Feature extraction method and system for CAR-NK cell data

Abstract

The present invention relates to the field of feature extraction technology, specifically to a feature extraction method and system for CAR-NK cell data, comprising the following steps: collecting CAR-NK cell samples, performing protein analysis on the CAR-NK cell samples in batches, recording changes in protein expression in each batch of CAR-NK cell samples, and performing quantitative analysis on metabolite samples to obtain metabolite concentration data, and standardizing the concentration data to obtain basic data of the cell samples. In the present invention, by quantitatively analyzing proteins and metabolites in CAR-NK cell samples, the response characteristics of cells under different immune environments can be fully captured, immune response grouping and cluster analysis clarify the impact of the immune environment on cell function, the protein-metabolite interaction network reveals the interaction between molecules inside and outside the cell, and by screening and verifying immune markers, provides an accurate basis for immune function assessment. The quantitative analysis of immune reactivity characteristics can track the immune response process in real time, thereby improving the accuracy and depth of research.

Description

Technical Field

[0001] The present invention relates to the field of feature extraction technology, and in particular to a method and system for extracting features from CAR-NK cell data. Background Art

[0002] Feature extraction is a crucial step in data analysis and machine learning, aiming to extract meaningful information from raw data and transform it into a form that is easier to process and analyze. Across various fields, especially in bioinformatics and medicine, feature extraction techniques are widely used in data dimensionality reduction, pattern recognition, and classification tasks. These techniques involve identifying and extracting representative features from data, effectively expressing the characteristics of samples and thereby improving the accuracy and efficiency of machine learning models. These techniques encompass a variety of methods, such as statistical feature extraction, frequency domain analysis, time domain analysis, image processing, and signal processing, often relying on domain knowledge to select and extract the most useful features for the target task.

[0003] Feature extraction methods for CAR-NK cell data are primarily used to analyze and process data from CAR-NK cell therapies. CAR-NK cells are genetically engineered natural killer (NK) cells with enhanced tumor-killing capabilities. Feature extraction can extract various biological characteristics of cells from experimental data, helping researchers understand the performance and responses of CAR-NK cells in different environments. This is of great significance for further analysis of CAR-NK cell function, evaluation of clinical efficacy, and optimization of CAR-NK cell therapy design.

[0004] Existing technologies rely heavily on a single data dimension during data processing, neglecting the interactions between proteins and metabolites. This makes it impossible to fully explore the multidimensional characteristics of cells and reveal detailed differences in cellular responses. Furthermore, they often lack quantitative analysis capabilities when assessing immune responses, resulting in an incomplete description of immune responses and difficulty in achieving accurate immune function monitoring. In particular, in the biomarker validation phase, existing technologies fail to provide effective multidimensional verification methods, making the functional confirmation of biomarkers susceptible to errors and unable to guarantee their reliability in complex immune environments. This results in insufficient depth and accuracy in research results, hindering the further optimization and clinical application of immunotherapy. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and propose a feature extraction method for CAR-NK cell data.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a feature extraction method for CAR-NK cell data, comprising the following steps:

[0007] S1: Collect CAR-NK cell samples, perform protein analysis on the CAR-NK cell samples in batches, record the changes in protein expression in each batch of CAR-NK cell samples, perform quantitative analysis on metabolite samples, obtain metabolite concentration data, and standardize the concentration data to obtain basic cell sample data;

[0008] S2: Based on the basic data of the cell samples, denoising the data, extracting key differences between samples through data dimensionality reduction, clustering the samples according to their characteristics, grouping the samples according to their immune responses, extracting key information, and obtaining integrated feature data;

[0009] S3: Based on the integrated feature data, a protein-metabolite interaction network is constructed. By connecting proteins with related metabolites, the interaction between proteins and metabolites in the immune response process is evaluated, the changes in metabolic pathways and protein interactions are identified, and a functional network diagram is generated;

[0010] S4: Based on the functional network diagram, screen key protein and metabolite markers associated with the immune response, confirm candidate markers based on literature and databases, and validate the candidate markers. Based on the validation results, detect the expression of each marker in CAR-NK cells, analyze the role of each marker in the immune response, and obtain marker validation data;

[0011] S5: Based on the validation marker data, quantitatively analyze the immune reactivity of CAR-NK cells, monitor the contact reaction between CAR-NK cells and target cells, evaluate the dynamic process of the immune response, quantify the reactivity of key markers, and obtain the CAR-NK cell immune reactivity feature extraction results.

[0012] As a further solution of the present invention, the steps for obtaining the basic data of the cell sample are:

[0013] S111: Collect CAR-NK cell samples, extract proteins from each batch of samples, and perform protein quantitative analysis to obtain protein expression data for each batch of samples;

[0014] S112: Comparing the protein expression data in each batch of samples, determining protein expression changes through differential analysis, recording the protein expression changes in each batch of samples, and obtaining protein change data in each sample;

[0015] S113: Extracting metabolite components from each batch of samples based on the protein change data in each sample, obtaining concentration data of each metabolite, and normalizing the metabolite concentration data. Combined with the protein expression change data in each batch of samples, basic cell sample data is obtained.

