A Network Pharmacology-Based Method for Screening Betulinic Acid Targets in the Treatment of Ischemic Stroke

By employing network pharmacology methods, betulinic acid and ischemic stroke targets were uniformly named. Combined with multi-level topological screening and molecular simulation, the problems of inconsistent naming and difficulty in tracing the hierarchy in multi-source target screening were solved, thus achieving the reliability and traceability of target screening.

CN123120018APending Publication Date: 2026-11-17HUAZHONG UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611185549.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-06
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing studies on betulinic acid as a target for ischemic stroke prevention suffer from inconsistent names for multiple target sources, difficulty in tracing screening levels, and reliance on a single calculation indicator to determine core targets, resulting in unreliable screening results.

Method used

Using a network pharmacology-based approach, we identified betulinic acid and ischemic stroke targets. After unifying gene nomenclature, we performed deduplication and intersection operations, screened protein-protein interaction networks using at least two levels of topological thresholds, conducted gene ontology function and pathway enrichment analysis, and gradually narrowed down the candidate target set by combining molecular docking and molecular dynamics simulations.

Benefits of technology

The candidate targets were gradually converged, ensuring that the screening results had a clear intermediate level and traceability. The core candidate targets also had evidence of network topology and pathway association. The combination of molecular simulation results and static scoring improved the reliability of the screening results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123120018A_ABST
    Figure CN123120018A_ABST
Patent Text Reader

Abstract

This invention relates to the fields of bioinformatics and computer-aided drug research, and discloses a method for screening betulinic acid targets for ischemic stroke based on network pharmacology. The method obtains and standardizes a betulinic acid target set and an ischemic stroke target set, constructs a protein-protein interaction network by taking the intersection, and performs hierarchical topological screening. Candidate targets are enriched functionally and through pathway-target cross-mapping. The binding tendency of candidate proteins to betulinic acid is evaluated through molecular docking, and molecular dynamics simulation analysis is performed on the selected candidate complexes. Core targets and their correlation analysis results are output according to hub target conditions, pathway mapping conditions, and binding energy screening conditions. This method can form a traceable target screening chain with clear origin, screening, and verification relationships, reducing the result bias caused by relying on a single indicator to determine core targets.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the fields of bioinformatics and computer-aided drug technology, specifically to a method for screening betulinic acid targets for ischemic stroke based on network pharmacology. Background Technology

[0002] Ischemic stroke is a group of diseases characterized by ischemic and hypoxic damage to brain tissue due to interruption or reduction of blood flow. The injury process in ischemic stroke is not caused by a single factor, but involves interrelated biological processes such as oxidative stress, neuroinflammation, apoptosis, cell cycle alterations, and abnormal energy metabolism. Different biological processes may also be linked through common proteins and signaling pathways. Therefore, examining only a pre-selected target is usually insufficient to reflect the multi-target effects that natural compounds may possess.

[0003] Betulinic acid is a natural compound obtained from plants such as birch. Existing research data document its anti-inflammatory and antioxidant pharmacological activities, and it has been used in research related to cerebral ischemia-reperfusion injury. Since natural compounds may simultaneously target multiple protein sites, and their mechanisms of action may involve multiple intertwined signaling pathways, it is necessary to organize and screen candidate targets from multiple levels, including compound targets, disease targets, protein-protein interactions, and pathway associations.

[0004] Existing studies have observed the effects of betulinic acid in cerebral ischemia-reperfusion models and analyzed the possible biological processes involved, including oxidative stress and inflammatory responses. Other studies have explored mechanisms such as the PI3K / Akt signaling pathway and NOX4 expression, suggesting that betulinic acid could be a candidate compound for research related to cerebral ischemia. Some studies have reported that betulinic acid treatment can reduce infarct volume and improve neurological deficits, indicating a pharmacological basis for the relationship between betulinic acid and cerebral ischemia injury. However, these existing studies primarily evaluate the efficacy of betulinic acid in animal models and do not cover the process of selecting candidate core targets by starting from multiple drug targets and disease targets, through intersection processing, protein-protein interaction network screening, enrichment mapping, and molecular simulation.

[0005] In addition, existing research mostly revolves around pre-selected biological processes or signaling pathways. For the multiple targets and pathway relationships that natural compounds may involve, it is still necessary to uniformly organize candidate targets from multiple levels such as compound targets, disease targets, protein interactions and pathway associations, and explain the source and screening basis of core targets through continuous data processing.

[0006] Network pharmacology analyzes the relationships between drugs, targets, diseases, and pathways, organizing target information scattered across different databases into interconnected networks. In the analysis of betulinic acid's effects on ischemic stroke, the target set for betulinic acid and the target set for ischemic stroke can be obtained separately, and then the candidate range can be narrowed down through intersection operations. However, because different databases have inconsistent record names, inclusion focuses, and output formats, a unified gene nomenclature system needs to be used before performing intersection operations; otherwise, the same target may not be identified due to different names, or may be counted multiple times due to duplicate records.

[0007] After obtaining the intersection targets, a protein-protein interaction network can be used to represent the associations between the proteins corresponding to the candidate targets. Nodes in the network represent proteins, edges between nodes represent protein-protein interactions, and the degree value of a node reflects the number of associations a node has with other nodes. If only a small number of nodes are selected directly based on a single sorting, the screening results are easily affected by the threshold setting. Using a progressive two-level topological threshold preserves the gradual convergence of the candidate set from large to small, forming an intermediate target set that can be verified.

