluster analysis, as the core means of grouping sentinel gene bank samples, currently employs a single processing logic of "one-time clustering determines the result," resulting in coarse grouping granularity and difficulty in refining sample heterogeneity. Traditional single-time clustering cannot achieve "homogeneity within groups and heterogeneity between groups" due to individual differences in samples, and it cannot meet the needs of "static classification into major categories and dynamic subdivision into subgroups."

[0047] Third, the accuracy of abnormal sample detection is low and lacks dynamic support. Abnormal samples in sentinel gene banks (such as samples with abnormal fluctuations in gene expression or those that deviate from the overall evolutionary trend) often correspond to key information such as specific disease phenotypes, experimental operation errors, or drug intervention failures, and are the focus of research. However, existing technologies mostly rely on static indicators such as "gene expression levels exceeding the mean ± n standard deviations" or "differential expression fold greater than a fixed threshold" to judge abnormalities, without considering the relative deviation of samples in the dynamic evolution process. Moreover, the judgment of abnormalities lacks statistical model support and relies only on subjective experience thresholds, resulting in a high rate of false positives (normal samples are misjudged as abnormal) and false negatives (abnormal samples are missed).

[0048] Fourth, the data processing results are isolated and lack dynamic correlation. Traditional methods only output group lists or lists of abnormal samples, failing to quantify or visualize the association between groups and samples. They cannot demonstrate the similarity of subgroup trajectories, nor can they clarify the association tendency between abnormal samples and normal subgroups. This is not conducive to researchers' global understanding of gene evolution networks and affects in-depth exploration.

[0049] In view of this, the present invention proposes a method for processing sentinel gene bank data based on cluster analysis, including:

[0050] S1. Extract gene expression data from samples in the sentinel gene bank and construct a static feature space and a dynamic temporal space;

[0051] S2. Perform preliminary clustering on the static feature space to obtain initial static groups; switch to the dynamic time series space within each static group, and perform secondary clustering on the samples within the group based on time series similarity to refine them into dynamic evolution subgroups.

[0052] S3. Fit a dynamic trajectory model of gene expression over time for each dynamic evolution subgroup, calculate the deviation between the sample and the predicted value of the corresponding dynamic trajectory model in the static feature space, and mark the sample with the deviation exceeding the preset confidence interval as an abnormal sample.

[0053] S4. Using dynamic evolution subgroups as core nodes and abnormal samples as independent nodes, calculate the correlation strength between nodes based on trajectory feature similarity, and construct a dynamic map of sentinel genes with correlation strength as edges.

[0054] Gene expression data includes transcriptome data and proteome data. Transcriptome data refers to the expression levels of RNA molecules produced by gene transcription, including mRNA (messenger RNA) and non-coding RNA expression levels. It is the most basic and commonly used type of gene expression data, used to reflect the gene's "transcriptional level" of expression activity (i.e., whether the gene is "activated" and to what extent). Proteome data refers to the expression levels of proteins produced by the translation of gene transcription products, including protein expression levels and protein modification levels. It is the final product reflecting the gene's "translation level" (RNA must be translated into protein to directly participate in physiological / pathological processes).

[0055] It should be noted that gene expression refers to the process by which the genetic information in a gene is transformed into molecules with biological functions through transcription (DNA→RNA) and translation (RNA→protein). Therefore, gene expression data is essentially a quantitative record of key molecules in this process, specifically including two core data types: transcriptome data and proteome data, each corresponding to different detection dimensions.

[0056] Among them, mRNA expression level directly reflects the degree to which a gene is activated for transcription and serves as a template for protein synthesis. It is commonly quantified using standardized metrics such as FPKM (few kilobase transcripts per million mapping readings) and TPM (transcripts per million).

[0057] Non-coding RNA expression levels include lncRNA (long non-coding RNA), miRNA (microRNA), circRNA (circular RNA), etc. Although they do not directly encode proteins, they participate in gene expression regulation, and changes in their expression levels can indirectly reflect the functional status of genes.

