Traditional Chinese medicine intelligent matching system based on network targeting comprehensive index and screening method thereof

Abstract

A traditional Chinese medicine intelligent screening method based on network targeting comprehensive indexes belongs to the technical field of traditional Chinese medicine intelligent screening. The method comprises the following steps: constructing a target set A and a protein interaction network database corresponding to each traditional Chinese medicine; constructing a disease core gene set B based on a protein interaction network; calculating a network targeting comprehensive index based on the target point set A and the disease core gene set B; calculating the network targeting comprehensive index of each traditional Chinese medicine and then performing ascending sorting, wherein the smaller the network targeting comprehensive index is, the stronger the network targeting relevance between the traditional Chinese medicine and the disease is; and matching the traditional Chinese medicines according to the obtained ascending order, and generating a structured and explainable final recommendation report. According to the method, the disease-related gene identification accuracy is improved, the biological interpretation of the result is enhanced, the standardized processing of integrating multi-source data into intelligent recommendation is realized, and the urgent demand of efficient and accurate screening in modern research of traditional Chinese medicines is met.

Description

Technical Field

[0001] This invention relates to a method for intelligent screening of traditional Chinese medicine based on a network-targeted comprehensive index, belonging to the field of intelligent screening technology for traditional Chinese medicine. Background Technology

[0002] With breakthroughs in systems biology and network pharmacology, drug development paradigms are shifting from traditional single-target research to multi-target, network-based approaches. Protein-protein interaction networks (PPINs), as core carriers of signal transduction and metabolic regulation in organisms, have become crucial analytical tools for elucidating drug mechanisms of action and disease progression. However, current technologies lack a systematic analytical framework capable of automating and intelligently integrating PPIN data, drug target information, and disease gene data. In particular, they lack the ability to deeply mine network topology, resulting in reliance on manual data processing and fixed-format input for key aspects such as network distance analysis and drug recommendation. This leads to technical shortcomings including low analytical efficiency, insufficient methodological flexibility, and weak interpretability of results.

[0003] Traditional Chinese medicine (TCM), as an important component of the traditional medical system, possesses a holistic regulatory characteristic involving multiple components, targets, and pathways that naturally aligns with the multifactorial pathological features of modern diseases. However, limited by the traditional single-target drug research paradigm, the elucidation of the complex mechanisms of action of TCM has long faced technical bottlenecks. In recent years, network pharmacology based on protein-protein interaction networks has provided a new technical approach for studying the mechanisms of action of TCM, and related tools and databases have been widely used in target identification, pathway analysis, and other aspects.

[0004] However, driven by the actual needs of modern research on traditional Chinese medicine, the existing technology system still lacks a systematic analysis framework that can automatically integrate multi-dimensional network pharmacology indicators (such as network topology parameters, target correlation, pathway coverage, etc.) and generate comprehensive quantitative scoring indicators. It is difficult to achieve multi-level intelligent matching and drug selection from target-pathway-network. There is an urgent need to build a more efficient and accurate intelligent screening method for traditional Chinese medicine through technology integration and intelligent upgrading. Summary of the Invention

[0005] To address the problems existing in the background technology, this invention provides a method for intelligent screening of traditional Chinese medicine based on a network-targeted comprehensive index.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for intelligent screening of traditional Chinese medicine based on network-targeted comprehensive index, the method comprising the following steps:

[0007] S1: Construct a target set A and a protein interaction network database for each Chinese herbal medicine;

[0008] S101: Obtain information on Chinese herbal medicine components from multiple Chinese herbal medicine databases, and take the union of the Chinese herbal medicine components from each database to form a dataset of Chinese herbal medicine components;

[0009] S102: Using the experimentally validated compound target data from the STITCH database, map the traditional Chinese medicine components in the component dataset to the validated protein targets to form a target dataset. After removing duplicates of the targets corresponding to all components contained in each traditional Chinese medicine, use it as the target set A for that traditional Chinese medicine.

[0010] S103: Construct or load human protein interaction networks based on experimentally validated data to form a protein interaction network database.

