Early gastric cancer lymph node metastasis risk prediction method and system, application and medium

Abstract

The invention relates to the technical field of methylation detection site detection, and particularly provides an early gastric cancer lymph node metastasis risk prediction method, system, application and medium, and the method comprises the following steps: obtaining a gastric cancer public data set containing a DNA methylation chip data set and an RNA sequencing data set, and carrying out sample screening, quality control and grouping processing to obtain eight quality control grouping samples; carrying out methylation and RNA difference analysis on the quality control grouped samples to obtain a related gene set of differential methylation sites and differential methylation regions and an RNA differential expression gene set; a gene set of differential methylation sites and an RNA differential expression gene set are integrated and screened to obtain eight target genes. The system comprises a sample acquisition module, a gene analysis module and a gene screening module. Target genes are analyzed and screened on the basis of database biological information, and methylation detection sites for histopathological specimens are screened in combination with lymph node metastasis positive and negative early gastric cancer histological sample verification.

Description

Technical Field

[0001] This invention relates to the field of methylation detection site technology, and in particular to a method, system, application and medium for predicting the risk of early gastric cancer lymph node metastasis. Background Technology

[0002] Gastric cancer is a malignant tumor that occurs in the stomach. Main symptoms include chronic indigestion, upper abdominal pain, and significant loss of appetite. Early-stage gastric cancer treatment generally involves endoscopic and surgical interventions, with the choice of treatment method based on the risk of lymph node metastasis. In clinical practice, the assessment of the risk of lymph node metastasis in early-stage gastric cancer is primarily based on pathological morphological evaluation and imaging assessment. Pathological morphological evaluation is based on current clinical and pathological guidelines, and pathological assessment indicators include tumor size, secondary ulceration, tumor differentiation and depth of invasion, vascular invasion, and surgical margins. Existing assessment methods are simple and easy to perform, requiring no expensive laboratory equipment, and are inexpensive and technically simple to operate. However, the accuracy of existing assessment methods is limited by the diagnostic experience of pathologists and radiologists / endoscopists, is subjective, and may be affected by factors related to the preparation of pathological slides. Imaging assessment includes endoscopic ultrasound, computed tomography (CT), and PET-CT, which can detect enlarged potentially metastatic lymph nodes. However, due to their non-invasive nature, they cannot definitively characterize lymph nodes and also have the potential for false positives and false negatives. Even excluding technical and evaluator-related factors, current pathological morphological assessments for lymph node metastasis risk still exhibit a certain degree of false positives and false negatives. Therefore, existing lymph node metastasis assessment techniques contain a risk of false positives and false negatives, potentially leading to unnecessary radical gastrectomy or endoscopic treatment in early-stage gastric cancer patients. With the development of molecular pathology, the introduction of molecular pathology techniques into the assessment of lymph node metastasis risk in early-stage gastric cancer is imperative.

[0003] DNA methylation is an important epigenetic phenomenon, involved in the development and progression of various tumors. Methylation detection has been applied in tumor-related diagnosis and treatment. For example, in 2018, the National Medical Products Administration (NMPA) approved peripheral blood Septin9 gene methylation for early screening of colorectal cancer; in 2022, BRCA1 and ESR1 methylation detection in breast cancer biopsy specimens received FDA Breakthrough Device Designation for guiding treatment plans. In 2020, the EU CE certified RASSFF1A and GSTP1 methylation detection for liver cancer tissue samples to guide surgical scope and assist in pathological diagnosis. In 2022, the NMPA approved a seven-in-one gastric cancer methylation detection kit for early gastric cancer screening; currently, there are no products that use methylation markers to assess the risk of lymph node metastasis in early gastric cancer. Currently, there is one patent for early lymph node assessment based on methylation detection, but it detects 20 methylation genes, which is a large number and results in high costs for clinical application. Methylation genes related to lymph node metastasis are obtained by analyzing cfDNA extracted from plasma of patients with and without lymph node metastasis (without specifying whether they are early-stage gastric cancer patients), rather than by directly screening through methylation detection of tumor tissue. The concentration of ctDNA (ng / ml) is lower than the concentration of DNA in tumor tissue (μg / ml), especially in early-stage cancers where ctDNA in peripheral blood extraction is often less than 1% of free DNA. Furthermore, ctDNA extracted from peripheral blood is highly fragmented and cannot fully reflect tumor heterogeneity.

[0004] Prior art 1, Chinese Patent Application No. 202510091223.1, discloses a model for predicting the risk of lymph node metastasis in early gastric cancer, its construction method, and its application. The model's input variables include gender, tumor location, maximum tumor diameter, pathological differentiation degree, pT stage, number of submitted lymph nodes, and lymphatic vessel invasion. The model's output value is the predicted probability of lymph node metastasis. The design of this model for predicting the risk of lymph node metastasis in early gastric cancer allows for accurate assessment of the risk in patients with early gastric cancer. Further, a model for predicting the number of metastatic lymph nodes is derived. Based on the predicted number of metastatic lymph nodes, a corrected pN staging (pNM) is proposed. This pNM staging can be applied to the diagnosis of early gastric cancer patients who cannot undergo adequate lymph node dissection for various reasons. While it helps clinicians assess and calculate the number of early-stage gastric cancer patients who have not submitted sufficient lymph nodes for examination, providing a new approach for accurate clinical evaluation, relying on traditional clinicopathological parameters, such as tumor diameter and differentiation degree, has two major drawbacks: strong subjectivity: indicators such as pathological grading are easily influenced by the observer's experience; and lack of molecular mechanisms: it cannot reflect the biological essence driving lymph node metastasis.

[0005] Prior art two, Chinese patent application number 202211280869.7, discloses a composition for assessing the risk of lymph node metastasis in early gastric cancer and its use. This composition detects the methylation levels of promoter regions of eight genes (CLGN, ZNF354C, LMO3, LDHB, ALDH1L1, PTPN13, PPP2R2B, and EYA4) in early gastric cancer tissue to assess the risk of lymph node metastasis in early gastric cancer (EGC), aiding clinicians in EGC risk stratification to determine whether endoscopic submucosal dissection (ESD) is necessary. While this method exhibits high sensitivity, specificity, and clinical concordance, and is highly significant for EGC risk stratification and treatment, it is a single-dimensional detection method, only detecting promoter methylation and neglecting functional verification at the transcriptome level; static analysis does not establish a regulatory relationship between methylation and gene expression.