[0058] Protein expression level: The abundance of proteins detected by technologies such as mass spectrometry and protein microarrays can directly reflect the functional effect of gene expression (avoiding the problem of "transcription-translation inconsistency").

[0059] Protein modification level: such as post-translational modifications like phosphorylation, acetylation, and methylation. These modifications significantly affect protein function, and their quantitative data (such as the abundance of phosphorylation sites in a protein) are an important supplement to the gene expression functional status.

[0060] Methods for constructing static feature spaces and dynamic temporal series spaces include:

[0061] Based on gene expression data, the gene expression values ​​of the sample at a preset fixed time point are used as feature inputs. After standardization and feature dimensionality reduction, the static feature vector of the sample is obtained, forming a static feature space that characterizes the molecular state of the sample at a static time section.

[0062] Meanwhile, in the dynamic time series space, the gene expression data of the samples at different continuous time points are arranged in chronological order and organized into a multidimensional time series. The processed time series is mapped into a low-dimensional trajectory vector by principal component analysis, which serves as the coordinate point of the sample in the dynamic time series space and is embedded to form the dynamic time series space.

[0063] Methods for preliminary clustering of static feature spaces include:

[0064] The static feature vectors of all samples in the static feature space are standardized to make the expression levels of different genes comparable. Based on the standardized static feature vectors, an unsupervised clustering method is used to group the samples. The unsupervised clustering method can be any one of K-means clustering, hierarchical clustering, or density-based clustering to obtain the initial static grouping of samples in the static feature space.

[0065] Methods for performing secondary clustering on samples within a group include:

[0066] For samples belonging to the same static group, gene expression data at different consecutive time points are extracted to form time series trajectories; the similarity between time series samples is measured by the cosine similarity method, and a similarity matrix of samples within the group is constructed based on the similarity measurement results;

[0067] Based on the similarity matrix, time series clustering methods are used to perform secondary clustering of samples. For example, K-means clustering based on dynamic time warping, time series clustering methods based on hierarchical clustering, or trajectory clustering based on Gaussian process mixture models are used to refine samples in the same static group into different dynamic evolution subgroups. Each dynamic evolution subgroup corresponds to a class of samples with similar gene expression evolution trajectories in the time dimension.

[0068] Methods for obtaining dynamic trajectory models include:

[0069] For each dynamic evolution subgroup, gene expression data at continuous time points are extracted. The gene expression data is preprocessed, including standardization, smoothing, and noise suppression. A preset function space is selected as the fitting form for the dynamic trajectory model. By minimizing the fitting error, the samples in the dynamic evolution subgroup are fitted in the function space. Finally, a dynamic trajectory model that can characterize the trend of gene expression over time is generated for each dynamic evolution subgroup.

[0070] Methods for calculating the deviation between a sample and the corresponding dynamic trajectory model prediction include:

[0071] At any given time point, the gene expression value of the sample in the static feature space is extracted and compared with the predicted value of the dynamic trajectory model to which the sample belongs at the corresponding time point. The difference between the two is calculated as the residual. The absolute value of the residual is normalized according to the standard deviation of the residual within the dynamic evolution subgroup to which the sample belongs at the corresponding time point (plus a zero-prevention constant) to obtain the normalized deviation value. The normalized deviation values ​​of all time points are averaged to obtain the overall deviation of the sample over the entire time series, which is then used as the deviation of the sample from the predicted value of the corresponding dynamic trajectory model.

[0072] In addition, the process of calculating the deviation can be expressed using a mathematical formula: Deviation ;in, Indicates sample Deviation from the corresponding trajectory model; This represents the total number of time points, i.e., the number of time points at which gene expression values ​​of the sample were collected; Indicates the first A point in time; Indicates sample At the point of time Actual observed gene expression values; Indicates the initial static group to which the sample belongs. Corresponding dynamic evolution subgroup The established dynamic trajectory model, at time point For the sample The predicted value; Indicates the dynamic evolution subgroup to which the sample belongs at time point The residual standard deviation is used to characterize the uncertainty of the model prediction; Index representing a point in time; This indicates a non-zero constant introduced to prevent the denominator from being zero; Indicates the number of the initial static group; This indicates the number of the dynamically evolving subgroup.

