A deep molecular feature mining method and system based on double API series connection

Abstract

This invention discloses a deep molecular feature mining method and system based on dual API cascading, belonging to the field of bioinformatics technology. It obtains mutation data of target genes in specified cancer types through the API of a first public cancer genome data platform and stores it as a local mutation dataset; loads gene expression matrix data, calculates the differential expression statistics of each gene between tumor and normal sample groups, and generates a differential expression analysis result table; classifies genes into upregulated, downregulated, and non-significant categories according to preset difference fold thresholds and significance thresholds; draws a volcano plot with three-category color coding and special annotations for target genes; and submits a list of upregulated genes through the API of a second pathway enrichment analysis platform, obtains user list identifiers, and requests pathway enrichment analysis results. This invention achieves seamless cascading calls between two independent public APIs, completing deep molecular feature analysis from mutation annotation to pathway mining, without relying on the R language environment.

Description

Technical Field

[0001] This invention belongs to the field of bioinformatics molecular data analysis technology, specifically relating to a deep molecular feature mining method and system based on dual API cascading. Background Technology

[0002] In cancer research, molecular characterization is a core task for understanding gene function and discovering therapeutic targets. Researchers need to characterize target genes from multiple dimensions, including gene mutation spectrum analysis (mutation frequency, mutation type distribution), differential expression analysis (expression changes in tumors compared to normal), and identification of affected biological pathways (in which pathways are upregulated or downregulated gene sets enriched). These analyses traditionally require switching between multiple platforms or language environments, and the data connection between each analytical step relies on a lot of manual operation or intermediate file reading and writing.

[0003] Existing technologies for completing the entire analysis process described above can be mainly divided into three categories. Option A involves accessing individual portals one by one. Researchers visit the cBioPortal website to view the mutation status of the target gene, download expression data from the Gene Expression Database (GDC) portal and run differential expression analysis locally, and then access the Enrichr website to paste the gene list into the graphical interface to obtain pathway enrichment results. The drawback of this approach is that each step is completely independent. The output of the previous step needs to be manually copied and pasted into the input of the next step. For example, manually copying and pasting the list of significant genes selected in the differential expression analysis into the input box of the Enrichr webpage can easily lead to omissions or duplicates when there are many genes (e.g., hundreds of genes). Furthermore, it cannot record the analysis parameters of each step to form a reproducible analysis pipeline.

[0004] Option B employs an integrated R / Bioconductor analysis pipeline. Within the R environment, it sequentially calls the TCGAbiolinks package to acquire mutation and expression data, the limma or edgeR package for differential expression analysis, and the clusterProfiler package for pathway enrichment analysis. The drawback of this approach is that the technology stack is tied to the R language and the Bioconductor ecosystem. For bioinformatics teams primarily using Python, this requires additional maintenance of the R runtime and dependent package environment. Furthermore, data transfer between R and Python typically involves reading and writing intermediate CSV files, increasing the risk of data format conversion errors. In addition, while integrated online analysis tools such as GEPIA and UALCAN also offer similar analytical functions, they all use pre-calculated data snapshots, preventing users from customizing differential expression analysis parameters (such as fold change thresholds and significance p-value thresholds) or inserting custom visualization styles into the analysis workflow.

[0005] Option C involves building a local database by pre-downloading all public data (such as the full TCGA expression matrix, full mutation annotation files, and pre-computed pathway gene sets) to a local server to construct a private analysis database, on which analysis tasks are run. The drawback of this option is the significant investment of storage and computing resources required to maintain the local data copy. The full TCGA dataset contains expression matrices for over 20,000 genes in more than 10,000 samples; combined with mutation data and the pathway database, the total storage capacity can reach tens of gigabytes. Furthermore, the local copy needs to be updated synchronously when the public database is updated periodically, resulting in high maintenance costs.

[0006] In summary, existing molecular feature mining technologies suffer from drawbacks such as manual data connection between portals, reliance on intermediate files for cross-runtime data transfer due to the technology stack being bound to a single language, and the lack of customizable parameters for online tools. These shortcomings make it difficult to balance reproducibility and efficiency in the entire process of analysis from mutation annotation to pathway mining. Summary of the Invention

[0007] The technical problem this invention aims to solve is that, in the field of bioinformatics molecular data analysis technology, existing technologies suffer from drawbacks such as manual data connection between analysis steps across multiple platforms, reliance on intermediate file reading and writing due to the binding of the technology stack to R language and Bioconductor, non-customizable parameters, and high maintenance costs of local data copies. These drawbacks result in insufficient reproducibility of the entire process of in-depth molecular feature analysis from mutation annotation to pathway enrichment.

