Network pharmacological analysis methods, systems and predictive models for health wine

Abstract

This invention provides a network pharmacological analysis method, system, and prediction model for health-enhancing wines. The method includes acquiring multi-dimensional parameters of the active ingredients in the health-enhancing wine. These parameters include exposure-related parameters characterizing the in vivo exposure level of the active ingredients, binding-related parameters characterizing the binding strength between the active ingredients and the target, and solvent effect-related parameters characterizing the regulatory effect of the solvent ethanol on the target. Based on these multi-dimensional parameters, an active ingredient-target weighted network is constructed. Through analysis of the weighted network, the potential functions of the health-enhancing wine or its active ingredients are predicted. This invention solves the problem that existing methods cannot handle the bidirectional regulatory effect and concentration-dependent effect of ethanol, which leads to the inability to distinguish functional differences at different drinking doses.

Description

Technical Field

[0001] This invention relates to the field of network pharmacology analysis, and in particular to a method, system and prediction model for network pharmacology analysis of health wine. Background Technology

[0002] Health-preserving wines are ethanol-containing beverages made by extracting medicinal plants and animals using baijiu (Chinese white liquor) or edible alcohol as a base. They have a long history of application. Network pharmacology, a commonly used tool for systematically analyzing the mechanisms of action of complex systems of traditional Chinese medicine, has been attempted to be applied to the study of health-preserving wines in recent years. However, existing network pharmacology methods have the following technical shortcomings when applied to the analysis of health-preserving wines: they rely solely on the qualitative existence of components to determine their association with targets, ignoring the differences in the actual content of components in health-preserving wines and the differences in oral bioavailability; they fail to quantify the binding strength between components and targets, treating weak and strong bindings equally; they completely ignore the regulatory effect of the health-preserving wine solvent ethanol on disease-related targets, or they simply presuppose ethanol as a single negative interfering factor, failing to distinguish its positive synergistic effect at low concentrations from its negative interfering effect at high concentrations.

[0003] More importantly, existing technologies lack network modeling methods for targets with "low concentration protection and high concentration damage." The few studies that have realized the influence of ethanol concentration only use binary threshold models or linear assumptions, which deviate significantly from the true S-type dose-response relationship. This makes it impossible to distinguish the function prediction of the same health wine at different drinking doses, which seriously restricts the clinical translation value of the analysis results.

[0004] Therefore, how to quantify the bidirectional regulatory effect of ethanol on the efficacy of health wine at different concentrations, so as to achieve accurate functional prediction of different drinking doses, has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] The purpose of this invention is to provide a network pharmacological analysis method, system, and prediction model for health wines, in order to solve the problem that existing solutions cannot handle the bidirectional regulatory effect and concentration-dependent effect of ethanol, which leads to the inability to distinguish functional differences under different drinking doses.

[0006] To achieve the above objectives, the present invention is implemented as follows: In a first aspect, the present invention provides a method for network pharmacological analysis of health wine, comprising: The multidimensional parameters of the functional components in health wine are obtained. The multidimensional parameters include exposure-related parameters characterizing the exposure level of the functional components in vivo, binding-related parameters characterizing the binding strength between the functional components and the target, and solvent effect-related parameters characterizing the regulatory effect of the solvent ethanol on the target. Based on the aforementioned multi-dimensional parameters, a weighted network of efficacy components and targets is constructed. By analyzing the weighted network, the potential functions of health wine can be predicted or its effective components can be identified.

[0007] Furthermore, the method may be applied in any of the following: (a) Screening key efficacy components with high contribution in health wines; (b) Predicting potential new functions of health wines; (c) Evaluate the balance between efficacy and risk of health wines with different ethanol concentrations.

[0008] Secondly, the method described in the first aspect is provided for the application in preparing efficacy evaluation models for health wines or screening key efficacy components of health wines.

[0009] Thirdly, a network pharmacological analysis system for health wine is provided, including: The parameter acquisition module is used to acquire multi-dimensional parameters of the functional components in the health wine. The multi-dimensional parameters include exposure-related parameters characterizing the exposure level of the functional components in vivo, binding-related parameters characterizing the binding strength between the functional components and the target, and solvent effect-related parameters characterizing the regulatory effect of the solvent ethanol on the target. The network construction module is used to construct an efficacy component-target weighted network based on the multi-dimensional parameters. The analysis and prediction module is used to predict the potential functions of health wine or analyze its effective components by analyzing the weighted network.

[0010] Fourthly, a predictive model for the function of health wine is provided, the model being stored in the form of a data structure in a computer-readable storage medium, including: The functional component node feature library is used to store exposure-related parameters for each functional component; A target node feature library is used to store solvent effect-related parameters for each target, which are dynamically determined based on an ethanol concentration-effect continuous function. The edge weight matrix is ​​used to store the binding-related parameters of each functional component-target pair; The weights of the edges connecting the functional component nodes and the target nodes in the model are generated according to a preset fusion rule by the exposure-related parameters of the corresponding functional components, the binding-related parameters of the corresponding component-target pairs, and the solvent effect-related parameters of the corresponding targets.