[0006] Prior art three, Chinese patent application number 202310424950.6, discloses a method for predicting the risk of lymph node metastasis in digital pathological sections of primary gastric cancer. First, gastric cancer pathological sections are magnified at different magnifications, and features are extracted using a feature extractor based on transfer learning. Then, these features are input into a weakly supervised network to generate WSI-level feature representations and late-stage fused clinical data, including the patient's gender, age, tumor location, and Lauren classification. This yields predictions of lymph node metastasis risk at different magnifications. Finally, model fusion is used to integrate the results from different scales, ultimately predicting whether the patient has developed lymph node metastasis. While this invention aims to analyze routine tissue sections and clinical data of patients with primary gastric cancer to predict the risk of preoperative lymph node metastasis, it can help patients choose appropriate surgical methods and improve prognosis, as well as enhance physician efficiency, thus possessing significant importance and broad prospects. However, the image features are disconnected from molecular mechanisms, failing to explain the molecular basis of metastasis; the computational complexity is high, requiring the handling of multi-scale image fusion problems.

[0007] Current technologies 1, 2, and 3 suffer from key shortcomings such as fragmented mechanisms, limited detection dimensions, and insufficient clinical interpretability. Therefore, this invention provides a method, system, application, and medium for predicting the risk of early gastric cancer lymph node metastasis. Summary of the Invention

[0008] To achieve the above objectives, the present invention adopts the following technical solution: One aspect of the present invention provides a method for predicting the risk of early gastric cancer lymph node metastasis, comprising the following steps: We obtained a public dataset of gastric cancer containing DNA methylation microarray data and RNA sequencing data. After sample screening, quality control and grouping, we obtained 8 quality control group samples. The quality control group samples were subjected to methylation and RNA differential analysis to obtain the gene sets related to differentially methylated sites and differentially methylated regions, as well as the RNA differentially expressed gene sets. After integration and screening of gene sets with differentially methylated sites and differentially expressed RNA genes, eight target genes were obtained.

[0009] In one optional implementation, the process of obtaining eight quality control group samples includes the following steps: Early gastric cancer samples with T1a and T1b stages were screened from a public gastric cancer dataset. Based on microsatellite status quality control, 6 MSI-H samples were removed, and 15 MSS / MSI-L samples were output. Fifteen MSS / MSI-L samples underwent lymph node grouping and PCA quality control, and were grouped according to lymph node metastasis status: lymph node negative LNN and lymph node positive LNP; Outlier samples were removed by PCA analysis, resulting in 8 quality control samples, 4 lymph node-negative LNNs and 4 lymph node-positive LNPs.

[0010] In one optional implementation, the process of subjecting the quality control group samples to methylation and RNA differential analysis includes the following steps: The methylation difference analysis of the 8 quality control group samples yielded the gene set related to differentially methylated sites (DMP) and differentially methylated regions (DMR); The ChAMP software package in R was used to perform differential methylation analysis on 4 LNN and 4 LNP samples; 108 genes associated with differential methylation sites and 53 genes associated with differential methylation regions were output. Eight quality control group samples were matched with RNA-seq samples to obtain seven matching samples; the samples in the RNA-seq dataset were cross-matched with the methylation dataset to confirm the usable sample subset of RNA-seq data; the seven matching samples were output as a subset of samples in which both methylation and RNA data existed. RNA differential expression analysis was performed on 7 matched samples to obtain a set of RNA differentially expressed genes. The DESeq2 software package of R was used to perform differential expression analysis between the lymph node negative LNN and lymph node positive LNP groups to identify highly expressed genes in the lymph node positive LNP group, which correspond to the DMR hypomethylation pattern, and output the set of RNA differentially expressed genes.

[0011] In one optional implementation, the process of outputting 7 matching samples includes the following steps: The co-variant change patterns of methylation sites in eight accusation group samples were analyzed by evolutionary differential analysis. The methylation fluctuation characteristics of methylation site regions were extracted as network nodes to analyze the regulatory cascade relationships between sites and construct a dynamic response network. The output is the regulatory network skeleton that characterizes the specificity of the accusation group samples. The regulatory network backbone is mapped to the transcript space of RNA-seq data, and cross-modal correspondences are established based on the regulatory path matching principle; the potential matching coordinates in the RNA-seq dataset are output. Calculate the topological distance between the anchor position and the original network skeleton, verify the consistency of the sample distribution pattern within the group, and the LNN / LNP clustering pattern; output 7 matching samples that pass the consistency verification.

[0012] In one alternative implementation, the process of obtaining eight target genes includes the following steps: Input 53 hypomethylated genes from the LNP group, quantify the methylation inhibition intensity of each hypomethylated gene, construct a regulatory priority ranking based on the amplitude of methylation fluctuation, and output a gene vector characterizing epigenetic regulatory potential. Input differentially expressed genes that are significantly overexpressed in the LNP group, calculate the confidence interval of gene expression fold change, screen for expression patterns with sustained activation characteristics, and output gene vectors characterizing transcriptional activity; Establish a covariance space between gene vectors characterizing epigenetic regulatory potential and gene vectors characterizing transcriptional activity, identify gene regions with highly synchronized dual-track signals, and output the gene subspace of epigenetic-transcriptional co-regulation. Gene nodes with the highest gradients in the gene subspace were extracted, and lymph node metastasis status was used as a biological validation filter to output eight target genes: CLGN, ZNF354C, LMO3, LDHB, ALDH1L1, PTPN13, PPP2R2B, and EYA4.

[0013] In one alternative embodiment, methylation markers associated with early gastric cancer lymph node metastasis include any one or a combination of four of PPP2R2B, LMO3, ZNF354C, and CLGN.

[0014] In another aspect, the present invention provides an early gastric cancer lymph node metastasis risk prediction system based on the aforementioned early gastric cancer lymph node metastasis risk prediction method, comprising: The sample acquisition module is used to acquire a public dataset of gastric cancer containing DNA methylation microarray dataset and RNA sequencing dataset. After sample screening, quality control and grouping, 8 quality control group samples are obtained. The gene analysis module is used to perform methylation and RNA differential analysis on the quality control grouped samples to obtain the gene sets related to differentially methylated sites and regions and the RNA differentially expressed gene sets. The gene screening module integrates and screens gene sets with differentially methylated sites and differentially expressed RNA genes to obtain 8 target genes.

[0015] In another aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by at least one processor, implement the method for predicting the risk of early gastric cancer lymph node metastasis.

[0016] In another aspect, the present invention provides a methylation PCR detection product for early gastric cancer lymph node metastasis, comprising substances for detecting methylation markers PPP2R2B, LMO3, ZNF354C, and CLGN.

[0017] In another aspect, the present invention provides a methylation marker for use as a target in screening drugs for the prevention or treatment of early gastric cancer lymph node metastasis.