[0016] As a further solution of the present invention, the step of acquiring the integrated feature data is:

[0017] S211: Based on the basic data of the cell sample, each data point is evaluated, a denoising threshold is determined according to the distribution characteristics of the sample data, and outliers exceeding the denoising threshold are eliminated to obtain denoised sample data;

[0018] S212: performing data dimensionality reduction on the sample data after the denoising process and extracting key differences between samples, retaining key features between samples, and eliminating redundant information to obtain dimensionality-reduced data;

[0019] S213: Based on the data after dimensionality reduction, sample cluster analysis is performed using the formula:

[0020]

[0021] Calculate the Euclidean distance d between sample i and sample j ij The smaller the Euclidean distance between samples, the higher the similarity, and samples with similar immune response characteristics are grouped into the same group to obtain the integrated feature data, where x ik and x jk are the kth eigenvalues ​​of sample i and sample j respectively, and n is the total number of features.

[0022] As a further solution of the present invention, the steps of obtaining the functional network diagram are:

[0023] S311: Based on the integrated feature data, by analyzing the pairing relationship between each protein and its related metabolites, detecting the relationship between proteins and metabolites, establishing the connection between each protein and related metabolites, and generating a preliminary interaction network structure;

[0024] S312: Based on the protein-metabolite interaction network structure, the formula is used:

[0025]

[0026] Calculate the interaction strength S between protein a and metabolite b ab , to obtain the interaction strength data of each pair of protein-metabolite, where x ak and y bk are the expression levels of protein a and metabolite b on the kth feature, respectively, and n represents the total number of features;

[0027] S313: Based on the interaction strength data of each pair of protein-metabolite, identify metabolic pathways related to the immune response, analyze the changes in the interaction between metabolic pathways and proteins during the immune response, reveal the key protein-metabolite interactions in the immune response mechanism, and generate a functional network diagram.

[0028] As a further embodiment of the present invention, the verification step of the candidate marker is:

[0029] S411: Based on the functional network diagram, key proteins and metabolite markers associated with the immune response are screened. Combined with key biological events in the immune response process, the screened markers are ranked according to their potential role in the immune response to obtain preliminary screened candidate markers;

[0030] S412: Based on the candidate markers initially screened, research literature is consulted to analyze and compare the immune relevance of the candidate markers to confirm whether they have biological support in immune response, and whether each candidate marker has an expression pattern of known immune function to obtain confirmed candidate markers;

[0031] S413: Based on the confirmed candidate markers, the candidate markers are verified, the expression data of the markers are obtained through experiments, the actual role of the markers in the immune response is verified, the changes in the immune cells are detected, and a list of candidate markers is obtained.

[0032] As a further embodiment of the present invention, the steps for obtaining the marker verification data are as follows:

[0033] S421: Based on the candidate marker list, perform expression detection of each marker in the CAR-NK cells, analyze the expression level of the marker in the cells, reveal the changes of the marker in the immune response, and generate marker expression data;

[0034] S422: Based on the marker expression data, compare the expression changes of the markers before and after the immune response and under different stimulation conditions, evaluate the performance during the immune response, confirm the correlation between each marker and the key process of the immune response, and generate marker immune effect analysis data;

[0035] S423: Based on the immune effect analysis data, identify the potential function of each marker in the immune response, analyze the potential biological mechanism in combination with the experimental data, and obtain marker verification data.

[0036] As a further embodiment of the present invention, the step of obtaining the CAR-NK cell immunoreactivity feature extraction result is:

[0037] S511: Based on the validation marker data, quantitatively analyzing the CAR-NK cell immune reactivity, monitoring the different stages of the immune response through immune stimulation and target cell contact reaction, and performing real-time quantitative analysis of the cell contact reaction to obtain quantitative data of the CAR-NK cell immune reactivity;

[0038] S512: Observe and record the interaction between target cells and CAR-NK cells based on the quantitative data of CAR-NK cell immune reactivity, quantify the reactivity of the marker, analyze the performance of the marker in the cell contact reaction under different immune stimulation conditions, analyze the relationship between the marker and the cell contact reaction, and obtain quantified marker reactivity data;

[0039] S513: In combination with the quantified marker reactivity data, analyze the changing trend of the markers in the differential immune response stage and their relationship with the CAR-NK cell response, determine the temporal dynamics of the cell response and marker changes, and obtain the CAR-NK cell immune reactivity feature extraction results.

[0040] The feature extraction system for CAR-NK cell data includes:

[0041] The sample data processing module collects CAR-NK cell samples, extracts proteins from each batch of samples, performs protein quantitative analysis, determines protein expression changes through differential analysis, extracts metabolite components from each batch of samples, obtains concentration data, performs standardization, and combines the protein expression change data to obtain basic cell sample data;

[0042] The feature extraction module determines the denoising threshold based on the basic data of the cell samples and the distribution characteristics of the sample data, performs data dimensionality reduction and extracts the key differences between samples, calculates the Euclidean distance between samples, and groups samples with similar immune response characteristics into the same group to obtain integrated feature data;

[0043] The mutual relationship analysis module detects the mutual relationship between proteins and metabolites based on the integrated feature data, calculates the interaction strength between proteins and metabolites, identifies metabolic pathways related to immune response, reveals key protein-metabolite interactions in the immune response mechanism, and generates a functional network diagram;

[0044] The marker validation module screens key protein and metabolite markers associated with the immune response based on the functional network diagram, verifies whether each candidate marker has an expression pattern of known immune function, detects changes in immune cells, evaluates performance during the immune response, confirms the association of each marker with key immune response processes, identifies the potential function of each marker in the immune response, and obtains marker validation data;

[0045] Based on the validation marker data, the immune response monitoring module monitors the differential stages of the immune response through immune stimulation and target cell contact reaction, observes and records the interaction between target cells and CAR-NK cells, analyzes the changing trends of markers in the differential immune response stages and their relationship with CAR-NK cell response, determines the temporal dynamics of cell response and marker changes, and obtains the CAR-NK cell immune reactivity feature extraction results.