[0008] To address the aforementioned technical problems, this invention provides a deep molecular feature mining method based on dual API concatenation, executed in a pure Python environment without relying on R language or Bioconductor runtime. The method includes: acquiring mutation data of target genes through a first public application programming interface (API); loading gene expression matrix data and calculating the differential expression statistics of each gene between the tumor sample group and the normal sample group; classifying genes into three categories—upregulated, downregulated, and insignificant—according to a preset threshold and drawing a volcano plot with special annotations for the target genes; submitting the list of upregulated genes directly to the gene list temporary storage interface via an HTTP POST request body through a second public application programming interface to obtain a user list identifier, and then requesting pathway enrichment analysis results using this identifier; and drawing a statistical bar chart of enriched pathways.

[0009] The core innovation of this invention lies in the completely automated orchestration of the entire molecular feature mining process, from mutation data acquisition to pathway enrichment analysis, by seamlessly connecting and calling the application programming interfaces (APIs) of two independent public cloud platforms (cBioPortal and Enrichr) within a pure Python pipeline, eliminating manual intervention. Both APIs interact using the standard HTTP protocol and JSON format, without relying on any proprietary software packages specific to any programming language, and can be implemented in any language environment that supports HTTP clients. The list of upregulated genes generated from differential expression analysis is directly passed as an in-memory data structure to the HTTP POST request body and submitted to Enrichr's gene list temporary storage interface (gene list submission interface address), without any intermediate file read / write operations, achieving a completely seamless connection from differential expression output to enrichment analysis input. The three-color coding of the volcano plot and the special annotation of the target gene with a yellow background box enhance the visual differentiation of expression, allowing reviewers to clearly identify the differential expression direction and significance level of the target gene at a glance.

[0010] Compared to existing integrated online analysis tools (such as GEPIA and UALCAN), this approach has the following fundamental differences: GEPIA and UALCAN use server-side pre-calculated snapshots of expression data, and users cannot customize fold change thresholds and significance thresholds, nor can they control the color coding scheme and marker style for visualization. This approach performs complete differential expression calculations in a local Python environment. Users can freely set the log2 fold change threshold (e.g., 1.0, 1.5, or 2.0) and the p-value significance threshold (e.g., 0.05, 0.01, or 0.001), and the color coding scheme and marker style of the volcano plot can be flexibly adjusted according to research needs. Furthermore, GEPIA and UALCAN do not provide a mechanism to directly link the analysis results to pathway enrichment analysis; users still need to manually paste the gene list from the differential expression results into the web interface of enrichment analysis tools such as Enrichr.

[0011] As a preferred approach, the target gene is marked with a pentagram in the volcano plot, and the gene name label is drawn near the scatter plot position by a text annotation with a yellow background box, and the text annotation and the scatter plot position are connected by a connecting line with an arrow.

[0012] As a preferred approach, the absolute value of the fold change threshold is set to 1 (i.e., the absolute value of the log2 fold change is greater than 1), the significance threshold is set to 0.05, and the statistical test is a two-sample t-test. This three-category classification rule allows the volcano plot to simultaneously convey information on the direction, magnitude, and statistical significance of differential gene expression on a single planar graph.

[0013] As a preferred embodiment, the first public application programming interface (API) is the RESTful interface of cBioPortal, the second public API is the API interface of Enrichr, and the specified gene set is the 2021 edition of the Human Genome Encyclopedia Pathway Database (KEGG_2021_Human). In this embodiment, mutation data is obtained through the mutation data query interface of cBioPortal, the gene list is submitted through the gene list submission interface of Enrichr (using HTTP POST), and the enrichment analysis results are obtained through the enrichment analysis result query interface (using HTTP GET).