[0018] The present invention. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the method for predicting the risk of early gastric cancer lymph node metastasis provided in Embodiment 1 of the present invention; Figure 2 This is a flowchart illustrating the process of obtaining eight quality control grouped samples as provided in Embodiment 2 of the present invention. Figure 3 This is a flowchart illustrating the process of methylation and RNA differential analysis of quality control grouped samples provided in Embodiment 3 of the present invention. Figure 4 This is a flowchart illustrating the process of obtaining eight target genes as provided in Example 6 of the present invention; Figure 5 This is a schematic diagram of the early gastric cancer lymph node metastasis risk prediction system provided in Embodiment 5 of the present invention; Figure 6 A block diagram of the electronic device provided by the present invention; Figure 7 A block diagram of a computer-readable storage medium provided for this invention; Figures 8-15 This is a schematic diagram of the differential methylation analysis between LNN / LNP groups in the dataset provided by the present invention; Figure 16 This is a schematic diagram of the multi-factor combined ROC analysis provided by the present invention. Detailed Implementation

[0020] The technical solutions of the present invention will now be described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Hereinafter, the terms "first," "second," etc., are used for descriptive convenience only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0022] In this invention, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be a fixed mechanical connection, a detachable mechanical connection, or an integral part; or, "connection" can be a direct connection or an indirect connection through an intermediate medium. Furthermore, unless otherwise explicitly specified and limited, the term "coupling" should be interpreted broadly. For example, "coupling" can be a direct electrical connection, such as physical contact and electrical conduction between two components; it can also be understood as an electrical connection between different components in a circuit structure through physical lines capable of transmitting electrical signals, such as copper foil or wires on a printed circuit board (PCB), to transmit electrical signals; or, "coupling" can be an indirect electrical connection between two components through an intermediate medium; or, "coupling" can be an electrical connection between two components in a non-contact manner, such as an electrical connection between two components using capacitive coupling to transmit electrical signals.

[0023] In this embodiment of the invention, directional terms such as "up," "down," "left," and "right" may be defined relative to the orientation of the components shown in the accompanying drawings. It should be understood that these directional terms can be relative concepts, used for relative description and clarification, and can change accordingly depending on the orientation of the components in the accompanying drawings.

[0024] The inventive concept of this invention is as follows: By integrating DNA methylation microarray and RNA sequencing data, a two-dimensional epigenetic-transcriptome screening system is constructed. A multi-stage dimensionality reduction strategy (clinical staging screening, molecular subtype quality control, PCA outlier elimination, and differential analysis cross-validation) is employed to identify eight target gene combinations that synergistically regulate methylation and expression. Its core innovation lies in: identifying epigenetic driver genes through phase synchronization analysis of methylation fluctuations and gene expression oscillations; achieving interpretable transformation of high-dimensional data into clinical biomarkers based on manifold learning and dynamic field models; and ultimately establishing a lymph node metastasis risk assessment system with both molecular mechanism explanatory power and clinical predictive efficacy.

[0025] Example 1: like Figure 1 As shown, this embodiment of the invention provides a method for predicting the risk of early gastric cancer lymph node metastasis, comprising the following steps: Step S100: Obtain a public dataset of gastric cancer containing DNA methylation microarray dataset and RNA sequencing dataset. After sample screening, quality control and grouping, 8 quality control group samples are obtained. Step S200: Perform methylation and RNA differential analysis on the quality control grouped samples to obtain the gene set related to differential methylation sites and differential methylation regions and the RNA differentially expressed gene set; Step S300: The gene set of differentially methylated sites and the RNA differentially expressed gene set were integrated and screened to obtain 8 target genes.

[0026] In the above embodiments, this embodiment establishes a dual validation system of epigenetics (methylation) and transcriptomics (gene expression) by simultaneously analyzing DNA methylation microarray and RNA sequencing data. Cross-validation significantly improves the reliability of biomarkers, and the association analysis between differentially expressed methylation regions and differentially expressed genes can reveal the potential mechanisms by which epigenetic regulation affects gene expression. From initial quality control (8 grouped samples) to differential analysis, and finally to the identification of 8 target genes, a strict dimensionality reduction logic is presented. This stepwise screening retains sufficient data dimensions while improving specificity through gradual convergence, conforming to the principle of broad entry and strict exit in biomarker development. The final combination of 8 target genes may constitute key nodes in the methylation-expression co-regulatory network; compared with single biomarkers, multi-gene panels can better reflect the complex pathway interactions involved in tumor metastasis, such as epigenetic modifications driving malignant phenotypes and metastasis-related gene expression dysregulation, providing a more stable predictive model basis for clinical practice.

[0027] In summary, this embodiment establishes a predictive system with molecular mechanism explanation through cross-mining of epigenetics and transcriptomics. The selected target gene combinations contain both methylation regulatory information and are associated with changes in expression levels, providing a scientifically sound and practical candidate biomarker group for subsequent development of diagnostic reagents or prognostic models. This invention screens target genes based on database bioinformatics analysis and further validates the results using early gastric cancer histological samples with positive and negative lymph node metastasis, identifying methylation detection sites suitable for histopathological specimens.

[0028] Example 2: like Figure 2 As shown, based on Example 1, the process of obtaining 8 quality control group samples in step S100 of this embodiment of the invention includes the following steps: Step S101: Screen early gastric cancer samples of T1a and T1b stages from the public gastric cancer dataset. Based on microsatellite status quality control: remove 6 MSI-H samples; output 15 MSS / MSI-L samples. Step S102: The 15 MSS / MSI-L samples were grouped by lymph node and PCA quality control, and grouped according to lymph node metastasis status: lymph node negative LNN and lymph node positive LNP; Step S103: Outlier samples were removed by PCA analysis, resulting in 8 quality control samples, 4 lymph node-negative LNNs and 4 lymph node-positive LNPs.

[0029] In the above embodiments, this embodiment employs a four-layer progressive design: staging screening, molecular marker screening, clinicopathological grouping, and data noise reduction. This sequentially excludes: non-early-stage gastric cancer samples (T1a / T1b staging screening), subtypes with significant differences in molecular mechanisms (MSI-H exclusion), groups with significantly different clinical prognoses (lymph node status grouping), and technical noise interference (PCA outlier removal). The final eight samples exhibit highly homogeneous biological characteristics. Integrating three key dimensions—clinical staging (T1a / T1b), molecular characteristics (MSS / MSI-L stability), and metastatic potential (LNN / LNP grouping)—the cross-validation strategy maximizes the clinical relevance of subsequent analysis results. The final output of four LNN and four LNP samples satisfies both inter-group comparability (1:1 sample size matching) and intra-group consistency (removal of intra-group heterogeneity through PCA), laying a statistical foundation for subsequent differential analysis.

[0030] In summary, this embodiment utilizes multimodal data fusion to construct a pure sample set with distinct molecular characteristics and contrasting clinical phenotypes in early-stage gastric cancer, making it particularly suitable for exploring molecular mechanisms related to lymph node metastasis. All technical choices (such as retaining MSI-L while excluding MSI-H) reflect strict control over the molecular heterogeneity of gastric cancer.