[0046] Compared with the prior art, the advantages and positive effects of the present invention are:

[0047] In the present invention, by quantitatively analyzing the proteins and metabolites in CAR-NK cell samples, the response characteristics of cells in different immune environments can be fully captured. Immune response grouping and cluster analysis clarify the impact of the immune environment on cell function. The protein-metabolite interaction network reveals the interaction between molecules inside and outside the cell. By screening and verifying immune markers, an accurate basis for immune function assessment is provided. Quantitative analysis of immune reactivity characteristics can track the immune response process in real time, improving the accuracy and depth of research. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flow chart of the main steps of the present invention;

[0049] Figure 2 This is a flowchart for obtaining basic data of cell samples according to the present invention;

[0050] Figure 3 A flowchart for obtaining integrated feature data for the present invention;

[0051] Figure 4 A flowchart for obtaining the functional network diagram of the present invention;

[0052] Figure 5 is a flow chart for the validation of candidate markers of the present invention;

[0053] Figure 6 This is a flow chart for obtaining marker validation data of the present invention;

[0054] Figure 7 Flowchart for obtaining the results of CAR-NK cell immune reactivity feature extraction of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention 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 invention and are not intended to limit the present invention.

[0056] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0057] See also Figure 1 , a feature extraction method for CAR-NK cell data, comprising the following steps:

[0058] S1: Collect CAR-NK cell samples, perform protein analysis on the CAR-NK cell samples in batches, record the changes in protein expression in each batch of CAR-NK cell samples, perform quantitative analysis on metabolite samples, obtain metabolite concentration data, and standardize the concentration data to obtain basic cell sample data;

[0059] S2: Based on the basic data of cell samples, the data is denoised and the key differences between samples are extracted through data dimensionality reduction. The samples are clustered according to their characteristics and grouped according to their immune responses to extract key information and obtain integrated feature data.

[0060] S3: Based on the integrated feature data, a protein-metabolite interaction network is constructed. By connecting proteins with related metabolites, the interactions between proteins and metabolites in the immune response process are evaluated, the changes in metabolic pathways and protein interactions are identified, and a functional network diagram is generated.

[0061] S4: Based on the functional network diagram, screen key protein and metabolite markers associated with the immune response, confirm candidate markers based on literature and databases, and validate the candidate markers. Based on the validation results, detect the expression of each marker in CAR-NK cells, analyze the role of each marker in the immune response, and obtain marker validation data;

[0062] S5: Based on the validation marker data, quantitatively analyze the immune reactivity of CAR-NK cells, monitor the contact reaction between CAR-NK cells and target cells, evaluate the dynamic process of the immune response, and combine flow cytometry data to quantify the reactivity of key markers to obtain the CAR-NK cell immune reactivity feature extraction results.

[0063] Basic cell sample data include protein expression data, metabolite concentration data and standardized data; integrated feature data include denoised data, key differential features and immune response grouping results; functional network diagrams include protein-metabolite interaction relationships, metabolic pathway change analysis results and protein interaction change analysis results; marker validation data include screened candidate markers, literature and database confirmation results and marker expression data in CAR-NK cells; immune reactivity feature extraction results include contact reaction data between CAR-NK cells and target cells, dynamic process analysis results of immune response and reactivity quantification data of key markers.

[0064] See also Figure 2 , the steps for obtaining basic data of cell samples are:

[0065] S111: Collect CAR-NK cell samples, extract proteins from each batch of samples, and perform protein quantitative analysis to obtain protein expression data for each batch of samples;

[0066] After collecting the CAR-NK cell sample, the protein in the cell is first extracted using an appropriate lysis solution. The extraction step involves mixing the cell sample with a lysis solution containing a buffer, and by controlling the appropriate temperature and time, the cell membrane is broken and the protein is released. In order to ensure that the extracted protein can be separated and measured stably and completely, it is necessary to use a lysis solution that does not contain enzymes to avoid protein degradation. In the extracted sample, the protein needs to undergo a centrifugation process to remove cell debris and other impurities to obtain a purified protein solution for analysis. Subsequently, the purified protein sample is transferred to the analysis platform for protein quantification analysis, using high-precision technology to detect the concentration and type of protein in each batch of samples, thereby recording the protein data of each sample. In this step, the detection of protein content must not only ensure the accurate protein expression level in the sample, but also require multiple rechecks to ensure that the protein data extracted from each batch of samples is sufficiently representative and consistent.

[0067] S112: Comparing the protein expression data in each batch of samples, determining the expression changes of the proteins through differential analysis, recording the changes in protein expression in each batch of samples, and obtaining the protein change data in each sample;

[0068] After obtaining the protein data for each batch of samples, the next step is to perform a differential analysis of the protein expression data. First, the protein data measured in each batch of samples are compared one by one, and the differences in the expression levels of each protein in different samples are recorded. When comparing, it is necessary to analyze whether there are obvious expression fluctuations or significant changes between different batches. During the analysis process, the selected protein expression differences should meet the set standards, and statistical tests are usually used to detect these changes. For example, a paired t-test or analysis of variance is performed to determine whether the differences in protein expression between different batches are statistically significant. In this process, the p-value of each protein expression level needs to be calculated, and the significance level needs to be set to determine which proteins have statistically significant differences in expression in different samples. After completing the differential analysis, the generated protein expression change data will be recorded in a special file or database to provide the necessary background data support for subsequent metabolite analysis.