[0011] Fifthly, a method for predicting the function of health wine is provided, including: A pre-built weighted network model for health wine is provided. The model includes nodes of functional components, nodes of target points, and edges connecting the nodes. Each edge has a basic weight, which is determined by the exposure-related parameters of the corresponding functional component and the binding-related parameters of the corresponding component-target pair. Obtain the ethanol concentration data of the health wine to be predicted; Based on the ethanol concentration data, the solvent effect-related parameters for each target in the model are determined; The basic weights are updated based on the solvent effect-related parameters to obtain the updated edge weights; Based on the updated edge weights, a topology analysis is performed on the weighted network model to predict the potential functions of the health wine to be predicted.

[0012] In a sixth aspect, the present invention also provides a computer-readable storage medium on which a computer program is stored, wherein the computer program, when executed by a processor, implements the steps of the method described in the first aspect.

[0013] The beneficial effects of this invention are as follows: This invention's method for network pharmacological analysis of health wines introduces solvent effect-related parameters characterizing the regulatory effect of solvent ethanol on targets. For the first time, it incorporates the bidirectional regulatory effect of ethanol into the quantitative framework of network pharmacological analysis of health wines, directly solving the technical problem of existing technologies being unable to handle ethanol concentration-dependent effects. Unlike existing technologies that presuppose ethanol as a single negative factor, this invention uses solvent effect-related parameters to characterize the positive synergistic or negative interference effects of ethanol on targets at different concentrations. This allows the constructed weighted network to dynamically reflect the impact of ethanol concentration changes on the efficacy of health wines, thereby achieving discriminative prediction of the functions of health wines at different drinking doses and overcoming the inherent defects of traditional methods that do not differentiate between doses and have ambiguous efficacy. Simultaneously, this invention introduces exposure-related parameters and binding-related parameters to address the problems of existing schemes neglecting differences in in vivo component exposure and target binding strength. Exposure-related parameters integrate component content and oral bioavailability information, upgrading network analysis from static in vitro chemical concentration to dynamic in vivo accessibility assessment; binding-related parameters quantify the affinity differences between components and targets, giving the network edge weights clear biological significance. By integrating the above three types of parameters to construct a weighted network, this invention achieves a technological leap from qualitative correlation to multidimensional quantification, significantly improving the accuracy of the analysis of the efficacy components of health wine and the reliability of the prediction of new functions. This provides a scientific basis for the rational development and clinical translation of health wine, thereby solving the problem that existing solutions cannot handle the bidirectional regulatory effect and concentration-dependent effect of ethanol, which leads to the inability to distinguish functional differences under different drinking doses.

[0014] Furthermore, this invention employs a tiered docking strategy in the acquisition of relevant parameters. Through a two-stage design of initial screening and precise docking, high-cost molecular docking calculations are performed only on the top 10%-20% of component-target pairs in the initial screening score, while the weights of the remaining pairs are assigned to 0 or preset minimum values. In this way, while ensuring the accuracy of core target identification, the number of component-target pairs requiring molecular docking calculations is reduced by more than 80%, compressing the analysis cycle for a single health wine variety from several weeks in traditional methods to within a few days. This significantly lowers the implementation threshold and computational cost of high-throughput network pharmacology methods. Unlike the conventional approach of full docking or random sampling in existing technologies, this invention achieves synergistic optimization of computational accuracy and implementation efficiency through a scientific tiered screening mechanism. This enables highly complex analyses that originally relied on supercomputing resources to run efficiently on conventional servers or cloud hosts, providing a feasible technical path for the industrialization and promotion of network pharmacology methods in small and medium-sized enterprises.

[0015] Furthermore, this invention employs the Hill equation as a continuous function of ethanol concentration-effect in the generation of solvent effect-related parameters, introducing the widely accepted receptor occupation theory in pharmacology into the framework of network pharmacology analysis. Unlike the binary threshold model or linear assumptions used in existing technologies, the Hill equation naturally satisfies the biological constraints of f(0)=0, value range [0,1], and monotonically increasing. Its parameters EC50 and Hill coefficient n have clear pharmacological significance and can be directly correlated with experimental data reported in the literature. Through this equation, this invention achieves a continuous, smooth, and saturable mathematical expression of the regulatory effect of ethanol concentration on targets, accurately depicting the true dose-effect relationship of "slow activation at low concentration - rapid response at medium concentration - saturation at high concentration." For concentration-dependent bidirectional targets such as NLRP3 and AKT1, this invention further incorporates piecewise directional symbols, enabling the network model to accurately distinguish between the protective effect at low concentrations and the damaging effect at high concentrations. This solves the technical problem that existing technologies cannot quantify the core characteristics of health wines, which are "beneficial in moderation and harmful in excess," providing a reliable mathematical basis for functional prediction at different drinking doses. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of a network pharmacological analysis method for health wine according to an embodiment of the present invention; Figure 2 This is a schematic flowchart of a network pharmacological analysis method for health wine according to another embodiment of the present invention; Figure 3 This is a schematic flowchart of a network pharmacological analysis method for health wine according to another embodiment of the present invention; Figure 4 This is a schematic flowchart of a network pharmacological analysis method for health wine according to another embodiment of the present invention; Figure 5This is a schematic flowchart of a network pharmacological analysis method for health wine according to another embodiment of the present invention; Figure 6 This is a schematic flowchart of a network pharmacological analysis method for health wine according to another embodiment of the present invention; Figure 7 This is a schematic flowchart of a network pharmacological analysis method for health wine according to another embodiment of the present invention; Figure 8 This is a schematic structural block diagram of a network pharmacological analysis system for health wine according to an embodiment of the present invention; Figure 9 This is a topology diagram of a computer-readable storage medium disclosed in this invention. Detailed Implementation