[0008] Functional enrichment analysis and pathway enrichment analysis are used to illustrate the biological functions and signaling pathways that appear frequently in a cluster of candidate targets. Gene ontology functional enrichment analysis typically organizes results from three dimensions: biological processes, cellular components, and molecular functions, while pathway enrichment analysis, as described in the Kyoto Encyclopedia of Genes and Genomes, is used to establish the correspondence between candidate targets and signaling pathways. However, the enrichment entries themselves are still statistical results for a set of targets, and it is necessary to remap the pivotal targets obtained from network screening to the top-ranked pathways to identify candidate core targets that simultaneously possess network centrality and pathway association.

[0009] Molecular docking can search for possible binding conformations given protein and small molecule structures, and output binding energies through a scoring function to compare the calculated binding propensities between different candidate proteins and betulinic acid. Molecular docking primarily reflects static conformations and scoring results under given structural conditions and cannot be directly equated to the effects at the cellular or animal level. Molecular dynamics simulations, on the other hand, can observe conformational changes in protein-ligand complexes within a set simulation time, and supplement the evaluation of dynamic changes in the complex using indicators such as root mean square deviation, root mean square fluctuation, solvent-accessible surface area, number of hydrogen bonds, radius of gyration, and free energy landscape.

[0010] Existing network pharmacology analyses typically include target prediction, intersection analysis, protein-protein interaction network analysis, and enrichment analysis. However, when applied to betulinic acid target screening for ischemic stroke, relying solely on a single database, a single network index, or a single molecular docking result to determine the core target can easily lead to a disconnect between target sources, screening levels, and validation relationships, making it difficult to verify the basis for the gradual narrowing of candidate targets from the initial set to the core set. Therefore, we propose a network pharmacology-based method for betulinic acid target screening for ischemic stroke. Summary of the Invention

[0011] The purpose of this invention is to provide a method for screening betulinic acid targets for ischemic stroke based on network pharmacology, in order to solve the problems of inconsistent names of multiple source targets, difficulty in tracing screening levels, and determination of core targets based on a single calculation indicator.

[0012] To achieve the above objectives, the present invention provides the following technical solution: a method for screening betulinic acid targets for ischemic stroke based on network pharmacology, executed by a computer, comprising the following steps (the data generated in the previous step is used as the basis for processing in the next step, and the data correspondence is maintained according to a unified target identifier): S1. Obtain compound information of betulinic acid and, based on this compound information, obtain a betulinic acid target set from a drug target database (forming an initial target set on the compound side); obtain disease information related to ischemic stroke and, based on this disease information, obtain a ischemic stroke disease target set from a disease target database; wherein the betulinic acid target set is used to characterize betulinic acid-related targets included in or predicted by the database, and the ischemic stroke disease target set is used to characterize disease targets related to the occurrence and development of ischemic stroke. The two target sets are saved separately during the acquisition stage for subsequent steps to unify names and verify correspondence.

[0013] S2. Convert the target names in the betulinic acid target set and the ischemic stroke target set to a unified gene nomenclature system; the gene symbols here are used for unified identification and data matching of targets, and do not indicate that the gene and its encoded protein are considered the same technical object. After deduplicating the two target sets, perform an intersection operation to retain targets that exist in both sets, forming a candidate target set, which is used as input for subsequent protein-protein interaction network analysis; S3. Input the candidate target set into the protein interaction database to obtain protein interaction relationship data and construct a protein-protein interaction network; calculate the topology parameters of each node in the network, and filter them sequentially according to at least two levels of topology thresholds to obtain the hub target set. Specifically, first retain the nodes that meet the conditions according to the first level of topology threshold, and then perform the next level of topology filtering on the network or subnetwork formed by the retained nodes. S4. Perform gene ontology function enrichment analysis and Kyoto Encyclopedia of Genes and Genomes enrichment analysis on at least one of the candidate target set and the hub target set. The former is used to organize related biological functions, and the latter is used to obtain the correspondence between targets and signaling pathways. Select a predetermined number of enriched entries according to the statistical ranking of the enrichment analysis, and establish the mapping relationship between enriched pathways and targets to provide a basis for further screening of candidate core targets. S5. From the targets that are simultaneously part of the hub target set and have a mapping relationship with the enriched pathways obtained from the screening, obtain the candidate core target set. The candidate core targets must simultaneously satisfy the network hub condition and the pathway mapping condition, rather than being determined based on a single topology parameter. S6. Obtain the three-dimensional structure of the protein corresponding to each target in the candidate core target set. Using the corresponding protein as the acceptor and betulinic acid as the ligand, perform molecular docking between betulinic acid and each protein to obtain the binding conformation and binding energy of the protein-betulinic acid complex. Retain candidate complexes with binding energies less than or equal to the preset binding energy threshold. The binding conformation and binding energy are molecular docking results obtained under the given protein structure, ligand structure and calculation model conditions, and are used to compare the calculated binding tendency between betulinic acid and different candidate proteins. S7. The target points corresponding to the candidate complexes retained in step S6 are identified as core targets. Using the target name after conversion to a unified gene nomenclature system as the association key, the topological parameters, mapped enrichment pathways, and binding energies of the corresponding candidate complexes of each core target are associated and recorded, so that network screening, pathway mapping, and molecular docking results are all attributed to the corresponding core targets. At least one candidate complex is selected from the candidate complexes for molecular dynamics simulation, and the kinetic parameters used to characterize the conformational changes of the complex are calculated. Only the selected candidate complexes are generated with corresponding kinetic analysis output, and the kinetic parameters are recorded in the association record of the corresponding core target, outputting the core target screening results of betulinic acid for ischemic stroke. Step S7 adopts a sequential screening and result association method, without weighted summation of topological parameters, enrichment pathways, binding energies, and kinetic parameters; the kinetic parameters are used as further simulation analysis results of the corresponding core targets, and are not used as a criterion for determining whether other core targets that have not undergone molecular dynamics simulation can enter the screening results, nor are they replaced by the kinetic results of other candidate complexes.