[0011] S2: Based on the protein interaction network constructed in S1, construct the core gene set B for the disease;

[0012] S201: Obtain disease-related omics datasets, perform quality control, standardization, and normalization preprocessing to obtain preprocessed gene expression data;

[0013] S202: Based on the preprocessed gene expression data, and using a preset differential expression threshold, differentially expressed genes between the disease group and the control group are screened to form the initial seed gene set S;

[0014] S203: Map the initial seed gene set S to a protein interaction network to form network nodes and construct the initial probability vector. If a network node belongs to the initial seed gene set S, then assign it the value... , This represents the number of genes in the initial seed gene set S; otherwise, it is assigned the value 0.

[0015] S204: Execute the restarted random walk algorithm iteratively and convergently until the probability distribution converges, thus obtaining the steady-state probability distribution. The iterative formula is:

[0016]

[0017] In the formula:

[0018] For the first The probability distribution vector of the step;

[0019] Let be the column random adjacency matrix of the network;

[0020] The preset restart probability;

[0021] S205: Based on steady-state probability distribution Sort all network nodes in descending order and select the top K network nodes to form the core gene set B of the disease.

[0022] S3: Calculate the network-targeted comprehensive index based on the target set A and the core disease gene set B. ;

[0023] S301: Calculate the target points of target set A. Genes related to the core gene set B of the disease Average shortest path distance in protein-protein interaction networks :

[0024]

[0025] In the formula:

[0026] For all target points with connected paths With genes The number of pairs;

[0027] Target With genes The shortest path length in a protein-protein interaction network;

[0028] S302: Calculate the average shortest path distance within the target point set A. :

[0029]

[0030] In the formula:

[0031] The number of targets in target set A;

[0032] For the target set A, the first The and the first One target point;

[0033] For the target set A The and the first The shortest path length to each target point;

[0034] S303: Calculate the average shortest path distance within the disease core gene set B. :

[0035]

[0036] In the formula:

[0037] The number of targets in target set B;

[0038] For the target set B, the first The and the first One target point;

[0039] For the target set B, the first The and the first The shortest path length to each target point;

[0040] S304: Calculate the network target comprehensive index :

[0041]

[0042] S4: Calculate the network-targeting comprehensive index for each Chinese herbal medicine. Then sort in ascending order, network target comprehensive index The smaller the value, the stronger the network target association between the herb and the disease;

[0043] S5: Match Chinese medicines according to the ascending order obtained from S4, and generate a structured and interpretable final recommendation report.

[0044] The present invention provides a method for intelligent screening of traditional Chinese medicine based on a network-targeted comprehensive index, which relies on an intelligent matching system for traditional Chinese medicine based on a network-targeted comprehensive index. This intelligent matching system includes:

[0045] The database construction module is used to execute S1 to construct the target set A and protein interaction network database corresponding to each Chinese herbal medicine.

[0046] The disease core gene analysis module is used to execute S2 and construct the disease core gene set B;

[0047] The topology analysis and calculation module is used to execute S3 and calculate the network target comprehensive index. ;

[0048] The traditional Chinese medicine matching and sorting module is used to execute S4, based on the network-targeted comprehensive index. Sort the Chinese medicines in ascending order;

[0049] The report generation module, used in S5, outputs recommendation results.

[0050] Compared with the prior art, the beneficial effects of the present invention are:

[0051] This invention achieves intelligent screening of traditional Chinese medicine (TCM) by constructing a network-targeted comprehensive index, breaking through the limitations of traditional single-target research. It establishes a multi-dimensional network pharmacology index fusion framework, automatically integrating protein-protein interaction network topology parameters, target correlation, and pathway coverage to form a quantitative comprehensive scoring system. It innovatively employs a restart-type random walk algorithm to accurately expand the core disease gene set, improving the accuracy of disease-related gene identification. Through average shortest path distance calculation and network-targeted comprehensive index formula design, it quantifies the correlation strength between the TCM target set and the disease gene set in the biological network, enhancing the biological interpretability of the results. It constructs a fully automated system module to achieve standardized processing of multi-source data integration into intelligent recommendation, solving the problems of fragmented processes and excessive manual intervention in existing systems. Finally, it generates a structured recommendation report with both objective quantitative scoring and visual interpretation of network topology features, meeting the urgent need for efficient and accurate screening in modern TCM research and promoting the upgrading of TCM intelligent screening technology towards systematization and intelligence. Attached Figure Description