[0073] Methods for obtaining abnormal samples include:

[0074] Statistical analysis was performed on the residuals of all samples within the dynamic evolution subgroup. The residual covariance matrix at that time point was obtained by calculating the covariance of the residuals in each dimension. The gene expression values ​​of the dynamic trajectory model and the samples were parametrically fitted to obtain the summary parameter vector of the dynamic trajectory model and the estimated parameter vector of the samples. Then, the cross-spatial consistency deviation was calculated.

[0075] Cross-spatial consistency deviation: ;in, Indicates sample Cross-space consistency deviation; Represents the estimated parameter vector of the sample; Indicates the first In the initial static grouping, the first Summary parameter vectors of dynamic trajectory models for each dynamic evolution subgroup; Indicates sample The inverse of the residual covariance matrix;

[0076] It should be noted that the cross-space consistency deviation here is achieved by parametrically fitting the samples and trajectory models and comparing the differences between the two in the parameter space. It reflects the consistency between the dynamic parameters predicted by static features and the actual trajectory parameters, and belongs to the overall parameter level evaluation index; it focuses on the consistency between static features and dynamic evolution.

[0077] Within a dynamic evolutionary subgroup, the gene expression of many samples exhibits similar trends over time. We aim to characterize this trend using a mathematical formula; for example, gene expression in a dynamic evolutionary subgroup might initially increase and then decrease. We would then fit this curve with a suitable function (such as a polynomial, exponential function, or spline curve) to obtain a dynamic trajectory model. This function contains several parameters (such as slope, inflection point location, and curve curvature), which together constitute the summary parameter vector of this dynamic evolutionary subgroup. This vector serves as a "typical trajectory ID card" for the subgroup, representing its typical pattern.

[0078] For a single sample, we can also use the same "mathematical formula" to fit its observation data, although there are fewer data points and more noise than in the dynamic evolution subgroup; the fitting result will yield a set of parameters (such as its expression level and direction of change), which is the estimated parameter vector of the sample; it can be understood as: the sample's "simplified representation of its corresponding trajectory shape in the static feature space", used for comparison and reference between individuals and groups;

[0079] When the deviation between a sample and the predicted value of the corresponding trajectory model, as well as the deviation from the cross-spatial consistency, both exceed the corresponding preset confidence interval, the sample is determined to be an abnormal sample.

[0080] This approach addresses the following issues with existing technologies: Traditional analysis methods based on static features or single time points struggle to capture the dynamic changes in gene expression over time, resulting in a lack of time-series sensitivity in identifying anomalous samples; existing methods typically ignore natural fluctuations and measurement errors within a group, easily leading to false positives or false negatives when identifying anomalous samples. Traditional methods usually analyze only a single dimension of dynamic or static features, failing to verify the consistency between static features and dynamic trajectories, potentially missing anomalous samples or misclassifying normal samples. Existing technologies often rely on single indicators to identify anomalous samples, easily leading to incomplete and unrobust annomalous sample detection results, failing to simultaneously consider dynamic deviations, residual uncertainties, and cross-spatial consistency.

[0081] The advantages over existing technologies are as follows: by fitting a dynamic trajectory model of gene expression changes over time within each dynamic evolution subgroup and calculating the deviation between the sample and the model prediction, dynamic trajectory deviation anomalous samples that show significant deviation from the trajectory pattern in the time series can be accurately identified, thus overcoming the limitations of existing technologies in dynamic anomaly detection.

[0082] By statistically analyzing the residuals of each time point sample and the fitted trajectory, the covariance matrix of the residuals within the group is calculated, thereby modeling the uncertainty of the residuals. This enables the identification of anomalous samples that are outside the range that the residual covariance can explain, thus enhancing the accuracy and reliability of anomaly detection.