[0014] Preferably, the drug target database includes at least one of the TCMSP database and the SwissTargetPrediction database; the disease target database includes at least one of the GeneCards database and the OMIM database; and the protein interaction database is the STRING database. In practice, search terms, access dates, original exported files, and data source fields can be retained to provide a traceable input basis for subsequent unified naming and intersection operations.

[0015] TCMSP and SwissTargetPrediction are used to identify betulinic acid targets, GeneCards and OMIM are used to identify targets for ischemic stroke, and STRING is used to identify protein-drug interactions corresponding to candidate targets. Different databases are used to acquire drug, disease, and protein interaction data respectively, and the output of a single database does not replace all data sources.

[0016] Preferably, in step S2, the names of the two types of target points are converted into unified gene symbols, and the unified gene symbols are used as the comparison key between the two target sets. Duplicate records are removed according to the unified gene symbols, and only gene symbols that exist simultaneously in both the betulinic acid target set and the ischemic stroke target set are retained to form a candidate target set. This reduces the situation where the same target is not identified due to different names or is counted multiple times due to duplicate records. The correspondence between the original name, unified gene symbol and database source of each target in the candidate target set is saved, and the candidate target set is used as the input for constructing the protein-protein interaction network in step S3.

[0017] Preferably, step S3 uses node degree value as a topology parameter. The node degree value reflects the number of associations between the node and other nodes in the current network. The degree value of each node in the current network to be screened is calculated, and the median of the degree values ​​is used to determine the topology threshold of the current level. Each level determines the threshold based on the distribution of node degree values ​​in the current network to be screened at that level, and nodes with degree values ​​greater than the topology threshold of the current level are retained. Based on the sub-network formed by the retained nodes (the subsequent level takes the screening result of the previous level as input and does not constitute a parallel substitution condition with the previous level), the node degree value and its median are recalculated, and at least one subsequent screening is performed according to the re-determined median to obtain a set of hub target points composed of hub target points, thereby forming a progressively advancing screening path. The hub target points are represented by their corresponding unified gene symbols and are used to match the associated genes of the subsequent enrichment pathways. The set of hub target points is used as the input for enrichment analysis and pathway mapping in step S4.

[0018] Preferably, the candidate target set consists of 130 intersection targets, which is the specific output after unified naming, deduplication and intersection operation of the database input; step S3 retains nodes with a degree value greater than 32 in the first-level topology screening, and based on the sub-network formed by the retained nodes, retains nodes with a degree value greater than 67 in the second-level topology screening to obtain a hub target set consisting of 23 hub targets.

[0019] Preferably, the pathway enrichment entries screened in step S4 from the Kyoto Encyclopedia of Genes and Genomes include at least one of cancer pathways, neurodegenerative pathways, PD-L1 / PD-1 pathways, and PPAR signaling pathways, which are used to limit the candidate range of this pathway mapping. Using the unified gene symbol corresponding to the hub target as the association key, it is matched with the list of associated genes corresponding to the pathway enrichment entries. Only targets that simultaneously meet the hub condition and have the selected pathway mapping relationship are retained to obtain a set of candidate core targets including MAPK3, NFKB1, STAT1, PTGS2, and PPARG. The above names serve as the unified identifier of the candidate core targets. The proteins corresponding to each target in the candidate core target set are used as the molecular docking objects in step S6, and each protein serves as the recipient object for subsequent docking tasks.

[0020] Preferably, in step S6, the three-dimensional structure files of the proteins corresponding to the candidate core targets and the three-dimensional structure files of betulinic acid are obtained and used as inputs for the receptor structure and ligand structure, respectively. Receptor-ligand docking tasks are established with the proteins corresponding to the candidate core targets as receptors and betulinic acid as ligands. A docking task is established for each candidate core target, and the receptor, ligand and output results are associated through a unified target identifier. The binding conformation and binding energy of each protein-betulinic acid candidate complex are output for subsequent comparison according to a unified binding energy threshold.

[0021] Preferably, the binding energy of each protein-betulinic acid candidate complex is compared with a preset binding energy threshold. The same threshold is used for judgment in each docking task. When the binding energy is less than or equal to -6.0 kcal / mol, the corresponding protein-betulinic acid candidate complex is retained. The binding energy belongs to the score under the given calculation model and is used to compare and calculate the binding tendency. It is not equivalent to the protein-ligand having been bound in the biological system.