[0017] The present invention will now be described in detail with reference to the embodiments shown in the accompanying drawings. However, it should be noted that these embodiments are not intended to limit the present invention. Equivalent changes or substitutions in function, method, or structure made by those skilled in the art based on these embodiments are all within the scope of protection of the present invention.

[0018] The technical solutions provided by the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0019] Example 1: like Figure 1 As shown, this embodiment provides a network pharmacological analysis method for health wines (hereinafter referred to as the "method" or "analysis method"), which is used in a network pharmacological analysis system for health wines or a functional prediction model for health wines. The method includes: Step 102. Obtain multi-dimensional parameters of the functional components in the health wine. These multi-dimensional parameters include exposure-related parameters characterizing the exposure level of the functional components in vivo, binding-related parameters characterizing the binding strength between the functional components and the target, and solvent effect-related parameters characterizing the regulatory effect of the solvent ethanol on the target.

[0020] This embodiment is based on the fundamental principles of multi-dimensional weighted network pharmacology, and constructs a quantitative basis for the efficacy analysis of health wine by introducing three types of parameters with clear biological significance. Among them, exposure-related parameters, based on pharmacokinetic principles, characterize the probability and extent to which the active ingredient enters the systemic circulation via oral administration; combination-related parameters, based on molecular recognition and interaction principles, quantify the affinity differences between the active ingredient and biological targets; and solvent effect-related parameters, based on receptor-ligand interaction theory and ethanol pharmacology principles, characterize the regulatory effect of the solvent ethanol on disease-related targets and its concentration-dependent characteristics. These three types of parameters comprehensively characterize the action process of the active ingredients in health wine from three levels: whether the target can be reached, whether the target can be bound, and how the solvent affects the target. This embodiment, by simultaneously acquiring three types of parameters with clear pharmacological significance, expands the dimensions of health wine analysis from single component-target correlation to a multi-dimensional quantitative evaluation framework of "in vivo exposure-target binding-solvent regulation" for the first time. This provides a complete data foundation for the subsequent construction of a weighted network that combines in vivo correlation and biological authenticity, fundamentally overcoming the one-sidedness of existing technologies that rely solely on the qualitative existence of components or single content weighting, and making the analysis results highly consistent with real biological processes.

[0021] like Figure 2 As shown, obtaining exposure-related parameters includes: Step 201. Perform qualitative and quantitative analysis on the chemical components in the health wine to obtain the content data of each component.

[0022] This embodiment utilizes ultra-high performance liquid chromatography-quadrupole-time-of-flight mass spectrometry (UPLC-Q-TOF-MS / MS) technology to separate and detect chemical components in the complex matrix of health wine, based on the principles of chromatographic separation and mass spectrometry in analytical chemistry. This technology achieves physical separation by leveraging the differences in partition coefficients of components on the chromatographic column, identifies structures through precise mass number determination and secondary fragmentation mode of the mass spectrometer, and achieves relative or absolute quantification based on the linear relationship between peak area and concentration. Combined with peak detection, alignment, and deconvolution algorithms from automated metabolomics software (such as MS-DIAL, XCMS, and Progenesis QI), and the spectral matching principle of molecular network platforms (such as GNPS) (automatically matching secondary mass spectrometry fragments with public databases), rapid component annotation from non-targeted to semi-targeted methods is achieved, significantly improving the efficiency of component analysis in complex systems. That is, by combining high-resolution liquid chromatography-mass spectrometry (LC-MS) technology with automated data processing tools, the rapid and accurate analysis of dozens to hundreds of chemical components in health wines can be achieved, reducing the time required for traditional manual analysis from several weeks to less than 2 days. At the same time, the accuracy of component identification is increased to more than 85%, providing reliable and comprehensive basic data for subsequent weight calculations and significantly improving the initial efficiency of the overall analysis process.

[0023] Step 202. Obtain oral bioavailability data for each component.