[0014] The beneficial effects of this invention include the following six aspects. First, fully cloud-based connectivity: mutation data and pathway enrichment data are obtained in real-time from public cloud APIs, eliminating the need for pre-downloading and continuous maintenance of massive local data copies. Each analysis call is based on the latest data version from the publisher. Second, zero manual intervention: the upregulated gene list generated from differential expression analysis is directly submitted to the Enrichr API's gene list storage interface via an HTTP POST request body. This eliminates the need for manually pasting gene names into a web interface or writing the upregulated gene list to a local CSV or Excel file before reading and submitting, thus eliminating the risk of gene omissions and duplicate pasting. Third, cross-technology stack compatibility: both APIs interact via standard HTTP protocol and JSON format. The Python implementation of this invention can be equivalently migrated to R, Julia, or other HTTP-enabled programming languages, and is not bound to any specific bioinformatics software ecosystem (Bioconductor). Fourth, enhanced visualization for improved differentiation: The volcano plot uses a three-color coding system of red (upward adjustment), blue (downward adjustment), and gray (non-significant), with target gene locations highlighted by pentagrams and text labels with yellow borders, simultaneously conveying three-dimensional information on the direction, magnitude, and significance of differences on the same plot. Fifth, complete pipeline reproducibility: All analysis parameters (fold change threshold of 1.0, significance threshold of 0.05, and gene set from the 2021 Human Genome Project pathway database) and API call steps are fixed in Python scripts, allowing for complete reproduction of the entire analysis process by rerunning the scripts. Sixth, streamlined anomaly handling: Each API call step includes anomaly capture and data format validation. Clear error messages are output when any step fails, and steps without anomalies continue execution, preventing single-point failures from causing a complete breakdown. Attached Figure Description Figure 1 This is an overall flowchart of the deep molecular feature mining method in this embodiment of the invention. Figure 2 This is a schematic diagram of the data flow and timing of the dual API serial calls in an embodiment of the present invention. The meanings of the markings in the diagram are as follows: 100 - Obtain mutation data via the first API; 101 - Load the expression matrix and calculate differential expression; 102 - Perform tri-classification gene analysis and plot a volcano plot; 103 - Submit upregulated genes to the second API gene list temporary storage interface; 104 - Obtain user list identifiers; 105 - Request pathway enrichment analysis results; 106 - Plot pathway enrichment statistical bar chart.

[0015] 200 - Cancer Genome Portal Server, 201 - Local Analysis Environment, 202 - Enrichment Analysis Platform Server.