[0031] Example 3: Based on Example 2, the PCA quality control process in step S102 provided in this embodiment of the invention includes the following steps: Step S1021: Using nonlinear dimensionality reduction, the methylation sites of each MSS / MSI-L sample are transformed into coordinate points in two-dimensional space through manifold learning technology; the direction of maximum variance of the data is automatically identified through eigenvalue decomposition, the optimal low-dimensional embedding of the topological relationship between samples is preserved, and the spatial distribution mapping of all MSS / MSI-L samples in the two-dimensional plane is output. Step S1022: Apply density-based clustering to quantify the relative distance between each sample point and the core region within the group, calculate the local density and the global optimum of the distance between the sample points within the group; dynamically establish a spatial distribution probability model of the samples within the group, automatically correct the influence of feature correlation through Mahalanobis distance, and output the outlier probability score for each sample. Step S1023: Set an adaptive outlier threshold, remove samples whose outlier probability scores exceed the adaptive outlier threshold, and automatically optimize the cutoff point based on extreme value theory to maintain the homogeneity of spatial distribution within the group.

[0032] In the above embodiments, this embodiment achieves precise quality control of high-dimensional methylation data; when mapping high-dimensional methylation sites to two-dimensional space, maximum variance direction decomposition and topological relationship preservation are used to avoid sample distribution distortion caused by neglecting local structure in linear PCA; ensuring that the spatial distribution of MSS / MSI-L subgroups can truly reflect their epigenetic heterogeneity, providing a reliable low-dimensional basis for subsequent outlier detection. By calculating the Mahalanobis distance (corrected feature correlation) between sample points and the core region within the group, the outlier tendency of each point is dynamically evaluated, avoiding the neglect of covariance structure by Euclidean distance; the outlier score quantifies the degree to which the sample deviates from the population distribution, providing a mathematical basis for adaptive thresholding. Based on the distribution characteristics of the outlier score, the threshold cutoff point is automatically determined to avoid manually setting bias; the removed samples have statistically significant outliers, while the retained sample set maintains intragroup consistency at the epigenetic level, reducing noise interference in subsequent analyses.

[0033] In summary, this embodiment features extremely high dimensionality of methylation sites, which traditional PCA may lose due to the loss of nonlinear structure. Manifold learning combined with density clustering is better suited to this type of data. Mahalanobis distance correction and adaptive thresholding avoid overfitting or underfitting of different datasets with fixed thresholds. This ensures that subsequent differential methylation analysis reflects only true biological variation, rather than technical outliers. By combining dimensionality reduction, clustering, and statistical inference into a closed-loop system, a highly reliable MSS / MSI-L sample set is ultimately output.

[0034] Example 4: Based on Example 3, the process of converting coordinate points in two-dimensional space in step S1021 provided in this embodiment of the invention includes the following steps: Step S10211: Analyze the nonlinear coupling relationship between methylation sites, generate a dynamic feature map, identify stable cross-sample associated pathways through latent variable decomposition, and output the basic dimensional framework characterizing molecular covariation patterns. Step S10212: Project the samples onto the affine space defined by the skeleton, and unfold the sample distribution topology along the direction of maximum discrimination; optimize the spatial curvature parameters based on the direction of maximum variance of the identification, and maintain the original network connectivity through iterative projection correction; output the relative position mapping of the samples in the compressed space; Step S10213: Obtain the position entropy value of each sample in the reconstruction space, generate unique spatial coordinates based on entropy stability, use the discrimination direction of the skeleton as the coordinate axis reference, and fuse the distribution density of samples within the group as the position constraint; output the two-dimensional spatial fingerprint coordinates of 15 samples.

[0035] In the above embodiments, this embodiment utilizes the nonlinear dependence between methylation sites to construct a biologically meaningful fundamental dimension; it transforms high-dimensional correlations into a resolvable spatial structure through affine transformation; and it quantifies the topological uniqueness of samples within the population distribution using positional entropy values. The implicit correlations between molecular features are transformed into measurable relationships in geometric space, and the spatial localization of samples is achieved through a dynamically generated discriminative dimensional framework, establishing a computable topological foundation for subsequent outlier detection. This process relies entirely on the intrinsic structure of the data, avoids pre-set spatial assumptions, and ensures the objectivity of subsequent PCA quality control.

[0036] Example 5: like Figure 3 As shown, based on Example 1, the process of performing methylation and RNA differential analysis on the quality control grouped samples in step S200 of this embodiment of the invention includes the following steps: Step S201: The methylation difference analysis of the 8 quality control group samples was performed to obtain the gene set related to differentially methylated sites (DMP) and differentially methylated regions (DMR); The ChAMP software package in R was used to perform differential methylation analysis on 4 LNN and 4 LNP samples; 108 genes associated with differential methylation sites and 53 genes associated with differential methylation regions were output. Step S202: The 8 quality control group samples are matched with RNA-seq samples to obtain 7 matching samples; the samples in the RNA-seq dataset are cross-matched with the methylation dataset to confirm the sample subset of RNA-seq data that is available; the 7 matching samples are output as a sample subset in which both methylation and RNA data exist. Step S203: RNA differential expression analysis was performed on 7 matched samples to obtain a set of RNA differentially expressed genes; differential expression analysis between lymph node negative LNN and lymph node positive LNP groups was performed using the DESeq2 software package of R (statistical computing and data visualization programming language) to identify highly expressed genes in the lymph node positive LNP group, which correspond to the DMR hypomethylation pattern, and the set of RNA differentially expressed genes was output.

[0037] In the above embodiments, this embodiment employs a three-tiered progressive design involving methylation (DMP / DMR), RNA-seq (differentially expressed genes), and epigenetic-transcriptional co-validation: The ChAMP package is used to identify epigenetic regulatory signals (108 DMP genes + 53 DMR genes), suggesting that lymph node metastasis (LNP) may be driven by methylation modifications of specific genes. Seven pairs of samples are retained through cross-omics data matching to ensure sample consistency for subsequent analyses and eliminate batch effect interference. DESeq2' is used to screen for differentially expressed genes at the expression level and cross-validate with methylation results (e.g., LNP high-expression genes correspond to DMR low-methylation), identifying epigenetic-transcriptional co-regulatory targets. Genes corresponding to DMR low-methylation regions in the LNP group show high expression (e.g., activation after demethylation of tumor suppressor genes), directly correlated with the regulatory role of epigenetic modifications on transcription. Simple methylation or RNA differential analysis may contain a large amount of noise, while co-occurrence signals (e.g., DMR low-methylation and high RNA expression) are more likely to reflect the true driving events. Seven RNA-seq samples were rigorously matched from eight methylated samples to ensure comparability of epigenetic and transcriptional data from the same patients. Methylation analysis was performed using ChAMP, an R package specifically designed for processing methylation microarray data, which includes the DMRcate algorithm. RNA-seq analysis was performed using DESeq2, the gold standard for differential expression based on a negative binomial distribution, to ensure reproducibility of the results.