[0024] This embodiment addresses the quantification of component bioavailability based on the oral absorption principle of pharmacokinetics, employing a multi-source data acquisition strategy. Oral bioavailability (OB) refers to the extent and rate at which a drug is absorbed into the systemic circulation after oral administration, influenced by the component's physicochemical properties such as lipid solubility, water solubility, molecular weight, and the number of hydrogen bond donors and acceptors. For components with reported findings, OB values ​​are directly retrieved from professional databases such as TCMSP and DrugBank. For unknown components, based on the quantitative structure-activity relationship (QSAR) principle, absorption parameters are predicted using computer prediction models such as SwissADME and pkCSM, leveraging the component's molecular structure characteristics. Furthermore, based on the threshold screening principle of OB ≥ 30% and drug-likeness DL ≥ 0.18, candidate components with high exposure potential are identified to guide subsequent targeted validation of components entering the bloodstream. In this way, by acquiring data from multiple sources and using a priority screening strategy based on pharmacokinetic principles, we can not only solve the problem of missing OB values ​​for some components, but also reduce the workload of identifying blood-entering components by more than 70%, allowing analytical resources to focus on high-exposure potential components that are truly likely to exert in vivo activity, thus avoiding the waste of time and costs caused by blind screening.

[0025] Step 203. Integrate the content data with the oral bioavailability data to generate the in vivo exposure weight coefficients for each component as exposure-related parameters.

[0026] This step organically integrates information from two independent dimensions—in vitro content and oral bioavailability—based on the principle of multi-index fusion and normalization. First, the original content and OB values ​​are normalized to the range [0.1, 1] to eliminate dimensional differences. Second, according to the basic pharmacokinetic principle of "in vivo exposure level = in vitro content × absorption degree," the normalized content weight is multiplied by the bioavailability weight. Finally, the product is normalized again to generate the in vivo exposure weight coefficient for each component. This coefficient comprehensively reflects the combined effect of the actual amount of a component in the health tonic and its potential for absorption and utilization by the body, conforming to the basic pharmacological principle that "a component must both exist and be absorbable to exert its effect in vivo." Thus, through the dual-dimensional fusion and normalization of content and oral bioavailability, this embodiment upgrades network pharmacology from static in vitro chemical concentration analysis to dynamic in vivo accessibility assessment, effectively correcting the systematic bias in traditional methods where "high-content, low-absorption components are overestimated, and low-content, high-absorption components are underestimated." The generated in vivo exposure weight coefficients make the subsequent network analysis results closer to the actual in vivo exposure situation, significantly improving the accuracy of screening of functional components and the biological relevance of the prediction results, and providing a scientific quantitative basis for the precise efficacy analysis of health wine.

[0027] like Figure 3 As shown, obtaining the relevant parameters includes: Step 301. Predict the potential targets of each active ingredient to obtain component-target pairs.

[0028] It should be understood that this embodiment imports the chemical structures of each active ingredient into professional databases or prediction platforms such as SwissTargetPrediction, STITCH, and SEA. Utilizing pharmacophore matching algorithms or chemical structure similarity comparison principles, the target ingredient is matched with the structural features of known ligands in the database. Based on the fundamental principle of medicinal chemistry that "structurally similar substances often have similar biological activities," the potential target proteins that the ingredient may bind to are predicted. The prediction results are then processed using gene symbol normalization in the UniProt database to form a standardized set of component-target association pairs, providing foundational data for subsequent quantification of binding strength. Through target prediction based on structural similarity and pharmacophore matching principles, high-throughput identification of potential targets for dozens of active ingredients is achieved, avoiding the limitations of relying solely on existing literature reports and significantly expanding the scope of exploration into the mechanisms of action of health wines. Simultaneously, the normalization of gene symbols ensures the accuracy and cross-comparability of subsequent data analysis, laying the foundation for constructing a complete component-target interaction network.

[0029] Step 302. Perform molecular docking simulations on the component-target pair to obtain binding energy data.

[0030] like Figure 4 As shown, molecular docking simulations of component-target pairs include hierarchical docking strategies: Step 401. Perform initial screening on all component-target pairs to obtain an initial screening score.

[0031] This step employs lightweight algorithms such as reverse pharmacophore matching (e.g., PharmMapper) or similarity ensemble methods (e.g., SEA) to rapidly score the similarity between the two-dimensional or three-dimensional structural features of the components and the known ligands of the target. These methods, based on the fundamental principle of medicinal chemistry that "molecular structure determines biological activity," can perform preliminary evaluations of a large number of component-target pairs in a short time, obtaining initial screening scores as the basis for subsequent precise docking. By rapidly screening all component-target pairs, a preliminary binding tendency ranking of each pair is obtained at a relatively low computational cost, providing a scientific basis for resource allocation in subsequent graded docking and avoiding the waste of computational resources caused by directly performing high-cost molecular docking on all combinations.

[0032] Step 402. Perform molecular docking simulations only on the component-target pairs that rank in the top 10%-20% of the initial screening scores to obtain binding energy data.