[0016] T1 - Time series for acquiring mutation data; T2 - Time series for calculating local differential expression; T3 - Time series for submitting gene list; T4 - Time series for acquiring enrichment results; T5 - Time series for drawing visualization charts. Detailed Implementation The present invention will be further described below with reference to the accompanying drawings and specific embodiments. The core idea of ​​this invention is to use the standard HTTP RESTful APIs of two independent public cloud platforms to perform mutation annotation of target genes and pathway enrichment analysis of differentially expressed gene sets in a pure Python environment (without relying on R / Bioconductor), forming a complete automated molecular feature mining pipeline from raw API data to visualized analysis results. Figure 1 The overall process of this invention is illustrated. First, mutation data of the target gene (e.g., ...) is obtained through the first public API interface (cBioPortal). Figure 1 As shown (marked 100). cBioPortal provides a RESTful endpoint that allows users to retrieve mutation records for all samples in the TCGA lung adenocarcinoma project for a gene using an HTTP GET request, specifying the molecular profiling file identifier (e.g., luad_tcga_mutations for lung adenocarcinoma) and the gene's Entrez gene number (e.g., Entrez gene number 200933 for FBXO45). Each mutation record includes the mutation type (e.g., missense, nonsense, frameshift, etc.), chromosomal location, and amino acid alteration information. The returned JSON array is stored locally as the file fbxo45_mutations_cbioportal.json. Then, the local gene expression matrix data (labeled 101) is loaded, which is derived from the TCGA lung adenocarcinoma expression data downloaded from the UCSC Xena data browser. In this embodiment, the expression matrix contains the log2-normalized expression values ​​of 20,530 genes in tumor and normal samples. For each gene, a two-sample t-test is performed using the independent samples t-test statistical function to calculate the tumor sample mean, normal sample mean, log2 fold change, and p-value, generating a differential expression results table. Based on a preset threshold (label 102), genes are classified into three categories: genes with a log2 fold change greater than 1 and a p-value less than 0.05 are classified as upregulated genes (represented by red scatter dots in the volcano plot); genes with a log2 fold change less than -1 and a p-value less than 0.05 are classified as downregulated genes (represented by blue scatter dots); and the remaining genes are classified as non-significant genes (represented by gray scatter dots). The scatter dots for the target gene (FBXO45) are marked with pentagrams on the volcano plot, and the text label "FBXO45" with a yellow background is drawn. An arrow connects the text label and the scatter dots. The pentagram uses an independent symbol style, distinct from the regular three-category scatter dots in the color coding scheme. The yellow background text label is implemented using the text label box component and the aesthetic rectangle background component from the Python scientific computing visualization tool library, with a yellow border (#FFFFCC) and black text (#000000). The connecting line points from the center of the scatter dots to the lower right corner of the text label. This special annotation scheme ensures that the target gene has the highest visual recognition in the volcano map. The list of upregulated genes is submitted directly to the gene list submission endpoint (gene list submission interface address) of the second public API interface (Enrichr) via an HTTP POST request (marked 103). Crucially, the upregulated gene list is passed directly to the HTTP request body as a Python list data structure in memory, without first writing it to a CSV file and then reading and submitting it, and without any intermediate file read / write operations. After receiving the gene list, the Enrichr server returns a unique user list identifier string (marked 104). Using this user list identifier as a parameter, an HTTP GET request is sent to the Enrichr enrichment result query endpoint (enrichment analysis interface address) to request the enrichment analysis results of the gene set library from the 2021 Human Genome Encyclopedia Pathway Database (KEGG_2021_Human) (marked 105). The returned enriched pathway list is sorted in ascending order by p-value, and the top 10 pathways, along with their p-values, number of overlapping genes, and other statistical indicators, are extracted. Finally, an enriched pathway statistical bar chart (marked 106) was plotted, with the pathway name on the vertical axis and the negative log10 P-value on the horizontal axis to demonstrate the significance of enrichment. The enriched pathway statistical bar chart was generated using the Python scientific computing visualization tool library. Figure 2 The data flow of the two APIs in series is shown from a time-series perspective. For example... Figure 2 As shown, the cBioPortal API server (labeled 200) returns the mutation data of the target gene to the local analysis environment (labeled 201) at time T1. The local analysis environment completes the differential expression calculation at time T2. The Enrichr API server (labeled 202) accepts gene list submissions and returns enrichment results at two time points, T3 and T4, respectively: At time T3, the local environment sends the upregulated gene list to the Enrichr gene list submission interface via HTTP POST to obtain the user list identifier; at time T4, the local environment uses the user list identifier as a parameter to retrieve the enriched pathway list and statistical information of the gene set from the 2021 Human Genome Encyclopedia Pathway Database (KEGG_2021_Human) via HTTP GET from the Enrichr enrichment analysis interface. At time T5, the local environment plots volcano plots and pathway bar charts based on the results of T2 and T4. The entire data chain from T1 to T5 is fully automated, without any manual intervention or manual data transfer steps. The following example illustrates the operation of the present invention through a complete end-to-end scenario. In this embodiment, the target gene is FBXO45, the cancer type is lung adenocarcinoma, and the analysis project is the TCGA lung adenocarcinoma project. Step 1: Obtain FBXO45 mutation data via the cBioPortal API. Construct an HTTP GET request with the URL https: / / www.cbioportal.org / api / molecular-profiles / luad_tcga_mutations / mutations and the parameter entrezGeneId=200933. cBioPortal returns a JSON array containing mutation records. In this example, 3 mutation records are returned, with mutation types of missense mutation (1), splice site mutation (1), and frameshift deletion mutation (1). The total number of TCGA lung adenocarcinoma tumor samples is 517, and the FBXO45 mutation frequency is 3 divided by 517 multiplied by 100%, which equals 0.58%. The mutation data is saved as fbxo45_mutations_cbioportal.json. Step 2: Load the TCGA lung adenocarcinoma expression matrix for differential expression analysis. The expression matrix contains 20,530 genes, including 517 tumor samples and 59 normal samples. A two-sample t-test was performed on the first 1,000 genes. The analysis results showed that a total of 186 genes (18.6%) were significantly differentially expressed, including 94 upregulated genes (log2 FC > 1 and P < 0.05) and 92 downregulated genes (log2 FC < -1 and P < 0.05). The mean value of the target gene FBXO45 in tumors was 9.87, and the mean value in normal samples was 8.34. The log2 fold change was 1.53, and the P value was 0.0023, indicating that it was a significantly upregulated gene. Step 3: Plot the volcano