[0022] Preferably, in step S7, the candidate complexes corresponding to PPARG and MAPK3 are selected from the candidate complexes, and molecular dynamics simulations are performed on them for 100 ns to observe the changes in the conformation of the protein, ligand, and complex during the simulation time. The dynamic parameters include at least one of root mean square deviation, root mean square fluctuation, solvent-accessible surface area, number of hydrogen bonds, radius of gyration, and Gibbs free energy landscape, which respectively form analytical information from at least one aspect of overall offset, local fluctuation, solvent exposure, hydrogen bond formation, structural compactness, and free energy distribution.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows: The method of this application obtains betulinic acid target and ischemic stroke target respectively, and performs deduplication and intersection operation after unified gene naming. Records from different databases can be compared according to the same target identifier. Then, at least two levels of topological thresholds are used to sequentially reduce the candidate nodes in the protein interaction network, so that the process of candidate target convergence from the intersection set to the hub target set retains a clear intermediate level.

[0024] Furthermore, after cross-mapping enrichment pathways with hub targets, candidate core targets simultaneously possess network topology and pathway correlation evidence. Subsequently, the binding conformation and binding energy of candidate complexes are obtained through molecular docking, and at least one candidate complex is selected for molecular dynamics simulation, so that the static scoring and the kinetic analysis results of the selected candidate complexes can be mapped to the aforementioned screening process.

[0025] Furthermore, the target sets, topological parameters, enrichment entries, pathway mapping relationships, and molecular simulation results at all levels are interconnected according to a unified target identifier. This allows for the reverse tracing of data sources and screening processes from core targets, and provides a computational basis for selecting targets that require in vitro or in vivo validation. Attached Figure Description

[0026] Figure 1 This is a graph showing the intersection of the target set of betulinic acid and the target set of ischemic stroke. Figure 2 A diagram showing the protein-protein interaction network and the results of the two-level topology screening. Figure 3 A graph showing the results of gene ontology function enrichment analysis; Figure 4 A graph showing the results of pathway enrichment analysis from the Kyoto Encyclopedia of Genes and Genomes. Figure 5 A diagram showing the cross-mapping results between enriched pathways and pivotal targets; Figure 6 This is a molecular docking conformation diagram of betulinic acid with the protein corresponding to the candidate core target. Figure 7 A diagram showing the structure numbering, binding energy, and interaction information of betulinic acid with the corresponding proteins of the candidate core targets; Figure 8 The figures show the molecular dynamics simulation results of the PPARG-betulinic acid candidate complex and the MAPK3-betulinic acid candidate complex. Detailed Implementation

[0027] The present invention will be further described below with reference to the embodiments. The database name, number of target points, structure number, binding energy and simulation time used in the embodiments are all information actually used or output in this analysis.

[0028] For ease of expression and understanding, some basic dataset symbols are defined in the specification of this application.

[0029] For example, the so-called betulinic acid target set T1 refers to the set of targets that are potentially associated with betulinic acid, predicted or included in the compound target database. The target set T2 for ischemic stroke refers to the set of targets in the disease database that are related to ischemic stroke or cerebral ischemia. The candidate target set T3 refers to the set formed by unifying the names, removing duplicates, and taking the intersection of the betulinic acid target set T1 and the ischemic stroke target set T2. The pivotal target set T4 refers to the set formed after the candidate target set T3 is screened through the protein interaction network topology; The candidate core target set T5 refers to the set formed by further cross-mapping the hub target set T4 with the enriched pathways obtained from screening.

[0030] Among them, hub targets refer to protein targets that meet the topological screening conditions in the protein-protein interaction network. Each hub target is recorded and matched with its corresponding unified gene symbol. The unified gene symbol is an association identifier in data processing, and the concepts of gene, its encoded protein and target are not regarded as completely synonymous.

[0031] It should be noted that the calculation results referred to in this specification include target records exported from the database, target tables with standardized naming, set operation results, network node and edge data, topological parameters, enrichment entries, pathway-target mapping relationships, molecular docking conformations, binding energies, and molecular dynamics analysis indicators. These results are outputs of data processing or molecular simulation and have different evidentiary properties than wet experimental results such as cell viability, protein expression, and cerebral infarction volume.

[0032] Example 1: This example addresses the issue that betulinic acid targets and ischemic stroke targets originate from different databases, have inconsistent naming conventions, and involve a large number of candidates. To ensure a verifiable intermediate level in the screening process, this example sequentially completes target acquisition, name unification, intersection calculation, protein-protein interaction network construction, and two-level topology screening.

[0033] S1. On a computer, search the TCMSP and SwissTargetPrediction databases for betulinic acid compound information and export the predicted or included targets. Simultaneously, search the GeneCards and OMIM databases using "ischemic stroke" and "cerebral ischemia" as disease search terms and export disease-related targets. The results from different databases retain their source fields and are not directly mixed after export.

[0034] The drug target data corresponding to the betulinic acid target set T1 and the disease target data corresponding to the ischemic stroke target set T2 are established as separate data tables. Each record should at least correspond to its original target name and database source, so that after unified naming, it is still possible to trace back to the source of the target from drug target search, disease target search, or both. When the same database returns multiple records pointing to the same target, the original records are retained first, and then deduplication is performed according to the unified gene symbol after the name conversion is completed.

[0035] The use of "ischemic stroke" and "cerebral ischemia" as disease search terms is to maintain consistency with the disease research object of this invention. The disease targets obtained from the two search terms are first merged and deduplicated in the disease target data table to form the ischemic stroke disease target set T2, without using the frequency of search terms as a criterion for judging the importance of targets; then, an intersection operation is performed with the betulinic acid action target set T1.