[0069] S113: Based on the protein change data in each sample, extract the metabolite components in each batch of samples, obtain the concentration data of each metabolite, standardize the metabolite concentration data, and combine it with the protein expression change data to obtain the basic data of the cell samples;

[0070] To extract the metabolite components of each batch of CAR-NK cell samples, the cell samples are first placed in a solution, and the metabolites within the cells are separated using a solvent. To ensure that the separated metabolites can be accurately detected in subsequent analysis, an appropriate solvent system is selected to dissolve the cellular contents. The sample is then filtered or centrifuged to remove unwanted impurities and insoluble matter, ensuring that only metabolites remain in solution. During the analysis process, the processed sample is introduced into the analytical instrument for quantitative metabolite analysis using mass spectrometry. This process requires the establishment of an appropriate detection standard curve to ensure accurate metabolite concentration measurement. The concentration of each metabolite in the sample is calculated by comparing it to the standard curve, ensuring the stability and reproducibility of the metabolite concentration data for each sample across multiple experiments. Finally, the obtained concentration data is normalized, typically using the Z-score method, to make the concentration data of each metabolite in each batch of samples comparable, ensuring data consistency between different batches and providing reliable quantitative results for further analysis.

[0071] See also Figure 3 , the steps for obtaining integrated feature data are:

[0072] S211: Based on the basic data of the cell sample, each data point is evaluated, a denoising threshold is determined according to the distribution characteristics of the sample data, and outliers exceeding the denoising threshold are eliminated to obtain denoised sample data;

[0073] Based on the basic data of CAR-NK cell samples, the data is denoised to ensure the accuracy and reliability of the analysis results. The denoising process starts with the detection of each data point. First, statistical analysis is performed on each data in the sample to calculate basic statistics such as its mean and standard deviation. Then, according to the set denoising criteria, abnormal data points that exceed the normal distribution range or exceed a certain standard deviation are removed. Specifically, if the deviation of a data point from the mean exceeds the set threshold (such as mean ± 3 times standard deviation), it is considered an outlier. After preliminary screening, continue to check for any duplicates, missing data or significantly deviated values ​​in the data, and perform corresponding processing, such as supplementing missing values ​​or removing duplicate values. After the denoising process is completed, the remaining data set will be cleaner and meet the analysis requirements, and can provide real and effective sample data for subsequent dimensionality reduction and clustering analysis.

[0074] S212: Based on the sample data after denoising, perform data dimensionality reduction and extract key differences between samples, retain key features between samples, and eliminate redundant information to obtain data after dimensionality reduction;

[0075] Dimensionality reduction is performed on the denoised cell sample data to more clearly reveal the key differences between samples. The core of the dimensionality reduction process lies in extracting the most representative features from the multi-dimensional data. First, each data point is analyzed across all dimensions to identify the key dimensions that explain the data variance. During this process, each feature in the dataset is evaluated in relation to other features and its role in distinguishing samples, selecting those that are most helpful in distinguishing between samples. By calculating the contribution of each feature, the impact of each dimension on the data variation is determined, retaining the feature dimensions with the greatest contribution and reducing redundant data. Next, the data points are reconstructed, re-representing each sample using the selected key features. This compresses the original multi-dimensional data into a lower-dimensional space, ensuring that sufficient information is retained after dimensionality reduction. The reduced dimensionality data facilitates the identification of significant differences between samples and provides a more streamlined and representative dataset for subsequent cluster analysis.

[0076] S213: Based on the data after dimensionality reduction, perform sample cluster analysis using the formula:

[0077]

[0078] Calculate the Euclidean distance d between sample i and sample j ij The smaller the Euclidean distance between samples, the higher the similarity, and samples with similar immune response characteristics are grouped into the same group to obtain the integrated feature data, where x ik and x jk are the kth eigenvalues ​​of sample i and sample j respectively, and n is the total number of features;

[0079] There are two samples, sample i and sample j, and the values ​​of the two features (feature 1 and feature 2) are:

[0080] Sample i: Feature 1 = 10, Feature 2 = 5;

[0081] Sample j: Feature 1 = 12, Feature 2 = 6;

[0082] Calculate the Euclidean distance according to the above formula:

[0083]

[0084] Calculate the square of the variance for each feature:

[0085] (10-12) 2 =(-2) 2 =4

[0086] (5-6) 2 =(-1) 2 =1

[0087] Add the two squared results:

[0088] 4+1=5

[0089] Calculate the square root:

[0090]

[0091] The calculated result is 2.236, indicating that the similarity between sample i and sample j is low. A smaller Euclidean distance indicates that the difference between the two samples is smaller and the more similar they are; a larger distance indicates a greater difference and a lower similarity between the samples. A distance of 2.236 indicates that the difference between sample i and sample j on feature 1 and feature 2 is relatively small, but the difference still exists, making it suitable as a similarity metric for subsequent cluster analysis or other classification tasks.