[0083] By introducing a cross-spatial consistency deviation, and comparing the dynamic summary parameters predicted from the static features of a sample with the summary parameters of the actual dynamic trajectory model, this invention can identify samples with excessively large differences between static and dynamic features. This ensures that the determination of anomalous samples considers multi-dimensional spatial information, improving the comprehensiveness of identification. By simultaneously considering dynamic trajectory deviation, residual uncertainty, and cross-spatial consistency, this invention can identify comprehensive anomalous samples that exhibit abnormalities in all three dimensions. This method overcomes the shortcomings of existing technologies that rely on only a single indicator, which can easily lead to one-sided judgments, making anomalous sample detection more comprehensive and robust.

[0084] Methods for obtaining the association strength between nodes include:

[0085] Each dynamic evolution subgroup is used as a core node, and each abnormal sample is used as an independent node. For each node, dynamic trajectory features are extracted in the time series. Between any two nodes, cosine similarity is used to calculate the similarity of dynamic trajectory features, thereby obtaining the correlation strength between each node.

[0086] Methods for obtaining dynamic sentinel gene maps include:

[0087] The calculated association strength is used as the weight of the connecting edges between nodes in the sentinel gene dynamic map. When the association strength between two nodes is greater than or equal to a preset association strength threshold, a connecting edge is established between the two nodes, and the weight of the connecting edge is set to the association strength value between the two nodes. If the association strength between the independent node corresponding to the anomalous sample and any core node is less than the preset association strength threshold, it is kept isolated to reveal potential unique anomalous patterns. All nodes and their connecting edges are combined into a weighted undirected graph to obtain the sentinel gene dynamic map. In the sentinel gene dynamic map, the connecting edges between core nodes reveal the similarity of typical evolutionary patterns, while the connecting edges between core nodes and independent nodes reveal the source and degree of anomalous deviations, thus revealing both typical evolutionary patterns and anomalous branch paths in the map.

[0088] The preset confidence interval is set by staff based on the results of historical data analysis. This historical analysis process includes the system collecting the deviation of multiple data points and calculating their average value as a reference. It can be adjusted by staff during system operation according to the actual situation. Similarly, preset correlation strength thresholds are set.

[0089] In this embodiment, by simultaneously constructing a static feature space and a dynamic temporal space, and combining two-level clustering to achieve "static grouping-dynamic subdivision", the evolutionary trajectory of gene expression can be accurately depicted on the basis of global hierarchical structure. This effectively avoids the misclassification problem of statically similar but dynamically different samples in existing methods, thereby revealing the key subtypes and inflection points of disease progression.

[0090] Secondly, by fitting a dynamic trajectory model within a dynamic evolution subgroup and calculating a dual deviation based on model residuals and cross-spatial consistency, a statistical discrimination mechanism for abnormal samples is established. This mechanism is more objective and robust than the traditional empirical threshold method, and can significantly reduce false positives and false negatives, improve the accuracy of abnormal sample detection, and facilitate researchers in capturing important information related to disease phenotypes, drug responses, or experimental biases.

[0091] Furthermore, a dynamic atlas construction method is introduced, with dynamic evolutionary subgroups as core nodes and anomalous samples as independent nodes. The node associations are quantified through trajectory feature similarity to form a dynamic atlas of sentinel genes with structured expression capabilities. This atlas can not only intuitively display typical evolutionary patterns, but also reveal the branching paths and attribution tendencies of anomalous samples, providing an intuitive tool for a global understanding of gene networks.

[0092] In summary, this invention overcomes the limitations of traditional methods, providing more refined, dynamic, and visualized data processing tools for sentinel gene bank research, thereby enhancing the value of disease mechanism exploration and clinical application.

[0093] Example 2

[0094] Please see Figure 2 As shown, parts not described in detail in this embodiment are described in Embodiment 1. A sentinel gene bank data processing system based on cluster analysis is provided, including:

[0095] The feature mapping module extracts gene expression data from samples in the sentinel gene bank and constructs a static feature space and a dynamic temporal space.