[0052] Figure 1 This is a flowchart of the present invention;

[0053] Figure 2 This is a schematic diagram of the GO enrichment function analysis results of genes in the disease core gene set B in Example 1. Detailed Implementation

[0054] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the invention, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0055] A method for intelligent screening of traditional Chinese medicine based on network-targeted comprehensive index, the method comprising the following steps:

[0056] S1: Construct a target set A and a protein interaction network database for each Chinese herbal medicine;

[0057] S101: Obtain information on Chinese medicinal ingredients listed in the Chinese Pharmacopoeia from multiple Chinese medicine databases (TCMSP, HERB2.0, TCMIO, TCMID and SymMap), and take the union of the Chinese medicinal ingredients in each database to form a dataset of Chinese medicinal ingredients;

[0058] S102: Using the experimentally validated compound target data from the STITCH database, map the traditional Chinese medicine components in the component dataset to the validated protein targets to form a target dataset. After removing duplicates of the targets corresponding to all components contained in each traditional Chinese medicine, use it as the target set A for that traditional Chinese medicine.

[0059] S103: Construct or load human protein interaction networks based on experimentally validated data to form a protein interaction network database.

[0060] S2: Based on the protein interaction network constructed in S1, construct the core gene set B for the disease;

[0061] S201: Obtain disease-related omics datasets from public gene expression databases, perform quality control (removing invalid data such as sample contamination and weak detection signals), standardization (eliminating systematic errors from different detection batches and instruments), and normalization (making the expression levels of different genes comparable) preprocessing to ensure data reliability and obtain preprocessed gene expression data;

[0062] S202: Based on the preprocessed gene expression data, using a preset differential expression threshold (probability P<0.05 and |log2FC|>1, where FC is the gene expression level in the disease group), genes with significant differences in expression between the disease group (the set of biological samples corresponding to the group suffering from the target disease, which is the target of the study) and the control group (the set of biological samples corresponding to the healthy group without the target disease, which is the benchmark of the study) are selected to form the initial seed gene set S.

[0063] S203: Map the initial seed gene set S to the protein-protein interaction network to form network nodes, verify that all genes in the initial seed gene set S exist in the protein-protein interaction network, and construct the initial probability vector. If a network node belongs to the initial seed gene set S, then assign it the value... , This represents the number of genes in the initial seed gene set S; otherwise, it is assigned the value 0.

[0064] S204: Execute the restarted random walk algorithm iteratively and convergently until the probability distribution converges, thus obtaining the steady-state probability distribution. The iterative formula is:

[0065]

[0066] In the formula:

[0067] For the first The probability distribution vector of the step;

[0068] Let be the column random adjacency matrix of the network;

[0069] The preset restart probability ranges from 0.5 to 0.8, with 0.7 being the preferred value.

[0070] S205: Based on steady-state probability distribution Sort all network nodes in descending order and select the top K (a preset value, a positive integer, preferably 100) network nodes to form the disease core gene set B.

[0071] S3: Calculate the network-targeted comprehensive index based on the target set A and the core disease gene set B. ;

[0072] S301: Calculate the target points of target set A. Genes related to the core gene set B of the disease Average shortest path distance in protein-protein interaction networks :

[0073]

[0074] In the formula:

[0075] For all target points with connected paths With genes The number of pairs;

[0076] Target With genes The shortest path length in a protein-protein interaction network;

[0077] S302: Calculate the average shortest path distance within the target point set A. :

[0078]

[0079] In the formula:

[0080] The number of targets in target set A;

[0081] For the target set A, the first The and the first One target point;

[0082] For the target set A The and the first The shortest path length to each target point;

[0083] S303: Calculate the average shortest path distance within the disease core gene set B. :

[0084]

[0085] In the formula:

[0086] The number of targets in target set B;

[0087] For the target set B, the first The and the first One target point;

[0088] For the target set B, the first The and the first The shortest path length to each target point;

[0089] S304: Calculate the network target comprehensive index :

[0090]

[0091] S4: Calculate the network-targeting comprehensive index for each Chinese herbal medicine. Then sort in ascending order, network target comprehensive index The smaller the value, the stronger the network target association between the herb and the disease;

[0092] S5: Match Chinese medicines according to the ascending order obtained from S4, and generate a structured and interpretable final recommendation report.