[0092] See also Figure 4 , the steps to obtain the functional network diagram are:

[0093] S311: Based on the integrated feature data, the pairing relationship between each protein and its related metabolites is analyzed to detect the relationship between proteins and metabolites, establish the connection between each protein and related metabolites, and generate a preliminary interaction network structure;

[0094] Based on the integrated feature data, it is first necessary to construct a protein-metabolite interaction network. First, all proteins and metabolites are extracted from the existing cell sample data, and the expression data of each protein and metabolite are collected and organized. Next, the relationship between each protein and its related metabolites is evaluated, and pairings between proteins and metabolites are established to ensure that the metabolites in each protein pairing have biological relevance. In order to ensure the accuracy of the network, it is necessary to combine clinical data or experimental data related to the immune response to further confirm the connection between proteins and metabolites. Through these data, the connection between proteins and metabolites is gradually constructed, and the importance of each pair of proteins and metabolites is evaluated based on their interaction strength, functional relevance, and immune response performance. Finally, a complete interaction network is formed. The interaction network will be able to reflect the interaction relationship between each protein and its related metabolites in the immune response and provide basic data for subsequent analysis.

[0095] S312: According to the protein-metabolite interaction network structure, the formula is used:

[0096]

[0097] Calculate the interaction strength S between protein a and metabolite b ab , to obtain the interaction strength data of each pair of protein-metabolite, where x ak and y bk are the expression levels of protein a and metabolite b on the kth feature, respectively, and n represents the total number of features;

[0098] There are two protein-metabolite pairs, protein a and metabolite b, and the values ​​of the two features (feature 1 and feature 2) are:

[0099] Protein a: Feature 1 = 10, Feature 2 = 5;

[0100] Metabolite b: feature 1 = 12, feature 2 = 6;

[0101] Compute the dot product:

[0102] x a1 ·y b1 +x a2 ·y b2 =(10×12)+(5×6)=120+30=150

[0103] Calculate the modulus length:

[0104]

[0105] Calculate the interaction strength:

[0106]

[0107] The calculation result is S ab ≈0.999, indicating a very high interaction strength between protein a and metabolite b. The closer the interaction strength value is to 1, the stronger the interaction between the protein and metabolite pair. A value of 0.999 indicates a very close biological association between the protein and metabolite pair, playing a key role in the immune response.

[0108] S313: Based on the interaction strength data of each pair of protein-metabolite, identify metabolic pathways related to immune response, analyze the changes in the interaction between metabolic pathways and proteins during the immune response, reveal the key protein-metabolite interactions in the immune response mechanism, and generate a functional network diagram;

[0109] After establishing the protein-metabolite interaction network and evaluating its relationships, the next step is to identify metabolic pathways related to the immune response and the interactive changes between metabolic pathways and proteins. First, it is necessary to analyze all metabolic pathways involved in the network, and analyze the role played in the immune response process by identifying the key nodes and key proteins of the metabolic pathways. This process includes identifying which metabolic pathways play an important role in the immune response and analyzing the changing trends of proteins and metabolites in each pathway. By combining the performance data of the immune response, analyze how the pathway interacts with other biological events in the immune response process (such as cell activation, immune cell response, etc.). Next, based on the identified metabolic pathways and their key protein-metabolite interactions, a functional network diagram is generated to show the interactive changes between different metabolic pathways and proteins during the immune response. This functional network diagram not only helps to understand the key interactive relationships in the immune response process, but also provides visual support for further research on the mechanism of the immune response.

[0110] See also Figure 5 , the validation steps for candidate markers are:

[0111] S411: Based on the functional network diagram, key proteins and metabolite markers associated with the immune response are screened. Combined with key biological events in the immune response process, the screened markers are ranked according to their potential role in the immune response to obtain preliminary candidate markers;

[0112] Based on the functional network diagram, the first step is to conduct a comprehensive analysis of the interactions between proteins and metabolites. In this process, the data of proteins and metabolites contained in all networks are first collected and organized, and the data are compared with the key features of the immune response process to screen out markers that may play an important role in the immune response process. The relationship between each protein and metabolite and the immune response needs to be confirmed by analyzing its expression changes, and the expression changes should be compared with the data of known immune-related biomarkers. During the screening process, the existing immune response mechanism analysis should be combined to ensure the importance of the selected markers in the immune response. Finally, markers with high research potential are selected, and a list of candidate markers is established to prepare for subsequent verification experiments.

[0113] S412: Based on the initially screened candidate markers, review research literature, analyze and compare the immune relevance of the candidate markers to confirm whether they have biological support in immune response, verify whether each candidate marker has an expression pattern of known immune function, and obtain confirmed candidate markers;

[0114] Next, the selected candidate markers are confirmed, a process mainly carried out through extensive literature research and database comparison. Specifically, first, by consulting relevant research literature, we look for experimental data related to the candidate markers, confirm the known functions of these markers in the immune response and their expression patterns. At the same time, combined with the bioinformatics database of immune response, we verify whether the candidate markers have high biological plausibility and whether they have been widely studied and confirmed. In the database verification stage, it is necessary to ensure that the candidate markers have clear performance under immune-related conditions and can be reproduced in different experiments, so as to confirm the immune relevance of the candidate markers.