[0096] The hierarchical temporal clustering module performs preliminary clustering on the static feature space to obtain initial static groups; within each static group, it switches to the dynamic temporal space and performs secondary clustering on the samples within the group based on time series similarity, refining them into dynamic evolutionary subgroups.

[0097] The trajectory deviation detection module fits a dynamic trajectory model of gene expression changes over time for each dynamic evolution subgroup, calculates the deviation between the sample and the predicted value of the corresponding dynamic trajectory model in the static feature space, and marks samples with deviation exceeding the preset confidence interval as abnormal samples.

[0098] The gene association construction module uses dynamic evolution subgroups as core nodes and abnormal samples as independent nodes. It calculates the association strength between nodes based on trajectory feature similarity and constructs a dynamic map of sentinel genes using the association strength as edges.

[0099] Since the electronic device described in this embodiment is the electronic device used to implement the sentinel gene bank data processing method based on cluster analysis in the embodiments of this application, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the sentinel gene bank data processing method based on cluster analysis described in the embodiments of this application. Therefore, how the electronic device implements the method in the embodiments of this application will not be described in detail here. Any electronic device used by those skilled in the art to implement the sentinel gene bank data processing method based on cluster analysis in the embodiments of this application falls within the scope of protection of this application.

[0100] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0101] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for processing sentinel gene bank data based on cluster analysis, characterized in that, include: S1. Extract gene expression data from samples in the sentinel gene bank and construct a static feature space and a dynamic temporal space; The method for constructing the static feature space and the dynamic temporal space includes: Based on gene expression data, the gene expression values ​​of the sample at a preset fixed time point are used as feature inputs. After standardization and feature dimensionality reduction, the static feature vector of the sample is obtained, forming a static feature space that characterizes the molecular state of the sample at a static time section. Meanwhile, in the dynamic time series space, the gene expression data of the samples at different continuous time points are arranged in chronological order and organized into a multidimensional time series. The processed time series is mapped into a low-dimensional trajectory vector by principal component analysis, which serves as the coordinate point of the sample in the dynamic time series space and is embedded to form the dynamic time series space. S2. Perform preliminary clustering on the static feature space to obtain initial static groups; switch to the dynamic time series space within each static group, and perform secondary clustering on the samples within the group based on time series similarity to refine them into dynamic evolution subgroups. The method for performing secondary clustering on samples within a group includes: For samples belonging to the same initial static group, gene expression data at different consecutive time points are extracted to form time series trajectories; the similarity between time series samples is measured by the cosine similarity method, and a similarity matrix of samples within the group is constructed based on the similarity measurement results; Based on the similarity matrix, a time series clustering method is used to perform secondary clustering of the samples, refining the samples in the same initial static group into different dynamic evolution subgroups. Each dynamic evolution subgroup corresponds to a class of samples with similar gene expression evolution trajectories in the time dimension. S3. Fit a dynamic trajectory model of gene expression over time for each dynamic evolution subgroup, calculate the deviation between the sample and the predicted value of the corresponding dynamic trajectory model in the static feature space, and mark the sample with the deviation exceeding the preset confidence interval as an abnormal sample. The method for calculating the deviation between the sample and the corresponding dynamic trajectory model prediction value includes: At any given time point, the gene expression value of the sample in the static feature space is extracted and compared with the predicted value of the dynamic trajectory model to which the sample belongs at the corresponding time point. The difference between the two is calculated as the residual. The absolute value of the residual is normalized according to the standard deviation of the residual within the dynamic evolution subgroup to which the sample belongs at the corresponding time point to obtain the normalized deviation value. The normalized deviation values ​​of all time points are averaged to obtain the overall deviation of the sample over the entire time series, which is then used as the deviation of the sample from the predicted value of the corresponding dynamic trajectory model. S4. Using dynamic evolution subgroups as core nodes and abnormal samples as independent nodes, calculate the correlation strength between nodes based on trajectory feature similarity, and construct a dynamic map of sentinel genes with correlation strength as the edge. The method for obtaining the abnormal samples includes: Statistical analysis was performed on the residuals of all samples within the dynamic evolution subgroup. The residual covariance matrix at that time point was obtained by calculating the covariance of the residuals in each dimension. The gene expression values ​​of the dynamic trajectory model and the samples were parametrically fitted to obtain the summary parameter vector of the dynamic trajectory model and the estimated parameter vector of the samples. Then, the cross-spatial consistency deviation was calculated. When the deviation between a sample and the predicted value of the corresponding trajectory model, as well as the deviation from the cross-spatial consistency, both exceed the corresponding preset confidence interval, the sample is determined to be an abnormal sample. Each dynamic evolution subgroup is used as a core node, and each abnormal sample is used as an independent node. For each node, dynamic trajectory features are extracted in the time series. Between any two nodes, cosine similarity is used to calculate the similarity of dynamic trajectory features, thereby obtaining the correlation strength between each node. The calculated association strength is used as the weight of the connecting edge between nodes in the sentinel gene dynamic graph. When the association strength between two nodes is greater than or equal to the preset association strength threshold, a connecting edge is established between the two nodes, and the weight of the connecting edge is set to the association strength value between the two nodes. If the association strength between the independent node corresponding to the abnormal sample and any core node is less than the preset association strength threshold, it is kept isolated. All nodes and their connecting edges are combined into a weighted undirected graph to obtain the sentinel gene dynamic graph.