[0093] The present invention provides a method for intelligent screening of traditional Chinese medicine based on a network-targeted comprehensive index, which relies on an intelligent matching system for traditional Chinese medicine based on a network-targeted comprehensive index. This intelligent matching system includes:

[0094] The database construction module is used to execute S1 to construct the target set A and protein interaction network database corresponding to each Chinese herbal medicine.

[0095] The disease core gene analysis module is used to execute S2 and construct the disease core gene set B;

[0096] The topology analysis and calculation module is used to execute S3 and calculate the network target comprehensive index. ;

[0097] The traditional Chinese medicine matching and sorting module is used to execute S4, based on the network-targeted comprehensive index. Sort the Chinese medicines in ascending order;

[0098] The report generation module, used in S5, outputs recommendation results.

[0099] Example 1:

[0100] Prepare resources and computing environment, including basic databases (an integrated set of targets for each Chinese herbal medicine, a protein interaction network database, and a STITCH database).

[0101] The dataset GSE25724, containing data on type II diabetes and healthy individuals, was obtained from the NCBI-GEO public database. The chip data underwent quality control, standardization, and normalization to ensure data reliability.

[0102] Based on the preprocessed gene expression data, differential expression analysis was performed using the limma package with P < 0.05 and |log2FC| > 1 as the threshold to screen differentially expressed genes and form the initial seed gene set S.

[0103] The system automatically checks whether the genes in the initial seed gene set S exist in the preloaded protein interaction network and retains the genes that exist in the network.

[0104] Set restart probability The restart-type random walk algorithm is executed iteratively until convergence is obtained to obtain the steady-state probability distribution. The top 100 genes constitute the core gene set B of the disease.

[0105] like Figure 2 As shown, GO enrichment function analysis was performed on genes in the core gene set B of the disease, systematically revealing the biological characteristics of differentially expressed genes.

[0106] The bar chart results indicate that these genes are widely involved in key processes such as insulin metabolism, amino acid metabolism, and vascular endothelial growth factor receptor binding, suggesting that their influence is multifaceted and extensive.

[0107] Further bubble diagram analysis provided stronger evidence in terms of statistical significance and functional specificity. The above functions not only showed high gene enrichment ratios, but also had significant statistical significance, jointly outlining a complete dysfunction network from molecular binding activity, vesicle transport to hormone metabolism regulation.

[0108] These findings consistently point to typical metabolic disorders and insulin signaling impairments in type 2 diabetes and its vascular complications, providing important functional evidence for elucidating the pathological mechanisms of type 2 diabetes.

[0109] For each Chinese herbal medicine, its network targeting comprehensive index is calculated. .

[0110] The system traversed 616 Chinese medicinal herbs, as shown in Table 1, based on the network targeting comprehensive index. The top eight Chinese medicinal herbs selected were *Gynostemma pentaphyllum*, *Agkistrodon halys*, *Scrophularia ningpoensis*, *Dracaena cochinchinensis*, *Citrus aurantium*, *Polygonum aviculare*, *Zaocys dhumnades*, and leech. All eight herbs have pharmacological or clinical associations with type II diabetes and its complications, but the strength and manner of these associations differ. *Gynostemma pentaphyllum*, *Scrophularia ningpoensis*, and *Citrus aurantium* have been supported by numerous studies regarding their role in regulating blood sugar. Leech, *Dracaena cochinchinensis*, *Agkistrodon halys*, and *Zaocys dhumnades* are primarily used to prevent and treat diabetic complications. *Polygonum aviculare* can be used as an adjunct therapy.