[0115] S413: Based on the confirmed candidate markers, verify the candidate markers, obtain the expression data of the markers through experiments, verify the actual role of the markers in the immune response, detect changes in immune cells, and obtain a list of candidate markers;

[0116] After confirming the candidate markers, the experimental validation phase begins. This phase first involves designing and executing a series of validation experiments. The experiments include detecting the expression of candidate markers in CAR-NK cells, usually using experimental methods such as flow cytometry, Western Blot, and qPCR. These experimental techniques can accurately detect the expression levels of markers in immune cells. Compare the expression changes of markers under different immune states to further analyze their role in the immune response. At the same time, it is necessary to compare the performance of markers in different experimental groups to confirm whether these markers play a key role in the immune response, verify their actual role in the immune response through experimental data, and obtain a list of candidate markers to provide support for subsequent analysis of the immune response mechanism.

[0117] See also Figure 6 , the steps for obtaining marker verification data are:

[0118] S421: Based on the candidate marker list, perform expression detection of each marker in CAR-NK cells, analyze the expression level of the marker in the cells, reveal the changes of the marker in the immune response, and generate marker expression data;

[0119] Based on the list of candidate markers, CAR-NK cell samples are collected for immune stimulation treatment. When collecting samples, it is necessary to ensure that the number of cells in each sample is sufficient and that their biological activity is maintained. Next, according to the specific conditions of the immune response, the cells are treated at different time points to ensure that the cells can express the expected immune response under specific immune stimulation. The treated cells are then analyzed, and cell samples are collected and lysed to extract intracellular proteins and RNA. These extracts will then be used for subsequent quantitative testing. During the experimental operation, appropriate control group settings are also required to ensure that the stimulatory effect of the immune response can be accurately monitored. The processing of all experimental samples should be consistent to ensure the reliability of the data and ultimately obtain accurate marker expression data.

[0120] S422: Based on the marker expression data, compare the expression changes of the markers before and after the immune response and under different stimulation conditions, evaluate the performance during the immune response, confirm the correlation between each marker and the key process of the immune response, and generate marker immune effect analysis data;

[0121] After obtaining the expression data of the marker, a more in-depth analysis is needed to evaluate the role of the marker in the immune response. First, the cell samples are grouped according to different stages of the immune response, and the marker expression data in each group are compared. Specifically, it is necessary to analyze the role of the marker in the immune response process based on the change pattern of markers such as the activation state of immune cells and the secretion of immune factors. If the marker has significant expression changes during the immune response and is highly correlated with the functional changes of immune cells, it can be further inferred that it may be involved in the activation or regulation of immune cells. In addition, for different time nodes of the immune response, it is necessary to compare the expression changes of the markers and finally obtain the potential function of each marker in the immune response.

[0122] S423: Based on the immune effect analysis data, identify the potential function of each marker in the immune response, analyze the potential biological mechanism in combination with experimental data, and obtain marker validation data;

[0123] After completing the preliminary analysis of the immune effects of the markers, a comprehensive assessment of the markers is required to further confirm their specific biological functions in the immune response. At this point, it is first necessary to integrate all previously obtained marker expression and immune effect data and compare the performance of the markers under different immune response states. In particular, attention should be paid to the changes in the markers at key time points of the immune response, and combined with the functional status of immune cells, the specific role in the immune response can be inferred. By comparing with other known key factors of the immune response, it is confirmed whether the marker plays a regulatory or control role in the immune response. Finally, by systematically integrating all data, the validation data of the marker is obtained, providing strong support for subsequent research on the mechanism of the immune response.

[0124] See also Figure 7 , the steps for obtaining the CAR-NK cell immune reactivity feature extraction results are:

[0125] S511: Based on the validation marker data, quantitative analysis of CAR-NK cell immune reactivity is performed. Through immune stimulation and target cell contact reaction, the different stages of the immune response are monitored, and the cell contact reaction is quantitatively analyzed in real time to obtain quantitative data of CAR-NK cell immune reactivity.

[0126] Quantitative analysis of CAR-NK cell immune reactivity is required based on validation marker data. This process includes detailed recording of the quantity and magnitude of changes in each marker during the immune response. By organizing and compiling quantitative marker data from the experiment, their performance at different stages of the immune response can be determined. Using the corresponding marker detection values, an immune reactivity index is constructed. By analyzing the relationship between marker changes and immune response intensity, and combining the timing and intensity of interactions between immune cells and target cells, the characteristics of the immune response can be further deduced, ultimately providing data support for the evaluation of CAR-NK cell immune responses. The analysis results help understand marker expression and its impact on the immune response, providing quantitative immune reactivity data and laying the foundation for subsequent immune response analysis.

[0127] S512: Based on the quantitative data of CAR-NK cell immune reactivity, observe and record the interaction between target cells and CAR-NK cells, quantify the reactivity of the markers, analyze the performance of the markers in the cell contact response under different immune stimulation conditions, analyze the relationship between the markers and the cell contact response, and obtain quantified marker reactivity data;

[0128] To monitor the contact reaction between CAR-NK cells and target cells, it is first necessary to regularly collect cell samples at different time points. The contact events at each time point are recorded, and the frequency, intensity, and duration of contact reactions between CAR-NK cells and target cells are analyzed. During flow cytometry experiments, the expression of cell surface markers is observed in real time to quantify cell-cell contact events. The quantity and quality of these contact events can be further analyzed by monitoring the activation of intracellular signaling pathways to understand the progression of the immune response. By combining the changes in various markers during the reaction process with data from multiple stages of the immune response, the intensity of the immune response and the target cell response can be ultimately assessed. The relationship between immune reactivity and target cell contact reactions can be determined, providing a basis for further analysis of immune effects.