[0033] In this embodiment, molecular docking is based on the "lock-key model" principle. An algorithm simulates the geometric matching and energy complementarity between the ligand (active component) and the receptor (target protein) in three-dimensional space, calculating their binding conformation and binding free energy (ΔG, unit kcal / mol, a negative value; the larger the absolute value, the stronger the binding). This process involves complex force field calculations and conformation searches, resulting in computational costs far exceeding those of initial screening methods. By focusing computational resources on the top 10%-20% of the most promising component-target pairs in the initial screening, an optimal balance between computational accuracy and resource investment is achieved. By limiting molecular docking calculations to the top 10%-20% of component-target pairs in the initial screening, the number of component-target pairs requiring high-cost molecular docking calculations is reduced by more than 80%, compressing the molecular docking calculation time for a single health wine variety from 7-14 days using traditional methods to less than 1 day. As shown in Specific Implementation Example 2, under 100-core parallel computing conditions, the number of docking tasks was reduced from 5031 pairs to 1001 pairs, and the computing time was reduced from 36 hours to 7.2 hours, while the core target identification accuracy remained above 93%, achieving a dual optimization of computing accuracy and efficiency.

[0034] Step 403. The binding force weighting coefficient of the component-target pair that has not entered the molecular docking simulation is set to 0 or a preset minimum value.

[0035] It should be understood that, based on the initial screening results, these component-target pairs have a low probability of binding and contribute very little to the network analysis. Therefore, their binding force weight coefficients are assigned to 0 or a preset minimum value (e.g., less than 0.01) to ensure network integrity while avoiding interference from these low-probability associations in subsequent analyses. This approach conforms to the Pareto principle (i.e., 80% of the results often come from the 20% of key factors), allowing network analysis to focus on truly important interactions. By zeroing out or minimizing the low-probability component-target pairs, this embodiment effectively suppresses the interference of noisy associations on the analysis results while maintaining the integrity of the network topology, enabling subsequent weighted network analysis to focus on truly biologically significant interactions. Simultaneously, this sparsity treatment also reduces the storage and computational burden of network data, further improving overall analysis efficiency.

[0036] Step 303. Generate the binding force weighting coefficients of each component-target pair based on the binding energy data as binding-related parameters.

[0037] First, an effective binding threshold (e.g., ≤-5.0 kcal / mol) is set, and only component-target pairs with binding energies exceeding this threshold are considered effective interactions. Second, the absolute values ​​of binding energies exceeding the threshold are normalized to map the binding strength to the [0,1] interval. Finally, a binding force weighting coefficient (Wb_ij) for each component-target pair is generated. This coefficient quantifies the binding affinity between the active ingredient and the target; the stronger the binding, the closer the weight is to 1. Thus, by converting molecular docking binding energy into normalized weighting coefficients, this key pharmacological parameter of component-target binding strength is quantified and integrated into the network analysis framework for the first time, overcoming the shortcomings of existing techniques that treat weak and strong bindings equally. Furthermore, the generated binding force weighting coefficients give the network edge weights clear physicochemical meaning, significantly improving the accuracy of the network model in depicting real biological interactions and providing a reliable affinity quantification basis for subsequent weighted network analysis.

[0038] like Figure 5 As shown, the parameters related to the solvent effect are obtained as follows: Step 501. Construct a target library for bidirectional regulation of ethanol. The target library includes positive synergistic targets, negative interference targets, and concentration-dependent bidirectional targets, and assigns a corresponding directional symbol to each target.

[0039] This embodiment is based on the principle of data mining from toxicology and pharmacology literature, and systematically constructs a target knowledge base covering multiple regulatory modes of ethanol. Specifically, firstly, ethanol-related gene interaction records were exported in batches from professional databases such as the Comparative Toxicogenomics Database (CTD) to obtain an initial list of targets and their types of action (upregulation / downregulation / activation / inhibition). Secondly, text mining tools such as PubTator and LitSuggest were used to automatically extract ethanol-target-disease triplets from PubMed literature to supplement new targets not included in the database. Then, the targets were scored and ranked according to the strength of literature support (citation count, publication year) and the clarity of the mechanism. Finally, the top 50-100 targets were manually reviewed for literature review, and based on the regulatory effect characteristics of ethanol on the target, they were classified into three target libraries and assigned corresponding directional symbols: positive synergistic targets (characterizing ethanol-activated protective targets), negative interference targets (characterizing ethanol-activated damaging targets), and concentration-dependent bidirectional targets (defined with directional symbols D_low for low concentration region and D_high for high concentration region, representing the opposite effects of ethanol at different concentrations), as shown in Table 1.

[0040]

[0041] Thus, by combining database export, text mining, and manual review, the initial database construction workload can be reduced by more than 80%. At the same time, a complete classification system of ethanol regulatory targets, including positive synergistic, negative interference, and concentration-dependent bidirectional targets, has been constructed. This system liberates ethanol from the technical bias of a single negative factor for the first time, scientifically acknowledges and quantifies the positive synergistic effect of ethanol at low concentrations, and provides a taxonomic basis for the subsequent dynamic quantification of the concentration-dependent regulatory role of ethanol.

[0042] Step 502. Based on the ethanol concentration data of health wine, establish an ethanol concentration-effect continuous function, the function satisfies f(0)=0, the range is [0,1] and it is monotonically increasing.

[0043] Specifically, the ethanol concentration-effect continuity function is represented by the Hill equation: f(A) = A n / (EC50 n + A n ), where A is the normalized value of the ethanol concentration of the health wine, EC50 is the normalized value of the half-number effect concentration, and n is the Hill coefficient.