[0038] In summary, this embodiment identifies epigenetic driver genes related to lymph node metastasis, reveals potential epigenetic regulatory mechanisms, and provides a multi-omics evidence chain for biomarker discovery.

[0039] Example 6: Based on Example 5, the process of outputting 7 matching samples in step S202 of this embodiment of the invention includes the following steps: Step S2021: Input the synergistic change patterns of methylation sites from the evolutionary differential analysis of 8 accusation group samples, extract the methylation fluctuation characteristics of methylation site regions as network nodes, analyze the regulatory cascade relationship between sites, and construct a dynamic response network; output the regulatory network skeleton characterizing the specificity of the accusation group samples. Step S2022: Map the regulatory network backbone to the transcript space of the RNA-seq data, establish cross-modal correspondences based on the regulatory path matching principle; output the potential matching coordinates in the RNA-seq dataset; Step S2023: Calculate the topological distance between the anchor position and the original network skeleton, verify the consistency of the sample distribution pattern within the group, and the LNN / LNP clustering pattern; output 7 matching samples that pass the consistency verification.

[0040] In the above embodiments, this embodiment establishes a sample-specific regulatory network based on the methylation synergistic change pattern obtained from DMR analysis; utilizes the epigenetic association principle of methylation-expression to achieve identifier-free mapping in the data space; and confirms matching reliability by inheriting the spatial characteristics of PCA quality control through the consistency of sample distribution in molecular space. Methylation data is parsed into gene regulatory network features, and a biologically driven (rather than identifier-driven) sample localization system is established in RNA-seq space through epigenetic regulatory rules. Finally, matching samples are obtained through spatial topological consistency verification. This method overcomes the limitations of conventional identifier matching, ensuring the biological interpretability of the matching results.

[0041] Example 7: Based on Example 6, the process of calculating the topological distance between the anchoring position and the original network skeleton in step S2023 of this embodiment of the invention includes the following steps: Step S20231: Analyze the core regulatory paths in the regulatory network skeleton as the basic trajectory axis, quantify the positional offset features of each sample on the trajectory axis; output the methylation spatial trajectory vector that represents the sample specificity; Step S20232: Reconstruct the expression pattern in the same spatial dimension defined by the trajectory axis, and map RNA-seq data to transcript dynamics defined by the regulatory pathway; output the transcript spatial trajectory vector isomorphic to the methylation trajectory; Step S20233: Obtain the covariance angle between the methylation trajectory and the transcript trajectory, quantify the convergence strength of the two tracks at key regulatory nodes; output the topological consistency score for each sample. Step S20234: Set the dynamic acceptance threshold: mean score within the group + 2 × standard deviation, retain samples with scores higher than the dynamic acceptance threshold, and output 7 verified matching samples.

[0042] In the above embodiments, this embodiment establishes a quantifiable spatial dimension based on the core path of the regulatory network backbone, ensuring the comparability of methylation and expression data in the same biological space. It captures the consistency of regulation at the molecular level through differences in trajectory directions, inheriting the adaptive principle of previous quality control to ensure the robustness of the results. The molecular regulatory network is transformed into a computable spatial trajectory system, and the convergence degree of the methylation-expression dual tracks in the same biological space verifies the reliability of sample matching; it overcomes the limitations of traditional identifier matching, achieving biologically driven verification based on regulatory consistency.

[0043] Example 6: like Figure 4 As shown, based on Example 1, the process of obtaining 8 target genes in step S300 of this embodiment of the invention includes the following steps: Step S301: Input 53 LNP group hypomethylated genes, quantify the methylation inhibition intensity of each hypomethylated gene, construct a regulatory priority ranking based on the methylation fluctuation amplitude, and output gene vectors characterizing epigenetic regulatory potential; Step S302: Input differentially expressed genes that are significantly overexpressed in the LNP group, calculate the confidence interval of gene expression fold change, screen expression patterns with sustained activation characteristics, and output gene vectors characterizing transcriptional activity; Step S303: Establish a covariance space between gene vectors characterizing epigenetic regulatory potential and gene vectors characterizing transcriptional activity, identify gene regions with highly synchronized dual-track signals, and output the gene subspace of epigenetic-transcriptional co-regulation. Step S304: Extract the gene node with the highest gradient in the gene subspace, apply the lymph node metastasis status as a biological validation filter, and output 8 target genes: CLGN, ZNF354C, LMO3, LDHB, ALDH1L1, PTPN13, PPP2R2B, and EYA4.

[0044] In the above embodiments, this embodiment quantifies methylation regulation intensity and expression activity into comparable dimensions, establishes a biological consistency metric through dual-track signal synchronicity, locates key genes based on signal intensity peaks in the covariance space, and uses lymph node metastasis status as the final screening filter. By constructing a covariance space of methylation regulation and gene expression, gene regions with highly synchronized epigenetic alterations and transcriptional activity are identified, and finally, eight target genes with biological consistency are screened based on signal gradient intensity and clinical relevance. This overcomes the limitations of conventional intersection analysis and achieves three-dimensional integrated screening based on regulatory dynamics.

[0045] Example 7: Based on Example 6, the process of outputting the gene subspace of epigenetic-transcriptional co-regulation in step S303 of the present invention includes the following steps: Step S3031: The regulatory intensity of the gene vector is converted into a three-dimensional field gradient, and the dynamic inhibition wave propagation mode is unfolded along the DNA topological structure axis to output and characterize the molecular field model of the epigenetic regulation dynamics. Step S3032: Map the expression intensity of the expression pattern with sustained activation characteristics to a transcriptional wave radiation field, establish a propagation model in which the expression amplitude decays with the regulatory distance, and output a molecular field model that characterizes transcriptional activity. Step S3033: Calculate the phase overlap between the epigenetic regulatory dynamic molecular field and the transcriptional activity molecular field, identify the resonance region where the field strength fluctuation frequency is synchronized, and output the set of gene coordinates that resonate with the dual-track signal. Step S3034: Extract the continuous oscillation bands in the double-track signal resonance region, verify signal convergence along the chromosome topological domain, and output the gene subspace that is co-regulated by epigenetic-transcriptional factors.

[0046] In the above embodiments, this embodiment transforms discrete gene vectors into a continuous field model to capture regulatory dynamics; it quantifies the consistency of epigenetic-transcriptional regulation through fluctuation frequency synchronicity; and it verifies biological rationality based on chromosome spatial location. By constructing a molecular dynamic field model of epigenetic regulation and transcriptional activity, the phase synchronization characteristics of dual-track signals are captured in three-dimensional biological space, and finally, gene subspaces with regulatory consistency are identified based on resonance intensity and spatial continuity. This breaks through the traditional correlation analysis framework and achieves four-dimensional integration of three-dimensional spatial and temporal fluctuations based on molecular dynamics.