[0036] S2. The targets in the betulinic acid target set T1 and the ischemic stroke target set T2 were converted to uniform gene symbols. Duplicate records were removed, and then the set intersection operation was performed. A total of 130 intersection targets were obtained in this calculation. See [link to results]. Figure 1 These 130 targets constitute the candidate target set T3 and serve as input for subsequent protein-protein interaction analysis.

[0037] When standardizing naming, the drug target table and the disease target table use the same gene symbol as the comparison key. If multiple original names are converted to the same gene symbol, they are merged into a single target record in the corresponding data table, while retaining their original source information. After deduplication, only gene symbols that exist in both the betulinic acid target set T1 and the ischemic stroke target set T2 are written into the data table corresponding to the candidate target set T3.

[0038] Figure 1The set relationship represents the target set T1 of betulinic acid, the target set T2 of ischemic stroke, and the candidate target set T3 formed by their intersection. 130 represents the number of targets obtained from this intersection operation. This number is the input scale for the subsequent construction of the protein interaction network and does not mean that all 130 targets have been confirmed to have a direct biological relationship with the neuroprotective effect of betulinic acid.

[0039] S3. Input the candidate target set T3 into the STRING database to obtain the protein-protein interaction relationships, and then construct a protein-protein interaction network in Cytoscape. Nodes represent the proteins corresponding to the candidate targets, and edges represent the interaction relationships returned by the database. Calculate the degree value of each node in the current network to be screened, and determine the topological threshold of this level based on the median of the node degree values. Nodes with degree values ​​greater than the topological threshold of this level are retained. Then, a sub-network is formed using the retained nodes, and the node degree values ​​and their medians are recalculated. Subsequent screening is performed based on the newly determined median. In this embodiment, this two-stage screening process is manifested as first retaining nodes with degree values ​​greater than 32, and then screening nodes with degree values ​​greater than 67 from the retained nodes, finally obtaining a hub target set T4 consisting of 23 hub targets. Each hub target is recorded with a corresponding unified gene symbol. The screening process is described in [reference needed]. Figure 2 .

[0040] The degree of a node is the number of edges directly connected to that node. In the same protein-protein interaction network, nodes with higher degree values ​​interact with more candidate proteins, and therefore can be used to form a set of hub nodes that require further analysis. The degree value only reflects network connectivity and does not directly indicate the expression level or regulatory strength of the corresponding protein in cell or animal models.

[0041] The two-stage screening is performed sequentially. The first stage filters 130 candidate nodes in the candidate target set T3, selecting nodes with a retention rate greater than 32. The second stage only filters the subnetwork formed by the nodes retained in the first stage, selecting nodes with a retention rate greater than 67, thus forming a hub target set T4 consisting of 23 hub targets. This embodiment uses the median filtering method and completes the screening according to the above two-stage conditions.

[0042] This analysis also used the CytoHubba plugin for Cytoscape to sort hub targets according to the MCC algorithm; the MCC score was used as a verification indicator for node sorting after two-level degree value screening, and was recorded separately from the degree value screening results.

[0043] S4. Gene ontology function enrichment analysis and pathway enrichment analysis were performed on 130 candidate targets in candidate target set T3 and / or 23 hub targets in hub target set T4. Gene ontology function enrichment results were categorized according to biological process, cellular component, and molecular function, and the top 10 statistically ranked items were output. See [link to results]. Figure 3 The pathway enrichment results were selected from the top 20 items according to the statistical ranking of the analysis output. See the results below. Figure 4 .

[0044] In the gene ontology function enrichment results, biological processes are used to represent the biological activities involved by candidate targets, cellular components are used to represent the cellular structures in which candidate target-related products are located or function, and molecular functions are used to represent the molecular-level functions that the corresponding gene products may perform. This result includes entries related to inflammation, cell cycle, and metabolism, but these entries indicate the direction of enrichment of the candidate target set and do not necessarily mean that betulinic acid has already regulated the corresponding processes in the experiment.

[0045] Pathway enrichment analysis establishes a correspondence between candidate targets and pathway entries included in the database. Figure 4 The results shown include cancer pathways, neurodegenerative pathways, PD-L1 / PD-1 pathways, and PPAR signaling pathways. In this embodiment, the top 20 pathway entries are used as the pathway set for subsequent cross-mapping; entries not included in the display set can still be retained in the complete enrichment results table for review and analysis output.

[0046] S5. Cross-mapping was performed on the 23 hub targets in hub target set T4 with the selected pathway entries to establish the correspondence between each pathway and the hub targets it contains. This mapping involved cancer pathways, neurodegenerative pathways, PD-L1 / PD-1 pathways, and PPAR signaling pathways. Targets that simultaneously met the hub target criteria and corresponded to the above-mentioned enriched pathways were retained as the candidate core target set T5, resulting in five candidate core targets: MAPK3, NFKB1, STAT1, PTGS2, and PPARG. The mapping results are shown below. Figure 5 .

[0047] During cross-mapping, the unified gene symbol corresponding to the hub target in the hub target set T4 is used as the association key to check whether each hub target appears in the associated gene list of the selected pathway. When a hub target establishes a correspondence with one or more screening pathways, the gene and its pathway name are written into the pathway-target mapping table; hub targets that do not establish a correspondence with screening pathways are not included in the current candidate core target set T5.