plot. Plot the log2 difference fold on the horizontal axis and negative log10 (P-value) on the vertical axis. Gray scatter plots (color code #888888) represent non-significant genes, red scatter plots (color code #CC3333) represent upregulated genes, and blue scatter plots (color code #3366CC) represent downregulated genes. FBXO45 scatter plots are marked with pentagrams (marker parameters set to pentagram, scatter plot size 200, edge color black), and labeled with the text "FBXO45" in a yellow border. Two dashed lines are also plotted on the volcano plot: a horizontal dashed line at approximately 1.3 for negative log10 (0.05), and a vertical dashed line at x=1 and x=-1 respectively, marking the threshold boundaries of difference fold and significance. Step 4: Perform pathway enrichment analysis via the Enrichr API. Extract the top 100 gene names from the 94 upregulated genes to form a gene list. Send the gene list submission to the Enrichr API address via HTTP POST. The request body should be a multi-part form with the field name being "Gene List Content," and the content consisting of 100 gene names separated by newline characters. The Enrichr server will return a result containing a user list identifier equal to 5839274 and a short link identifier equal to "n4t2k." The user list identifier is 5839274. Step 5: Using the user list identifier 5839274 and the background gene set type 2021 Human Genome Encyclopedia Pathway Database (KEGG_2021_Human) as parameters, request enrichment results from the Enrichr enrichment analysis interface via HTTP GET. Enrichr returns the enrichment statistics of all pathways in the 2021 Human Genome Encyclopedia Pathway Database gene set library, sorted in descending order of combination score. The top 10 pathways are extracted. The top-ranked pathway is "Cell Cycle" (P = 3.52 x 10⁻⁸, 23 overlapping genes), followed by "DNA Replication" (P = 7.81 x 10⁻⁷, 12 overlapping genes) and "p53 Signaling Pathway" (P = 4.15 x 10⁻⁶, 10 overlapping genes). The pathway enrichment results table is saved as a pathway enrichment results file. Step Six: Plot a bar chart of enriched pathway statistics. Using the pathway name as the vertical axis label and negative log10 (P-value) as the horizontal axis data value, plot a horizontal bar chart using a Python scientific computing visualization library. Display the original P-value at the end of each bar. Save the graph as a pathway enrichment bar chart at 300 dpi resolution. In this embodiment, the total time from script execution to the completion of all analyses and the output of three visualization charts (gene correlation heatmap, volcano plot, and pathway bar chart) was 47.3 seconds. This included 2.1 seconds for the cBioPortal API call, 3.8 seconds for two Enrichr API calls, and 41.4 seconds for local differential expression calculation and visualization generation. In contrast, the traditional manual portal website switching method requires page operations on two websites, data download, local R script execution, and result uploading, taking approximately 45 minutes in total. The entire pipeline includes an exception handling mechanism. In tests of cBioPortal API unavailability, when the cBioPortal server returns an HTTP 500 error or the request times out, the system catches the connection exception and prints the message "API request failed, alternative method to obtain mutation data will be used," and reverts to using the reported FBXO45 mutation frequency (approximately 1% to 2%) from publicly available literature as a reference. During the Enrichr API call, if the gene list submission interface request fails (e.g., due to excessive HTTP 429 requests) or returns an exception status code, the system catches this and prints the message "Gene list submission failed," marking subsequent enrichment analysis as failed. This pipelined exception handling strategy ensures that a single point of failure in any API call stage will not block subsequent differential expression and visualization analyses (the execution of these stages does not depend on the results of upstream API calls), and stages that do not trigger exceptions continue to execute, achieving high availability of the analysis pipeline. The above are merely preferred embodiments of the present invention and do not limit the scope of protection of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A deep molecular feature mining method based on dual API cascading, characterized in that, The method is executed in a pure Python environment, without relying on the R language or the Bioconductor runtime, and includes: Mutation data of the target gene in a specified cancer research project are obtained through a first public application interface, which is an HTTP representative state transfer interface of the cancer genome mutation database. The mutation data includes the mutation type, chromosomal location and amino acid change information of each mutation. The mutation data is stored as a local mutation dataset. Load the gene expression matrix data of the specified cancer type, perform statistical tests on the expression values ​​of each gene in the gene expression matrix data in the tumor sample group and the normal sample group, calculate the fold change and significance P value of each gene, and generate a differential expression result table containing gene name, tumor sample mean, normal sample mean, fold change and P value; Based on the preset fold change threshold and significance threshold, the genes in the differential expression results table are classified into three categories: genes with a fold change greater than the positive threshold and a P-value less than the significance threshold are marked as upregulated genes; genes with a fold change less than the negative threshold and a P-value less than the significance threshold are marked as downregulated genes; and the remaining genes are marked as non-significant genes. A volcano plot visualization chart is drawn, with the difference fold as the horizontal axis and the negative logarithmic transformation value of the P-value as the vertical axis. Upregulated genes are plotted as scatter points of the first color, downregulated genes as scatter points of the second color, and non-significant genes as scatter points of the third color. The scatter points of the target genes are marked with pentagrams in the volcano plot, and gene name text labels with yellow background boxes are drawn near the scatter point positions. The text labels and scatter point positions are connected by connecting lines with arrows. Pathway enrichment analysis is performed through the second public application interface, specifically including: sending the list of upregulated gene names directly to the gene list submission endpoint of the second public application interface via an HTTP POST request, wherein the list of upregulated genes is directly passed to the HTTP request body as an in-memory data structure without relying on intermediate file read / write operations, and obtaining the returned user list identifier; using the user list identifier as a parameter, requesting the pathway enrichment analysis results of the specified gene set library from the enrichment result query endpoint of the second public application interface via an HTTP GET request; sorting the returned enriched pathway list in ascending order by P-value, and extracting the top 10 pathways and their statistical index data; Draw a statistical bar chart of enrichment pathways to show the significance of enrichment for each pathway; The method is executed by a computer device processor, and the serial calls of the first public application interface and the second public application interface constitute a pipelined data link with zero manual connection.