[0047] Example 8: Based on Example 7, the process of outputting the set of gene coordinates resonating with the dual-track signal in step S3033 of this embodiment of the invention includes the following steps: Step S30331: Decompose the field intensity gradient into discrete wave units along the DNA topological axis. Quantify the periodic phase angle of the methylation wave within each discrete wave unit, and output the spatiotemporal phase feature matrix of the apparent field; Step S30332: Decompose the transcribed wave radiation mode in the same topological coordinate system, record the phase shift of the expressed amplitude fluctuation, and output the spatiotemporal phase feature matrix of the transcribed field; Step S30333: Align the two phase matrices at the spatial location of the chromosome, obtain the phase angle cosine similarity of the corresponding sites, and output the phase synchronization oscillation spectrum at the genome scale; identify the continuous high signal region in the synchronization oscillation spectrum, extract the resonance band spanning multiple DMR regions, and output the gene coordinate set of dual-track signal resonance.

[0048] In the above embodiments, this embodiment transforms the continuous molecular field into discrete phase units to achieve precise quantization; captures the dynamics of epigenetic-transcriptional regulation through fluctuation periodic features; directly compares the phase relationship between the two fields at the chromosomal spatial location; and requires the high synchronization signal to span at least two adjacent DMR regions. By decomposing the dynamic fluctuations of the molecular field into a phase feature matrix, the phase synchronization of the two fields is directly calculated in the chromosomal topological space, and finally, the coordinates of the resonant gene are identified based on the continuous high synchronization signal, breaking through the traditional correlation analysis and achieving accurate resonance detection based on spatiotemporal fluctuation features.

[0049] Example 9: Based on Example 8, the process of identifying continuous high-signal regions in the synchronous oscillation spectrum in step S30333 of the present invention includes the following steps: Step S303331: Based on the DNA topological structure axis coordinate system, the spatiotemporal phase feature matrix of the apparent field and the spatiotemporal phase feature matrix of the transcription field are matched point-to-point in the chromosome spatial position. Each discrete wave unit corresponds to a base position interval in the matrix. The point-to-point matching process is directly mapped through coordinate index. The methylation phase angle of the apparent field and the expression phase offset of the transcription field are aligned at the same spatial site. Step S303332: At each spatial site of point-to-point matching, dynamic coordination is calculated based on the methylation cycle phase angle of the apparent field phase feature matrix and the expression phase shift of the transcription field phase feature matrix. The consistency of phase angle direction is evaluated: the apparent phase angle and transcription phase shift of each site are regarded as the instantaneous angle of the wave trajectory. The sine or cosine projection of the phase difference between the two in the cycle is evaluated, and a coordination coefficient is output, which represents the synchronization strength of the dual-track signal at the site. The value range is -1 to 1, where 1 represents complete synchronization. The coordination coefficients of all sites are arranged in chromosomal position order to form a one-dimensional vector. Step S303333: Expand the coordination coefficient vector along the DNA topological axis to construct a genome-wide oscillation spectrum. With spatial location as the horizontal axis and coordination coefficient as the vertical axis, it presents the synchronous fluctuation mode of the dual-track signal between the apparent field and the transcriptional field. The generation of the oscillation spectrum uses vector data and retains the discrete unit resolution of the original phase feature matrix. High value regions or coordination coefficients close to 1 indicate that methylation fluctuations and expression amplitudes oscillate in high phase with each other in specific chromosomal regions. Step S303334: On the output phase-synchronized oscillation spectrum, scan continuous spatial segments where the coordination coefficient is consistently higher than the dynamic threshold. The threshold is adaptively set based on the distribution of the whole genome coordination coefficient. Continuous spatial segments are defined as resonance bands, which span multiple differential methylation fluctuation regions and have a length that covers at least two adjacent DMRs. The extraction of resonance bands is achieved through spatial continuity detection: adjacent high coordination coefficient sites are merged into a single segment, and isolated peaks are ignored. Step S303335: Map the identified resonance bands to genome annotation, extract the gene start and end sites covered by each resonance band through coordinate transformation; the output is a set of gene coordinates, which only includes gene locations where the resonance bands completely or partially overlap, ensuring that each coordinate corresponds to a strong synchronous oscillation region of the dual-track signals of the epigenetic and transcribed fields.

[0050] In the above embodiments, this embodiment achieves a precise analysis of the dynamic coupling relationship between epigenetic modification and transcriptional activity from a molecular dynamics perspective. By mapping the DNA topological axis, the periodic oscillation signal of the epigenetic field and the expression fluctuations of the transcriptional field are precisely anchored at base resolution. The alignment method based on physical spatial indexing establishes a unified molecular spatiotemporal coordinate system for both types of biological signals, ensuring that each data point in subsequent analysis corresponds to a precise chromatin microenvironment. The phase difference projection algorithm innovatively transforms the periodic fluctuations of epigenetic modification (phase angle) and the phase shift of transcriptional activity into vector operations. Sine / cosine projection simplifies complex biomolecular interactions into calculable coordination coefficients, whose value range (-1 to 1) characterizes both synchronization intensity and distinguishes between in-phase and out-of-phase oscillation modes, providing a standardized metric for subsequent resonance band identification. Data integration based on discrete unit resolution preserves the molecular-scale information of the original phase matrix, and the generated oscillation spectrum is essentially an "epiogenetic-transcriptional coupling heatmap." The high-value regions with coordination coefficients close to 1 are equivalent to marking physical sites on chromatin where two types of regulatory signals coordinate precisely like "gears meshing." The combined application of adaptive thresholding and spatial continuity detection enables the system to automatically filter noise and lock onto biologically significant continuous resonance bands. The design pays special attention to long-range coordination patterns spanning multiple differentially expressed regions (DMRs), suggesting that the captured regulatory units may be at the higher-order domain level of chromatin. The final mapping to gene coordinates establishes a correlation between molecular dynamics features and functional genomic annotation; the output gene set is essentially a group of candidate targets regulated by epigenetic-transcriptional dual-field synchronous oscillations, providing a spatially precise molecular evidence chain for the analytical modification of gene expression regulation mechanisms.

[0051] In summary, this embodiment re-integrates traditionally fragmented epigenetic and transcriptomic data through physical phase coordination analysis to form an interpretable dynamic regulatory relationship map; the analysis method based on molecular motion synchronicity can reveal the real-time regulatory effect of chemical modifications on gene expression better than static correlation studies.