2. The method for processing sentinel gene bank data based on cluster analysis according to claim 1, characterized in that, The gene expression data includes transcriptome data and proteome data; transcriptome data refers to the expression levels of RNA molecules produced by gene transcription, including mRNA expression levels and non-coding RNA expression levels; proteome data refers to the expression levels of proteins generated by the translation of gene transcription products, including protein expression levels and protein modification levels.

3. The method for processing sentinel gene bank data based on cluster analysis according to claim 2, characterized in that, The method for preliminary clustering of the static feature space includes: The static feature vectors of all samples in the static feature space are standardized to make the expression levels of different genes comparable. Based on the standardized static feature vectors, an unsupervised clustering method is used to group the samples to obtain the initial static groups of samples in the static feature space.

4. The method for processing sentinel gene bank data based on cluster analysis according to claim 3, characterized in that, The method for obtaining the dynamic trajectory model includes: For each dynamic evolution subgroup, gene expression data at continuous time points are extracted; the gene expression data are preprocessed; a preset function space is selected as the fitting form of the dynamic trajectory model, and the samples in the dynamic evolution subgroup are fitted in the function space by minimizing the fitting error. Finally, a dynamic trajectory model that can characterize the trend of gene expression over time is generated for each dynamic evolution subgroup.

5. The method for processing sentinel gene bank data based on cluster analysis according to claim 4, characterized in that, The methods for obtaining the correlation strength between the nodes include: Each dynamic evolution subgroup is used as a core node, and each abnormal sample is used as an independent node. For each node, dynamic trajectory features are extracted in the time series. Between any two nodes, cosine similarity is used to calculate the similarity of dynamic trajectory features, thereby obtaining the correlation strength between each node.

6. The method for processing sentinel gene bank data based on cluster analysis according to claim 5, characterized in that, The method for obtaining the sentinel gene dynamic map includes: The calculated association strength is used as the weight of the connecting edge between nodes in the sentinel gene dynamic graph. When the association strength between two nodes is greater than or equal to the preset association strength threshold, a connecting edge is established between the two nodes, and the weight of the connecting edge is set to the association strength value between the two nodes. If the association strength between the independent node corresponding to the abnormal sample and any core node is less than the preset association strength threshold, it is kept isolated. All nodes and their connecting edges are combined into a weighted undirected graph to obtain the sentinel gene dynamic graph.

Images