[0111] Table 1 System calculation of traditional Chinese medicine and The result

[0112]

[0113] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0114] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for intelligent screening of traditional Chinese medicine based on network-targeted comprehensive index, characterized in that: The method includes the following steps: S1: Construct a target set A and a protein interaction network database for each Chinese herbal medicine; S2: Based on the protein interaction network constructed in S1, construct the core gene set B for the disease; S3: Calculate the network-targeted comprehensive index based on the target set A and the core disease gene set B. ; S4: Calculate the network-targeting comprehensive index for each Chinese herbal medicine. Then sort in ascending order, network target comprehensive index The smaller the value, the stronger the network target association between the herb and the disease; S5: Match Chinese medicines according to the ascending order obtained from S4, and generate a structured and interpretable final recommendation report.

2. The intelligent screening method for traditional Chinese medicine based on network-targeted comprehensive index according to claim 1, characterized in that: S1 includes the following steps: S101: Obtain information on Chinese herbal medicine components from multiple Chinese herbal medicine databases, and take the union of the Chinese herbal medicine components from each database to form a dataset of Chinese herbal medicine components; S102: Using the experimentally validated compound target data from the STITCH database, map the traditional Chinese medicine components in the component dataset to the validated protein targets to form a target dataset. After removing duplicates of the targets corresponding to all components contained in each traditional Chinese medicine, use it as the target set A for that traditional Chinese medicine. S103: Construct or load human protein interaction networks based on experimentally validated data to form a protein interaction network database.

3. The intelligent screening method for traditional Chinese medicine based on network-targeted comprehensive index according to claim 2, characterized in that: S2 includes the following steps: S201: Obtain disease-related omics datasets, perform quality control, standardization, and normalization preprocessing to obtain preprocessed gene expression data; S202: Based on the preprocessed gene expression data, and using a preset differential expression threshold, differentially expressed genes between the disease group and the control group are screened to form the initial seed gene set S; S203: Map the initial seed gene set S to a protein interaction network to form network nodes and construct the initial probability vector. ; If a network node belongs to the initial seed gene set S, then assign it the value... , This represents the number of genes in the initial seed gene set S; otherwise, it is assigned the value 0. S204: Execute the restarted random walk algorithm iteratively and convergently until the probability distribution converges, thus obtaining the steady-state probability distribution. The iterative formula is: In the formula: For the first The probability distribution vector of the step; Let be the column random adjacency matrix of the network; The preset restart probability; S205: Based on steady-state probability distribution Sort all network nodes in descending order and select the top K network nodes to form the core gene set B of the disease.

4. The intelligent screening method for traditional Chinese medicine based on network-targeted comprehensive index according to claim 3, characterized in that: S3 includes the following steps: S301: Calculate the target points of target set A. Genes related to the core gene set B of the disease Average shortest path distance in protein-protein interaction networks : In the formula: For all target points with connected paths With genes The number of pairs; Target With genes The shortest path length in a protein-protein interaction network; S302: Calculate the average shortest path distance within the target point set A. : In the formula: The number of targets in target set A; For the target set A, the first The and the first One target point; For the target set A The and the first The shortest path length to each target point; S303: Calculate the average shortest path distance within the disease core gene set B. : In the formula: The number of targets in target set B; For the target set B, the first The and the first One target point; For the target set B, the first The and the first The shortest path length to each target point; S304: Calculate the network target comprehensive index : 。 5. A method for intelligent screening of traditional Chinese medicine based on network-targeted comprehensive index according to any one of claims 1-4, characterized in that: The method relies on a TCM intelligent matching system based on a network-targeted comprehensive index, which includes: The database construction module is used to execute S1 to construct the target set A and protein interaction network database corresponding to each Chinese herbal medicine. The disease core gene analysis module is used to execute S2 and construct the disease core gene set B; The topology analysis and calculation module is used to execute S3 and calculate the network target comprehensive index. ; The traditional Chinese medicine matching and sorting module is used to execute S4, based on the network target comprehensive index. Sort the Chinese medicines in ascending order; The report generation module, used in S5, outputs recommendation results.

Images