[0052] Example 10: Based on Example 9, the process of extracting the gene start and termination sites covered by each resonance band in step S303335 of this embodiment of the invention includes the following steps: Step S3033351: Directly read the set of resonance bands, each resonance band containing its position range from the start site to the end site on the DNA topological axis; Step S3033352: Input the resonance band coordinates into the pre-constructed genome spatial topology framework, which is homologous to the DNA topological axis but divided into basic regions based on gene functional units. Each gene is defined as a spatial continuum with a clear start site to an end site. The conversion process is achieved through coordinate interval projection: the start and end positions of the resonance band are directly mapped to the linear coordinate axes of the genome spatial topology framework. Step S3033353: Within the genomic spatial topology framework, detect the spatial overlap between each resonance band coordinate interval and all gene boundary intervals. Based on the relative positions of the interval endpoints, determine if a resonance band start site is less than or equal to a gene termination site, and if the resonance band termination site is greater than or equal to the gene start site, then a coverage relationship exists, including complete coverage and partial overlap. Output a set of coverage relationship pairs, each pair containing a resonance band identifier and the boundary coordinates of the covered gene. Step S3033354: Traverse all coverage pairs and extract the start and end sites of each covered gene; perform deduplication for cases where the same gene is covered by multiple resonance bands, retaining only a unique coordinate record for each gene; output as a set of gene coordinates, with each entry containing the precise location range of the gene on the DNA topological axis. Step S3033355: Based on the definition of resonance bands, if the coordination coefficient is consistently higher than the dynamic threshold and spans multiple DMRs, each gene coordinate in the output gene coordinate set automatically inherits the synchronous oscillation attribute of the resonance band it covers; finally, the gene coordinate set retains only genes that have spatial overlap with the resonance band, located within the strong synchronous oscillation region of the apparent field and the transcriptional field.

[0053] In the above embodiments, this embodiment directly uses the projection of the DNA topological structure axis coordinate system onto the genome spatial topological framework, avoiding gene ID conversion or external annotation; it only relies on the positional relationship of the endpoints of the coordinate interval to determine overlap, without involving semantic information such as gene function or name; the synchronous oscillation attribute of the gene coordinate set is directly passed by the resonance band definition, without the need for recalculation; the input resonance band spatial coordinates are loaded with only the resonance band coordinates that are relied upon, and the output strong synchronous oscillation region verification is generated entirely by the gene coordinate set resolved by the coverage relationship, without the intervention of external data.

[0054] Example 11: like Figure 5 As shown, based on Examples 1-10, the early gastric cancer lymph node metastasis risk prediction system provided in this embodiment of the invention includes: Sample acquisition module 1 is used to acquire a public dataset of gastric cancer containing DNA methylation microarray dataset and RNA sequencing dataset. After sample screening, quality control and grouping, 8 quality control group samples are obtained. Gene analysis module 2 is used to perform methylation and RNA differential analysis on the quality control grouped samples to obtain the gene set related to differential methylation sites and differential methylation regions and the RNA differential expression gene set; Gene screening module 3 integrates and screens gene sets with differentially methylated sites and differentially expressed RNA genes to obtain 8 target genes.

[0055] In the above embodiments, the sample acquisition module 1 completes the standardization of multi-omics data (DNA methylation microarray and RNA sequencing), providing homogenized input data for subsequent analysis. The gene analysis module 2 captures transfer-related features at the epigenetic regulation and functional gene expression levels by analyzing methylation modifications and gene expression changes in parallel. The gene screening module 3 cross-validates differentially expressed genes with methylation and expression differences, ultimately outputting a combination of eight target genes that possess both epigenetic regulatory basis and functional expression changes. The cascaded three modules realize a complete transformation process from raw data to clinically usable biomarkers, and the established prediction system features multi-omics data collaborative analysis.

[0056] Figure 6 A block diagram of an exemplary electronic device suitable for implementing embodiments of the present invention is shown.

[0057] The electronic device may include a central processing unit / microprocessor / main control chip, etc. 4; and a storage medium 5, coupled to the central processing unit / microprocessor / main control chip, etc. 4, and storing computer-executable instructions therein for performing the steps of various methods of embodiments of the present invention when executed by the processor.

[0058] The central processing unit / microprocessor / main control chip, etc., can include, but are not limited to, one or more processors or microprocessors.

[0059] Storage medium 5 may include, but is not limited to, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, computer storage media (e.g., hard disk, floppy disk, solid-state drive, removable disk, CD-ROM, DVD-ROM, Blu-ray disc, etc.).

[0060] In addition, the electronic device may also include (but is not limited to) a data bus 6, an input / output bus / external bus / device bus 7, a display 8, and input / output devices 9 (e.g., keyboard, mouse, speaker, etc.).

[0061] The central processing unit / microprocessor / main control chip, etc. 4 can communicate with external devices (8, 9, etc.) via I / O bus 7 through wired or wireless network (not shown).

[0062] The storage medium 5 may also store at least one computer-executable instruction for performing the steps of various functions and / or methods in the embodiments described herein when the central processing unit / microprocessor / main control chip, etc., 4 is running.

[0063] In one embodiment, the at least one computer-executable instruction may also be compiled into or comprise a software product, wherein one or more computer-executable instructions are executed by a processor to perform the steps of the various functions and / or methods in the embodiments described herein.

[0064] Figure 7 A schematic diagram of a computer-readable storage medium according to an embodiment of the present invention is shown.

[0065] like Figure 7 As shown, the non-transitory computer-readable storage medium 11 stores instructions, such as computer-readable instructions 10. When the computer-readable instructions 10 are executed by a processor, the various methods described above can be performed. The non-transitory computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Non-transitory non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. For example, the non-transitory computer-readable storage medium 11 can be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions 10 stored on the computer-readable storage medium 11, the various methods described above can be performed.

[0066] In this embodiment, methylation biomarkers associated with early gastric cancer lymph node metastasis include PPP2R2B, LMO3, ZNF354C, and CLGN, comprising any one or a combination of four. The methylation PCR detection product for early gastric cancer lymph node metastasis includes substances for detecting the methylation biomarkers described above; the use of substances for detecting methylation biomarkers in the preparation of the detection product for early gastric cancer lymph node metastasis; and the use of methylation biomarkers as targets in screening drugs for the prevention or treatment of early gastric cancer lymph node metastasis.

[0067] The following analysis uses clinical samples to verify differentially expressed genes. The candidate gene set (8 genes in total) mined from the public STAD dataset for gastric cancer was validated on a real clinical sample set. Methylation differences between the LNN / LNP groups were analyzed, with the change in CT value (ΔCT) between groups used as the measure of methylation difference. PPP2R2B methylation expression showed a significant difference between the LNN and LNP groups (p value < 0.05). PPP2R2B methylation expression was significantly downregulated in the LNP group compared to the LNN group. Other genes with significant differences, from largest to smallest, were LMO3, ZNF354C, and CLGN, but no significant differences were observed between the groups. Figures 8-15 ).

[0068] ROC analysis was performed using clinical features with significant intergroup differences (tumor invasion depth, lymphovascular invasion) and the top 4 target genes with large differences (PPP2R2B, LMO3, ZNF354C, CLGN; methylation levels were categorized as high or low using the median as a threshold). The combined area under the ROC curve (AUC) was 0.83, and the optimal threshold for this prediction model was -0.139, at which point the sensitivity reached 85.71% and the specificity reached 73.33%. Figure 16 ).