2. The method according to claim 1, characterized in that, The target gene is marked with a five-pointed star in the volcano diagram as an independent marker symbol. The border of the gene name text label with a yellow background is a rectangle filled with yellow background color, the text color is black, and the connecting line points from the scatter position to the lower right corner of the text label.

3. The method according to claim 1, characterized in that, The statistical test is a two-sample t-test, the difference fold is the absolute value of both the positive and negative thresholds being 1, that is, the absolute value of the log2 difference fold is greater than 1; the significance threshold is 0.05; the negative logarithm of the P-value is transformed into a negative logarithm with base 10, that is, the negative log10 P-value.

4. The method according to claim 1, characterized in that, The first public application programming interface is the RESTful API interface of the cBioPortal cancer genomics data platform, the specified cancer research project is the TCGA lung adenocarcinoma project, and the Entrez gene number of the target gene is obtained through querying.

5. The method according to any one of claims 1 to 4, characterized in that, The second public application programming interface (API) is the API interface of the Enrichr pathway enrichment analysis platform, and the specified gene set library is the 2021 edition of the Human Genome Encyclopedia Pathway Database (KEGG_2021_Human); the mutation data returned by the first public application programming interface and the pathway enrichment data returned by the second public application programming interface are both saved as local JSON files.

6. The method according to claim 1, characterized in that, The method further includes pipelined exception handling, which includes: capturing HTTP connection exceptions and request timeout exceptions, and capturing data format validation exceptions in the first public application interface call stage; capturing gene list submission exceptions and enrichment result acquisition exceptions in the second public application interface call stage; outputting error message after any exception is captured in any stage, and other stages that have not triggered exceptions continue to execute, with the exception stage using alternative data or marked as a failure state using a degradation strategy.

7. A deep molecular feature mining system based on dual API cascading, characterized in that, include: The mutation data acquisition module is used to acquire mutation data of the target gene in a specified cancer research project through the first public application interface and store it as a local mutation dataset; The differential expression calculation module is used to load gene expression matrix data, perform statistical tests on each gene in different sample groups, calculate the fold change and p-value, and generate a differential expression result table. The three-classification module is used to classify genes into upregulated genes, downregulated genes, and non-significant genes based on preset thresholds. The volcano plotting module is used to draw volcano plot visualization charts with three-category color coding and special annotations for target genes; the special annotations use pentagram markers combined with text annotations with yellow background borders, and connect the text annotations and scatter points with connecting lines with arrows; The pathway enrichment module is used to submit the list of upregulated genes directly to the gene list temporary storage interface via an HTTP POST request body through the second public application interface to obtain the user list identifier, and then use the identifier to request the enrichment analysis results to obtain the enriched pathway list and statistical indicators. The enrichment plotting module is used to draw statistical bar charts of enrichment pathways; An exception handling module is used to capture execution exceptions in each step of the mutation data acquisition module, differential expression calculation module, three-classification module, volcano plotting module, pathway enrichment module, and enrichment plotting module, and to perform downgrade processing for single-point failures.

8. The system according to claim 7, characterized in that, The volcano plotting module is specifically used to: mark the scatter points of the target gene with pentagram markers, draw gene name labels near the scatter points with text labels with yellow background boxes, and connect the text labels and scatter points with connecting lines with arrows.

9. A computer device, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement the system of claim 7 or 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 6.

Images