[0044] It should be understood that the Hill equation originates from receptor occupation theory, and its basic form is f(A) = A n / (EC50 n +A n The equation is defined as follows: A is the normalized value of the ethanol concentration in the health wine (obtained by dividing the measured ethanol concentration by the highest reference concentration of 60% vol, with a value range of [0,1]), EC50 is the normalized value of the half-maximum effect concentration (characterizing the concentration at which ethanol produces a 50% maximum regulatory effect on the target), and n is the Hill coefficient (reflecting synergy; n>1 indicates positive synergy, n=1 indicates non-synergy, and n<1 indicates negative synergy). This equation naturally satisfies the biological constraints of f(0)=0 (zero effect without ethanol), value range [0,1] (saturable effect), and monotonically increasing, which highly matches the real S-shaped dose-response curve. The parameters EC50 and n have clear pharmacological significance and can be directly compared with the experimental data reported in the literature. That is, by introducing the Hill equation, which is recognized in the field of pharmacology, into the network pharmacology analysis framework, a continuous, smooth, and saturable mathematical expression of the ethanol concentration-response relationship can be achieved, completely overcoming the systematic modeling errors caused by the use of binary threshold models or linear assumptions in existing technologies. Furthermore, the interpretability of parameters EC50 and n allows model parameters to be directly obtained from literature data, significantly reducing the difficulty of model calibration and providing a unified quantitative standard for predicting the efficacy of different health wine varieties and different ethanol concentrations.

[0045] Step 503. Generate ethanol regulation factors for each target point based on the ethanol concentration-effect continuity function and direction sign, as parameters related to solvent effect.

[0046] This embodiment, based on the principle of regulatory factor generation, fuses the directional symbols (qualitative classification) in the target library with the concentration-effect continuous function (quantitative model) to generate the ethanol regulatory factor E for each target. Specifically, different generation rules are adopted according to the type of the target library: for a positive synergistic target library, E = 1 + f(A), representing the positive protective effect of ethanol on the target at that concentration; for a negative interference target library, E = 1 - f(A), representing the negative damaging effect of ethanol on the target at that concentration (achieved through the directional symbol); for a concentration-dependent bidirectional target library, E adopts a piecewise function form: when A ≤ Ac (concentration inflection point), E = 1 + D_low × f(A); when A > Ac, E = 1 + D_high × f(A), achieving a smooth switch between different regulatory directions in the low-concentration and high-concentration regions. If the target is not in any library, E = 1, indicating no ethanol regulatory effect. Therefore, by organically integrating the directional sign with the Hill equation, the bidirectional regulatory effect and concentration-dependent effect of ethanol are quantified into a specific ethanol regulatory factor E. This factor can be directly used as a multiplier for network edge weights in subsequent weighted network construction. For concentration-dependent bidirectional targets such as NLRP3, this embodiment can accurately distinguish between the protective effect at low concentrations (E=1+1×f(A) amplifies edge weights) and the damaging effect at high concentrations (E=1-1×f(A) inhibits edge weights). As shown in specific embodiment 3, in low-alcohol health wine (20% vol), the NLRP3 regulatory factor E=1.42 significantly enriches the anti-inflammatory pathway, while in high-alcohol health wine (45% vol), E=0.11 inhibits the pro-inflammatory pathway. This accurately characterizes the true biological characteristics of beneficial effects in appropriate amounts and harmful effects in excessive amounts, solving the core technical problem that existing technologies cannot distinguish the functional differences of health wines at different drinking doses.

[0047] Step 104. Construct a weighted network of efficacy components and targets based on multi-dimensional parameters.

[0048] like Figure 6 As shown, the construction of the efficacy component-target weighted network includes: Step 601. Multiply the exposure-related parameters, binding-related parameters, and solvent effect-related parameters to generate the edge weights of the active ingredient-target pairs.

[0049] According to the basic logic of biological action, the final regulatory effect of an active ingredient on a specific target must simultaneously satisfy three independent conditions: the ingredient must be able to reach an effective exposure level in vivo (exposure-related parameter), the ingredient must have sufficient binding affinity to the target (binding-related parameter), and the ethanol-regulated effect on the target must be conducive to the effect (solvent-related parameter). These three conditions are independent and indispensable. According to the joint probability principle in probability theory, when multiple independent events occur simultaneously, their joint effect should be the product of the probabilities of each event. Therefore, multiplying the exposure-related parameter (characterizing whether it can be reached), the binding-related parameter (characterizing whether it can bind), and the solvent-related parameter (characterizing how the solvent affects the target) yields the comprehensive regulatory weight of the ingredient-target pair at a specific ethanol concentration, as shown in Equation 1: W ij = Wc _i × Wb_ij × Ej (1) in, Wc _ i Components i In vivo exposure weight, Wb _ ij Components i With target j The binding force weight, Ej Target j Ethanol regulatory factors.