[0069] In existing technologies, pathological imaging assessment techniques are limited by the pathologist's experience, potentially leading to observer inconsistency; both pathological and imaging assessments are qualitative observations, lacking quantitative indicators; current early-stage cancer lymph node DNA methylation detection targets up to 20 genes, resulting in high detection costs; and the target genes screened for early-stage cancer lymph node DNA methylation detection are based on plasma cfDNA, which may not be applicable to tissue samples. This invention newly screens four target genes for methylation detection, classifying the methylation levels of these target genes in tumor tissue as high or low using methylation PCR; this can improve the sensitivity and specificity of predicting early gastric cancer lymph node metastasis.

[0070] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0071] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0072] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0073] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the methods of the various embodiments of this invention through a computer device (which may be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0074] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the risk of lymph node metastasis in early gastric cancer, characterized in that, Includes the following steps: A public dataset for gastric cancer containing DNA methylation microarray data and RNA sequencing data was obtained. After sample screening, quality control, and grouping, eight quality control group samples were obtained. The quality control group samples were subjected to methylation and RNA differential analysis to obtain gene sets related to differentially methylated sites and differentially methylated regions, as well as RNA differentially expressed gene sets. The gene sets of differentially methylated sites and RNA differentially expressed gene sets were integrated and screened to obtain eight target genes.

2. The method for predicting the risk of early gastric cancer lymph node metastasis as described in claim 1, characterized in that, The process of obtaining 8 quality control group samples includes the following steps: screening early gastric cancer samples of T1a and T1b stages from a public gastric cancer dataset, and quality control based on microsatellite status: removing 6 MSI-H samples; outputting 15 MSS / MSI-L samples; the 15 MSS / MSI-L samples were grouped by lymph node and PCA quality control, and grouped according to lymph node metastasis status: lymph node negative LNN and lymph node positive LNP; outlier samples were removed by PCA analysis, resulting in 8 quality control samples, 4 lymph node negative LNN and 4 lymph node positive LNP.

3. The method for predicting the risk of early gastric cancer lymph node metastasis as described in claim 1, characterized in that, The process of performing methylation and RNA differential analysis on the quality control group samples includes the following steps: Methylation differential analysis was performed on 8 quality control group samples to obtain differentially methylated sites (DMP) and differentially methylated regions (DMR) related gene sets; methylation differential analysis was performed on 4 LNN and 4 LNP samples using the R ChAMP package; 108 genes related to differentially methylated sites and 53 genes related to differentially methylated regions were output; RNA-seq sample matching was performed on the 8 quality control group samples to obtain 7 matching samples; the samples in the RNA-seq dataset were cross-matched with the methylation dataset to confirm the sample subsets for which RNA-seq data was available; the 7 matching samples, a subset of samples where both methylation and RNA data were present, were output; RNA differential expression analysis was performed on the 7 matching samples to obtain the RNA differentially expressed gene set; differential expression analysis was performed between the lymph node-negative LNN and lymph node-positive LNP groups using the R DESeq2 package to identify highly expressed genes in the lymph node-positive LNP group, corresponding to the DMR low methylation pattern, and the RNA differentially expressed gene set was output.

4. The method for predicting the risk of early gastric cancer lymph node metastasis as described in claim 3, characterized in that, The process of outputting 7 matched samples includes the following steps: inputting the methylation site co-change pattern of 8 accusation group samples with evolutionary differential analysis, extracting the methylation fluctuation characteristics of methylation site regions as network nodes, analyzing the regulatory cascade relationship between sites, and constructing a dynamic response network; Output the regulatory network backbone that characterizes the specificity of the grouped samples; map the regulatory network backbone to the transcript space of RNA-seq data, and establish cross-modal correspondences based on the principle of regulatory path matching; Output the potential matching coordinates in the RNA-seq dataset; calculate the topological distance between the anchoring position and the original network skeleton; verify the consistency of the sample distribution pattern within the group and the LNN / LNP clustering morphology; output 7 matching samples that pass the consistency verification.

5. The method for predicting the risk of early gastric cancer lymph node metastasis as described in claim 1, characterized in that, The process of obtaining 8 target genes includes the following steps: Inputting 53 hypomethylated genes from the LNP group, quantifying the methylation inhibition intensity of each hypomethylated gene, constructing a regulatory priority ranking based on methylation fluctuation amplitude, and outputting gene vectors characterizing epigenetic regulatory potential; inputting differentially expressed genes significantly overexpressed in the LNP group, calculating confidence intervals for gene expression fold changes, screening expression patterns with sustained activation characteristics, and outputting gene vectors characterizing transcriptional activity; establishing a covariance space between gene vectors characterizing epigenetic regulatory potential and gene vectors characterizing transcriptional activity, identifying gene regions with highly synchronized dual-track signals, and outputting a gene subspace for epigenetic-transcriptional co-regulation; extracting the gene nodes with the highest gradients in the gene subspace, applying lymph node metastasis status as a biological validation filter, and outputting 8 target genes: CLGN, ZNF354C, LMO3, LDHB, ALDH1L1, PTPN13, PPP2R2B, and EYA4.

6. The method for predicting the risk of early gastric cancer lymph node metastasis as described in claim 1, characterized in that, Methylation markers associated with early gastric cancer lymph node metastasis, comprising any one or a combination of four of PPP2R2B, LMO3, ZNF354C, and CLGN.

7. A system for predicting the risk of early gastric cancer lymph node metastasis, based on the method for predicting the risk of early gastric cancer lymph node metastasis as described in any one of claims 1 to 6, characterized in that, include: The sample acquisition module is used to acquire a public dataset of gastric cancer containing DNA methylation microarray dataset and RNA sequencing dataset. After sample screening, quality control and grouping, 8 quality control group samples are obtained. The gene analysis module is used to perform methylation and RNA differential analysis on the quality control grouped samples to obtain the gene set related to differentially methylated sites and differentially methylated regions, as well as the RNA differentially expressed gene set; the gene screening module is used to integrate and screen the gene set of differentially methylated sites and the RNA differentially expressed gene set to obtain 8 target genes.

8. A computer-readable storage medium, wherein, The computer-readable storage medium stores computer-executable instructions that, when executed by at least one processor, implement the method for predicting the risk of early gastric cancer lymph node metastasis according to any one of claims 1-6.

9. A methylation PCR detection product for early gastric cancer lymph node metastasis, characterized in that, It contains substances that detect methylation markers PPP2R2B, LMO3, ZNF354C, and CLGN.

10. A methylation marker, characterized in that, Application of targeting in screening drugs for the prevention or treatment of early gastric cancer lymph node metastasis.

Images