[0048] Figure 5This is used to represent the correspondence between pathways and core targets. One target can correspond to multiple pathways simultaneously, and one pathway can also correspond to multiple targets. This embodiment does not interpret all candidate core targets with a single pathway, but retains the mapping relationship between multiple targets and multiple pathways so that subsequent molecular simulation results can still be associated with network and pathway data according to the same target identifier.

[0049] The data output in this embodiment includes at least the following: the original target tables corresponding to the betulinic acid target set T1 and the ischemic stroke target set T2; the target table with unified naming; the list of 130 intersection targets corresponding to the candidate target set T3; the protein-protein interaction edge table; the node degree value table; the two-level threshold screening records; the list of 23 hub targets corresponding to the hub target set T4; the function and pathway enrichment result table; and the pathway-target mapping table. These intermediate files collectively determine the source of the five candidate core targets in the candidate core target set T5, and do not replace the above calculation process with animal efficacy results.

[0050] For ease of verification, the candidate core target output records can list whether a target belongs to one of the 130 overlapping targets, whether it has passed the two-level degree screening, the corresponding enrichment pathway, and whether it has entered molecular docking. All of these fields are filled in by the actual output of the preceding steps, and no inferences are made regarding missing analytical results. Through this correspondence, the origins of the initial drug and disease targets can be traced back from the final five targets.

[0051] Example 2 illustrates the molecular simulation verification process for specific candidate core targets. While the network locations and pathway associations obtained in Example 1 are used to reflect bioinformatics relationships, they do not yet provide the calculated binding between betulinic acid and the corresponding target protein. Therefore, after obtaining a candidate core target set T5 containing five candidate core targets, this example continues with molecular docking and selects at least one candidate complex for molecular dynamics simulation.

[0052] Specifically, the steps are as follows: S6. Obtain the three-dimensional structures of the proteins corresponding to the five candidate core targets in the candidate core target set T5 from the PDB database. The structure numbers used for NFKB1, MAPK3, PPARG, STAT1, and PTGS2 are 1SVC, 4QTB, 2PRG, 1BF5, and 5F19, respectively. AutoDock Vina is a molecular docking software used for conformational search and scoring given the three-dimensional structures of the receptor and ligand; in this embodiment, AutoDock Vina is used to perform molecular docking of betulinic acid with each protein, save the binding conformation, and output the binding energy. See [link to molecular docking conformation]. Figure 6 .

[0053] The PDB structure number is used to uniquely correspond to the 3D protein structure file read in this docking operation. Five docking tasks are established according to the correspondence between candidate core targets and structure numbers. Each task uses the same compound, betulinic acid, as the ligand and the corresponding candidate core target protein structure as the acceptor. The docking tasks independently output binding conformation and binding energy; the docking result of one target is not used to substitute for the results of other targets.

[0054] The binding conformation is used to represent the position and orientation of betulinic acid relative to the protein structure in the computational model, and the binding energy is used to score the conformation. Figure 6 Five sets of simulated molecular docking conformations are presented. Figure 7 Provide the corresponding structure number and binding energy.

[0055] The binding energies of NFKB1, MAPK3, PPARG, STAT1, and PTGS2 with betulinic acid were calculated to be -6.9 kcal / mol, -7.4 kcal / mol, -7.5 kcal / mol, -7.8 kcal / mol, and -8.0 kcal / mol, respectively. All of these are less than or equal to -6.0 kcal / mol, indicating a good calculated binding tendency under the adopted docking model. Therefore, these five target sites were retained as priority targets for further study. (See [link to binding energy results]). Figure 7 The values ​​above are outputs of the calculation model and should not be used as results of in vitro binding assays or in vivo efficacy studies.

[0056] The binding energy threshold was determined as follows: when the binding energy was less than or equal to -6.0 kcal / mol, the corresponding protein-betulinic acid complex was retained; when the binding energy was greater than -6.0 kcal / mol, the complex was not included in the retention set formed based on this threshold. Since all five docking results met the retention conditions, the complexes corresponding to NFKB1, MAPK3, PPARG, STAT1, and PTGS2 were all retained in the molecular docking results. A smaller docking binding energy value only indicates a lower calculated score under this docking scoring system. This ranking can be used to select subsequent analysis targets, but it cannot directly deduce the actual binding strength between the protein and betulinic acid, the direction of target regulation, or the magnitude of animal efficacy. Therefore, this embodiment treats molecular docking as a computational verification step after network pharmacology screening, rather than writing it as a biological experimental confirmation.

[0057] S7. The NFKB1, MAPK3, PPARG, STAT1, and PTGS2 corresponding to the five candidate complexes retained in step S6 are identified as core targets. Using the unified gene symbol corresponding to each core target as the association key, their node topology parameters, mapped enrichment pathways, and binding energies of the corresponding candidate complexes are linked and recorded. Among the five candidate complexes, the protein-betulinic acid candidate complex corresponding to PPARG and the protein-betulinic acid candidate complex corresponding to MAPK3 are selected, and 100 ns molecular dynamics simulations are performed using GROMACS. The protein-betulinic acid complex is placed in a cubic periodic simulation box, with the minimum distance between the complex and the simulation box boundary set to 1.0 nm. The system is solvated using the SPC water model, and Na is added. + and Cl - The system charge was neutralized, and the ion concentration was brought to 0.15 mol / L. Energy minimization and NPT pre-equilibrium were then performed, followed by formal molecular dynamics simulations. The simulation outputs the trajectory file of the protein-ligand complex, and the root mean square deviation, root mean square fluctuation, solvent accessible surface area, number of hydrogen bonds, radius of gyration, and Gibbs free energy landscape are calculated. See [link to relevant results] for details. Figure 8 .