[0129] S513: Combined with the quantified marker reactivity data, analyze the changing trends of markers during the differential immune response stage and their relationship with CAR-NK cell response, determine the temporal dynamics of cell response and marker changes, and obtain the CAR-NK cell immune reactivity feature extraction results;

[0130] In the process of combining flow cytometry data to quantify the reactivity of key markers, the first step is to convert the data collected in the experiment into standardized reactivity indicators. By analyzing the performance of the markers at each reaction stage, quantifying the range and rate of change of each marker, and measuring the fluctuation of marker expression levels as the immune response progresses, this quantitative reactivity data is used to construct a dynamic analysis curve of the intensity of the immune response. Furthermore, by comparing the data of the control and experimental groups, the key role of each marker in the immune response is evaluated. The relationship between the quantitative reactivity data of the markers and the intensity of the immune response provides a theoretical basis for further analysis of the immune response process. By summarizing and statistically analyzing the marker validation data, the results of the CAR-NK cell immune reactivity feature extraction are generated.

[0131] The feature extraction system for CAR-NK cell data includes:

[0132] The sample data processing module collects CAR-NK cell samples, extracts proteins from each batch of samples, performs protein quantitative analysis, determines protein expression changes through differential analysis, extracts metabolite components from each batch of samples, obtains concentration data, performs standardization, and combines the protein expression change data to obtain basic cell sample data;

[0133] The feature extraction module determines the denoising threshold based on the basic cell sample data and the distribution characteristics of the sample data, performs data dimensionality reduction, extracts key differences between samples, calculates the Euclidean distance between samples, and groups samples with similar immune response characteristics to obtain integrated feature data.

[0134] The relationship analysis module detects the relationships between proteins and metabolites based on integrated feature data, calculates the interaction strength between proteins and metabolites, identifies metabolic pathways related to immune response, reveals key protein-metabolite interactions in the immune response mechanism, and generates functional network diagrams;

[0135] The marker validation module screens key protein and metabolite markers associated with immune responses based on functional network diagrams, verifies whether each candidate marker has an expression pattern of known immune functions, detects changes in immune cells, evaluates performance during the immune response, confirms the association of each marker with key immune response processes, identifies the potential function of each marker in the immune response, and obtains marker validation data.

[0136] The immune response monitoring module is based on the validation marker data. It monitors the differential stages of the immune response through immune stimulation and target cell contact reaction, observes and records the interaction between target cells and CAR-NK cells, analyzes the changing trends of markers in the differential immune response stages and their relationship with CAR-NK cell response, determines the temporal dynamics of cell response and marker changes, and obtains the CAR-NK cell immune reactivity feature extraction results.

[0137] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A feature extraction method for CAR-NK cell data, characterized in that: The following steps are involved: S1: Collect CAR-NK cell samples, perform protein analysis on the CAR-NK cell samples in batches, record the changes in protein expression in each batch of CAR-NK cell samples, perform quantitative analysis on metabolite samples, obtain metabolite concentration data, and standardize the concentration data to obtain basic cell sample data; S2: Based on the basic data of the cell samples, denoising the data, extracting key differences between samples through data dimensionality reduction, clustering the samples according to their characteristics, grouping the samples according to their immune responses, extracting key information, and obtaining integrated feature data; S3: Based on the integrated feature data, a protein-metabolite interaction network is constructed. By connecting proteins with related metabolites, the interaction between proteins and metabolites in the immune response process is evaluated, the changes in metabolic pathways and protein interactions are identified, and a functional network diagram is generated; The steps for obtaining the functional network diagram are: S311: Based on the integrated feature data, by analyzing the pairing relationship between each protein and its related metabolites, detecting the relationship between proteins and metabolites, establishing the connection between each protein and related metabolites, and generating a preliminary interaction network structure; S312: Based on the protein-metabolite interaction network structure, the formula is used: Calculate the interaction strength S between protein a and metabolite b ab , to obtain the interaction strength data of each pair of protein-metabolite, where x ak and y bk are the expression levels of protein a and metabolite b on the kth feature, respectively, and n represents the total number of features; S313: Based on the interaction strength data of each pair of protein-metabolite, identify metabolic pathways related to the immune response, analyze the changes in the interaction between metabolic pathways and proteins during the immune response, reveal the key protein-metabolite interactions in the immune response mechanism, and generate a functional network diagram; S4: Based on the functional network diagram, screen key protein and metabolite markers associated with the immune response, confirm candidate markers based on literature and databases, and validate the candidate markers. Based on the validation results, detect the expression of each marker in CAR-NK cells, analyze the role of each marker in the immune response, and obtain marker validation data; S5: Based on the validation marker data, quantitatively analyze the immune reactivity of CAR-NK cells, monitor the contact reaction between CAR-NK cells and target cells, evaluate the dynamic process of the immune response, quantify the reactivity of key markers, and obtain the CAR-NK cell immune reactivity feature extraction results.

2. The feature extraction method for CAR-NK cell data according to claim 1, characterized in that: The steps for obtaining the basic data of the cell sample are: S111: Collect CAR-NK cell samples, extract proteins from each batch of samples, and perform protein quantitative analysis to obtain protein expression data for each batch of samples; S112: Comparing the protein expression data in each batch of samples, determining protein expression changes through differential analysis, recording the protein expression changes in each batch of samples, and obtaining protein change data in each sample; S113: Extracting metabolite components from each batch of samples based on the protein change data in each sample, obtaining concentration data of each metabolite, and normalizing the metabolite concentration data. Combined with the protein expression change data in each batch of samples, basic cell sample data is obtained.