[0050] Thus, by integrating the information from the three independent dimensions of "in vivo exposure, target binding, and solvent regulation" into unified network edge weights through product fusion, each edge carries complete biological significance. This fusion method strictly adheres to the multi-conditional dependence of biological effects, avoiding the biases that may arise from additive fusion (i.e., overestimation even when one dimension is extremely strong while others are extremely weak). As shown in Specific Example 1, ginsenoside Rg1, due to its high exposure weight, strong binding weight, and the combined advantages of being positively regulated by ethanol, ranks highly in the weighted network and is accurately identified as the core active ingredient, demonstrating the scientific validity and effectiveness of product fusion.

[0051] Step 602. Construct a bidirectional weighted network with functional components and targets as nodes and edge weights as edge weights.

[0052] Specifically, the various functional components in the health wine are treated as one type of node (source nodes), and the various potential targets are treated as another type of node (target nodes). The edge weights generated in step 601 are used as the weights of the edges connecting the two types of nodes to construct a bipartite weighted network. This network has a clear biological mapping relationship: nodes represent biological entities (chemical components and target proteins), edges represent the interaction relationships between entities, and the edge weights quantify the strength of this interaction. The network adopts a bidirectional structure, which not only retains the positive regulatory information from components to targets, but also facilitates subsequent reverse tracing of which components regulate a target. The construction process can be automated using Cytoscape's CyREST API or Python's NetworkX library, enabling batch import of node and edge files and rapid generation of network topology. By constructing a bidirectional weighted network of functional components and targets, abstract multidimensional parameters are transformed into a visualized network topology structure, providing a directly computable data foundation for subsequent topology analysis and functional prediction. This network fully preserves the quantitative information of the original data. Each node and edge has a clear biological meaning and numerical weight, enabling subsequent operations such as weighted degree centrality calculation, weighted betweenness centrality analysis, and weighted pathway enrichment to be performed while maintaining the weight information. As shown in Specific Implementation Example 4, using Python + NetworkX automated scripts, all network calculations and core node sorting can be completed within 10 minutes, reducing manual operation time from 3-5 days in traditional methods to less than 10 minutes, significantly improving analysis efficiency. The bidirectional structure of the network also facilitates multi-level association analysis of "component-target-pathway-disease," laying a topological foundation for revealing the multi-target and multi-pathway mechanisms of action of health wines.

[0053] As can be seen, through the product fusion and network construction in steps 601-602, this embodiment completes the full mapping from multi-dimensional parameters to a weighted network, achieving the organic integration of three types of independent information. The constructed bidirectional weighted network has three advantages: First, the weights are derived from experimental data (content, OB) and computational simulations (molecular docking, Hill equation), possessing a solid scientific foundation; second, the network structure retains complete topological information, supporting various weighted network analysis algorithms; and third, the construction process is highly automated, enabling rapid conversion from data to network. This network, as the core carrier for subsequent analysis, allows the identification of the functional components of health wine and the prediction of new functions to be carried out based on a comprehensive consideration of in vivo exposure, binding strength, and ethanol regulation, significantly improving the accuracy and biological relevance of the analytical results.

[0054] Step 106. By analyzing the weighted network, predict the potential functions of the health wine or analyze its effective components.

[0055] like Figure 7As shown, the analysis and prediction of weighted networks include: Step 701. Perform topology analysis on the weighted network, calculate the weighted degree centrality of the nodes, and sort and select core target points according to the weighted degree centrality.

[0056] This embodiment identifies key targets by calculating the weighted degree centrality of nodes in a weighted network. The degree of a node refers to the number of edges connected to it, reflecting the breadth of its connections within the network. In a weighted network, the weighted degree is the sum of the weights of all edges connected to that node, calculated as WD(i) = ΣWij, where Wij is the edge weight between node i and its neighbor j. Weighted degree centrality comprehensively reflects both the strength and number of connections a node has in the network: a high weighted degree for a target node indicates that it is regulated by multiple high-weight functional components, strongly regulated by a few extremely high-weight functional components, or positively amplified by the ethanol regulatory factor. Based on the principle of "the rich get richer" in network science, nodes with high weighted degrees are often key hubs in the network, playing a central role in information transmission and functional implementation. By calculating and ranking the weighted degrees of all target nodes, the top-ranked targets (e.g., the top 10%-20%) are selected as core targets for subsequent functional enrichment analysis.

[0057] By employing weighted degree centrality analysis, the differences in in vivo exposure, binding strength, and ethanol regulation of components are comprehensively reflected as network importance indicators for targets, achieving a leap from qualitative correlation to quantitative importance. Compared with the degree centrality of traditional unweighted networks, weighted degree centrality effectively avoids the bias of overestimating low-weight correlations and underestimating high-weight correlations, ensuring that the screened core targets truly represent key nodes in the mechanism of action of health wine. As shown in Specific Example 1, through weighted degree ranking, ginsenoside Rg1, ginsenoside Rb1, and gecko hypoxanthine were identified as core active ingredients, which highly agrees with subsequent experimental verification results, proving the scientific validity and accuracy of this screening method.

[0058] Step 702. Perform pathway enrichment analysis on the selected core targets using weighted degrees as input weights to obtain weighted enrichment results.