[0058] 100 ns represents the simulation time for each selected candidate complex. The molecular dynamics program records the changes in the complex structure over time according to the simulated trajectory and extracts the corresponding analytical parameters of the protein, ligand, or complex from the trajectory. In this embodiment, the protein-betulinic acid candidate complex corresponding to PPARG and MAPK3 has the output of this kinetic analysis, while the other three candidate complexes retain the corresponding molecular docking output.

[0059] Root mean square deviation (RMSD) represents the overall offset of the simulated structure relative to the reference structure over time; RMSD fluctuation represents the degree of local fluctuation at different residue positions during the simulation. They describe the simulation trajectory from the perspectives of overall conformational change and local residue change, respectively, and cannot be substituted for each other.

[0060] Solvent-accessible surface area is used to represent the change in the surface area of ​​the complex's related structures exposed to the solvent; the number of hydrogen bonds is used to represent the change in the hydrogen bond relationship between the protein and ligands during the simulation; and the radius of gyration is used to reflect the overall compactness of the structure. The changes of these indicators over time are used together to observe the structural state of the complex, rather than drawing a conclusion of stability or instability directly from a single value.

[0061] The Gibbs free energy landscape represents the relative free energy distribution in the simulated trajectory based on selected conformational variables and can be used to identify low free energy conformational regions that occur during the simulation. In this embodiment, the free energy landscape, along with root mean square deviation, root mean square fluctuation, solvent accessible surface area, number of hydrogen bonds, and radius of gyration, constitutes the kinetic analysis output.

[0062] For candidate complexes that have undergone molecular dynamics simulations, stability is not determined directly based on a single time point or a single parameter. Instead, the dynamic behavior of the selected candidate complexes is described by the trend of root mean square deviation over time, residue fluctuation distribution, solvent exposure changes, hydrogen bond number changes, structural compactness, and low free energy conformation distribution. The above dynamic analysis is only used to form further simulation analysis records for the corresponding core targets and does not participate in the selection of other core targets.

[0063] In specific evaluations, first check whether there are continuous changes, periodic fluctuations, or abnormal mutations in each indicator of the trajectory, and then compare the correspondence between the protein, ligand, and complex curves. When a certain indicator fluctuates, it is also necessary to combine it with other indicators and the corresponding time interval for interpretation, so as to avoid directly stating that the local changes of a single curve have formed a stable biological binding between betulinic acid and the target.

[0064] In the final output, for any target in the candidate core target set T5 of Example 1, it is first determined whether its corresponding protein-betulinic acid candidate complex meets the binding energy retention condition set in step S6; if the retention condition is met, the target corresponding to the candidate complex is determined as a core target. Subsequently, using a unified gene symbol as the association key, the node topology parameters, the mapped enrichment pathway, and the molecular docking binding energy of the core target are written into the same screening result record; for candidate complexes that have undergone molecular dynamics simulation, the corresponding dynamic parameters are further written. Core targets that have not undergone molecular dynamics simulation retain the network topology, pathway mapping, and molecular docking results, and are not replaced by the dynamic parameters of other core targets.

[0065] In summary, this implementation method uses betulinic acid and ischemic stroke as inputs. Through unified nomenclature, intersection operation, protein-protein interaction network topology screening, and enrichment pathway mapping, 23 pivotal targets were screened from 130 candidate targets, and five candidate core targets, namely MAPK3, NFKB1, STAT1, PTGS2, and PPARG, were further obtained. The binding energies of the proteins corresponding to these five candidate core targets and betulinic acid are -7.4 kcal / mol, -6.9 kcal / mol, -7.8 kcal / mol, -8.0 kcal / mol, and -7.5 kcal / mol, respectively, all of which meet the preset candidate complex retention conditions. Therefore, MAPK3, NFKB1, STAT1, PTGS2, and PPARG were identified as core targets. Among them, the candidate complexes corresponding to PPARG and MAPK3 were further analyzed using 100 ns molecular dynamics simulations and related parameter analyses. These dynamic parameters were recorded as further simulation analyses of the corresponding core targets and were not used as selection criteria for other core targets. Therefore, this method ultimately yields a hierarchical screening result consisting of core targets, pathway-target mapping relationships, molecular docking results, and kinetic analysis results of some core targets. This result can be used to determine the priority validation targets and subsequent experimental sequence in the study of betulinic acid's mechanism of action against ischemic stroke.