3. The feature extraction method for CAR-NK cell data according to claim 2, characterized in that: The steps for obtaining the integrated feature data are: S211: Based on the basic data of the cell sample, each data point is evaluated, a denoising threshold is determined according to the distribution characteristics of the sample data, and outliers exceeding the denoising threshold are eliminated to obtain denoised sample data; S212: performing data dimensionality reduction on the sample data after the denoising process and extracting key differences between samples, retaining key features between samples, and eliminating redundant information to obtain dimensionality-reduced data; S213: Based on the data after dimensionality reduction, sample cluster analysis is performed using the formula: Calculate the Euclidean distance d between sample i and sample j ij The smaller the Euclidean distance between samples, the higher the similarity, and samples with similar immune response characteristics are grouped into the same group to obtain the integrated feature data, where x ik and x jk are the kth eigenvalues ​​of sample i and sample j respectively, and n is the total number of features.

4. The feature extraction method for CAR-NK cell data according to claim 1, characterized in that: The verification steps of the candidate markers are: S411: Based on the functional network diagram, key proteins and metabolite markers associated with the immune response are screened. Combined with key biological events in the immune response process, the screened markers are ranked according to their potential role in the immune response to obtain preliminary screened candidate markers; S412: Based on the candidate markers initially screened, research literature is consulted to analyze and compare the immune relevance of the candidate markers to confirm whether they have biological support in immune response, and whether each candidate marker has an expression pattern of known immune function to obtain confirmed candidate markers; S413: Based on the confirmed candidate markers, the candidate markers are verified, the expression data of the markers are obtained through experiments, the actual role of the markers in the immune response is verified, the changes in the immune cells are detected, and a list of candidate markers is obtained.

5. The feature extraction method for CAR-NK cell data according to claim 4, characterized in that: The steps for obtaining the marker verification data are: S421: Based on the candidate marker list, perform expression detection of each marker in the CAR-NK cells, analyze the expression level of the marker in the cells, reveal the changes of the marker in the immune response, and generate marker expression data; S422: Based on the marker expression data, compare the expression changes of the markers before and after the immune response and under different stimulation conditions, evaluate the performance during the immune response, confirm the correlation between each marker and the key process of the immune response, and generate marker immune effect analysis data; S423: Based on the immune effect analysis data, identify the potential function of each marker in the immune response, analyze the potential biological mechanism in combination with the experimental data, and obtain marker verification data.

6. The feature extraction method for CAR-NK cell data according to claim 5, characterized in that: The steps for obtaining the CAR-NK cell immune reactivity feature extraction result are: S511: Based on the validation marker data, quantitatively analyzing the CAR-NK cell immune reactivity, monitoring the different stages of the immune response through immune stimulation and target cell contact reaction, and performing real-time quantitative analysis of the cell contact reaction to obtain quantitative data of the CAR-NK cell immune reactivity; S512: Observe and record the interaction between target cells and CAR-NK cells based on the quantitative data of CAR-NK cell immune reactivity, quantify the reactivity of the marker, analyze the performance of the marker in the cell contact reaction under different immune stimulation conditions, analyze the relationship between the marker and the cell contact reaction, and obtain quantified marker reactivity data; S513: In combination with the quantified marker reactivity data, analyze the changing trend of the markers in the differential immune response stage and their relationship with the CAR-NK cell response, determine the temporal dynamics of the cell response and marker changes, and obtain the CAR-NK cell immune reactivity feature extraction results.

7. A feature extraction system for CAR-NK cell data, characterized in that: The system is used to perform the feature extraction method for CAR-NK cell data according to any one of claims 1 to 6, comprising: The sample data processing module collects CAR-NK cell samples, extracts proteins from each batch of samples, performs protein quantitative analysis, determines protein expression changes through differential analysis, extracts metabolite components from each batch of samples, obtains concentration data, performs standardization, and combines the protein expression change data to obtain basic cell sample data; The feature extraction module determines the denoising threshold based on the basic data of the cell samples and the distribution characteristics of the sample data, performs data dimensionality reduction and extracts the key differences between samples, calculates the Euclidean distance between samples, and groups samples with similar immune response characteristics into the same group to obtain integrated feature data; The mutual relationship analysis module detects the mutual relationship between proteins and metabolites based on the integrated feature data, calculates the interaction strength between proteins and metabolites, identifies metabolic pathways related to immune response, reveals key protein-metabolite interactions in the immune response mechanism, and generates a functional network diagram; The marker validation module screens key protein and metabolite markers associated with the immune response based on the functional network diagram, verifies whether each candidate marker has an expression pattern of known immune function, detects changes in immune cells, evaluates performance during the immune response, confirms the association of each marker with key immune response processes, identifies the potential function of each marker in the immune response, and obtains marker validation data; Based on the validation marker data, the immune response monitoring module monitors the differential stages of the immune response through immune stimulation and target cell contact reaction, observes and records the interaction between target cells and CAR-NK cells, analyzes the changing trends of markers in the differential immune response stages and their relationship with CAR-NK cell response, determines the temporal dynamics of cell response and marker changes, and obtains the CAR-NK cell immune reactivity feature extraction results.

Images