[0066] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for screening target points of betulinic acid against ischemic stroke based on network pharmacology, characterized in that, Performed by a computer, including the following steps: S1. Obtain compound information of betulinic acid and, based on this compound information, obtain the target set of betulinic acid from the drug target database; Obtain disease information related to ischemic stroke, and obtain a set of disease targets for ischemic stroke from a disease target database based on this disease information; S2. Convert the target names in the betulinic acid target set and the ischemic stroke target set to a unified gene nomenclature system, remove duplicates, and perform intersection operation to obtain the candidate target set. S3. Input the candidate target set into the protein interaction database to obtain protein interaction relationship data and construct a protein-protein interaction network; calculate the topological parameters of each node in the network, and filter them sequentially according to at least two levels of topological thresholds to obtain the hub target set. S4. Perform gene ontology function enrichment analysis and Kyoto Encyclopedia of Genes and Genomes enrichment analysis on at least one of the candidate target set and hub target set. Select a preset number of enrichment entries according to the statistical ranking of the enrichment analysis and establish the mapping relationship between enriched pathways and targets. S5. Obtain a set of candidate core targets from targets that are simultaneously part of the hub target set and have a mapping relationship with the enriched pathways obtained through screening. S6. Obtain the three-dimensional structure of the protein corresponding to each target in the candidate core target set, perform molecular docking of betulinic acid with each protein to obtain the binding conformation and binding energy of the protein-betulinic acid complex, and retain the candidate complexes whose binding energy is less than or equal to the preset binding energy threshold. S7. The target points corresponding to the candidate complexes retained in step S6 are identified as core targets. Using the target names after conversion to a unified gene nomenclature system as the association key, the topological parameters of each core target point, the mapped enrichment pathways, and the binding energies of the corresponding candidate complexes are associated and recorded. At least one candidate complex is selected from the candidate complexes for molecular dynamics simulation. The kinetic parameters used to characterize the conformational changes of the complex are calculated and recorded in the association record of the corresponding core target. The core target screening results of betulinic acid for ischemic stroke are output.

2. The network pharmacology-based target screening method for anti-ischemic stroke of betulinic acid according to claim 1, characterized in that, The drug target database includes at least one of the TCMSP database and the SwissTargetPrediction database; the disease target database includes at least one of the GeneCards database and the OMIM database. The protein-protein interaction database is the STRING database. 3.The network pharmacology-based target screening method for anti-ischemic stroke of betulinic acid according to claim 1, characterized in that, Step S2 converts the two types of target names into unified gene symbols, removes duplicate records according to the unified gene symbols, and retains only the gene symbols that exist in both the betulinic acid target set and the ischemic stroke target set to form a candidate target set. Save the correspondence between the original name, uniform gene symbol and database source of each target in the candidate target set, and use the candidate target set as input for constructing the protein-protein interaction network in step S3. 4.The network pharmacology-based target screening method for anti-ischemic stroke of betulinic acid according to claim 1, characterized in that, Step S3 uses the node degree value as a topology parameter to calculate the degree value of each node in the current network to be screened, and determines the current topology threshold based on the median of the degree values, retaining nodes with a degree value greater than the current topology threshold. Based on the subnetwork formed by the retained nodes, the node degree values ​​and their medians are recalculated, and at least one subsequent screening is performed according to the recalculated median to obtain a set of hub target points composed of hub target points. The hub targets are represented by their corresponding unified gene symbols, and the set of hub targets is used as the input for enrichment analysis and pathway mapping in step S4.

5. The network pharmacology-based target screening method for anti-ischemic stroke of betulinic acid according to claim 4, characterized in that, The candidate target set consists of 130 intersecting targets; Step S3 retains nodes with a retention value greater than 32 in the first-level topology screening, and based on the sub-network formed by the retained nodes, retains nodes with a retention value greater than 67 in the second-level topology screening, thus obtaining a hub target set consisting of 23 hub target points. 6.The network pharmacology-based target screening method for anti-ischemic stroke of betulinic acid according to claim 1, characterized in that, The pathway enrichment entries screened in step S4 from the Kyoto Encyclopedia of Genes and Genomes include at least one of the following: cancer pathways, neurodegenerative pathways, PD-L1 / PD-1 pathways, and PPAR signaling pathways. Using the unified gene symbol corresponding to the hub target as the association key, it is matched with the list of associated genes corresponding to the enriched pathway entries to obtain a set of candidate core targets including MAPK3, NFKB1, STAT1, PTGS2 and PPARG, and the protein corresponding to each target in the candidate core target set is used as the molecular docking object in step S6.

7. The network pharmacology-based target screening method for anti-ischemic stroke of betulinic acid according to claim 1, characterized in that, Step S6: Obtain the three-dimensional structure files of the protein corresponding to the candidate core target and the three-dimensional structure file of betulinic acid. Establish receptor-ligand docking tasks with the protein corresponding to the candidate core target as the receptor and betulinic acid as the ligand, and output the binding conformation and binding energy of each protein-betulinic acid candidate complex. 8.The network pharmacology-based target screening method for anti-ischemic stroke of betulinic acid according to claim 7, characterized in that, The binding energy of each protein-betulinic acid candidate complex was compared with a preset binding energy threshold. When the binding energy was less than or equal to -6.0 kcal / mol, the corresponding protein-betulinic acid candidate complex was retained.

9. The method for screening betulinic acid targets for ischemic stroke based on network pharmacology according to claim 6, characterized in that, Step S7: Select the PPARG-betulinic acid candidate complex and the MAPK3-betulinic acid candidate complex from the candidate complexes, and perform 100 ns molecular dynamics simulations on each. The dynamic parameters include at least one of the following: root mean square deviation, root mean square fluctuation, solvent accessible surface area, number of hydrogen bonds, radius of gyration, and Gibbs free energy landscape.