[0059] Conventional pathway enrichment analyses (such as GO enrichment analysis and KEGG pathway enrichment analysis) are typically based on hypergeometric distributions or Fisher's exact test. Their input is merely a list of target names, assuming all targets have equal importance and ignoring weight differences between targets. This embodiment overcomes this limitation by using the weighted degree calculated in step 701 as the input weight for each target and employing a weighted enrichment algorithm (such as a variant of the weighted gene set enrichment analysis GSEA) for pathway enrichment calculation. Its core principle is that when assessing whether a pathway is significantly enriched, it considers not only the number of targets included in the pathway but also the proportion of the sum of the weighted degrees of these targets to the total weighted degrees of all core targets. Targets with higher weighted degrees receive a higher contribution weight in the enrichment statistics, making the enrichment results more focused on the biological pathways that health wine truly and strongly influences. Analysis can be performed using the custom weighting function of tools such as DAVID, Metascape, or ClusterProfiler.

[0060] Thus, by using weighted scores as input for pathway enrichment analysis, the enrichment results are highly correlated with the actual in vivo efficacy of the health tonic. Conventional enrichment analysis may dilute truly important pathway signals due to the inclusion of a large number of low-weight targets, while weighted enrichment analysis results in significantly higher enrichment scores for pathways containing high-weight targets, thereby more accurately revealing the core mechanism of action of the health tonic. As shown in Specific Example 1, weighted enrichment analysis successfully identified the osteoclast differentiation pathway and the FoxO signaling pathway as significantly enriched pathways. Based on this, it was predicted that ginseng and gecko health tonic has the potential to improve postmenopausal osteoporosis and has antioxidant and anti-aging effects. Animal experiments have verified the scientific validity and predictive accuracy of the weighted enrichment analysis.

[0061] Step 703. Map the weighted enrichment results to the disease database to predict the potential new functions of health wine.

[0062] This embodiment, based on the pathway-disease association mapping principle, matches the significantly enriched pathways obtained in step 702 with disease-gene association information in a human disease database to predict potential new functions of health wine. The theoretical basis is that abnormal regulation of biological pathways is the core mechanism of disease development; if a health wine can significantly affect a certain pathway, it may have an interventional effect on diseases related to that pathway. Specifically, the significantly enriched pathways are correlated with disease databases such as DisGeNET and CTD (Comparative Toxicogenomics Database) to obtain a list of diseases related to these pathways. Furthermore, the associated diseases can be sorted according to the significance level and weighting of pathway enrichment, and the diseases with the strongest associations can be selected as potential new functions of the health wine. Simultaneously, combined with the exposure-related parameters in step 601, content-weighted backtracking analysis can be used to locate the key functional components that contribute the most to the new function, providing precise guidance for new drug development or product upgrades.

[0063] It is evident that pathway-disease association mapping can directly transform the analysis results of network pharmacology into verifiable predictions of new functions of health wines, providing a scientific basis for the secondary development and expansion of new indications for health wines. Unlike existing technologies that only identify pathways, this embodiment further associates pathway information with diseases, giving the prediction results clear clinical application value. As shown in Specific Embodiment 1, based on the significant enrichment of the osteoclast differentiation pathway, it is predicted that ginseng and gecko health wine may improve osteoporosis, and this was verified in ovariectomized osteoporosis model mice. The medium-dose group significantly improved bone mineral density (P<0.05), forming a complete prediction-verification closed loop, fully demonstrating the predictive ability and practical value of this embodiment. At the same time, through content weighted backtracking analysis, the key efficacy components that contribute the most to the new functions can be accurately located, providing clear targets for the optimization of health wine formulations and quality control.

[0064] Through weighted topology analysis, weighted pathway enrichment, and disease association mapping in steps 701-703, this embodiment completes the full transformation from weighted networks to functional prediction, achieving the core objectives of analyzing the efficacy components of health wine and predicting new functions. This analytical process has three unique advantages: First, the concept of weight is used throughout the analysis, with weighting degree as the core parameter from node importance assessment to pathway enrichment calculation, ensuring a high degree of consistency between the analysis results and the actual in vivo effects of health wine; second, the analysis steps form a complete logical chain, progressing step-by-step from which targets are important (weighting degree ranking) to which pathways these targets affect (weighted enrichment) and then to which diseases these pathways are associated with (disease mapping), demonstrating rigorous logic; third, the prediction results are verifiable and traceable. As shown in specific embodiment 1, the prediction results for osteoporosis were verified through animal experiments and can be traced back to key components such as ginsenoside Rg1, providing a clear direction for subsequent product development. This entire analytical method upgrades traditional network pharmacology from a descriptive tool to a predictive tool, significantly improving the scientific rigor and accuracy of health wine research and development.

[0065] Specific Example 1: Analysis of the Efficacy Components and Prediction of New Functions of Ginseng and Gecko Health Wine 1.1 Chemical composition analysis and weighting Samples of ginseng and gecko health wine were collected, and total ion chromatograms were acquired using UPLC-Q-TOF-MS / MS technology. Peak detection, alignment, and deconvolution were performed using MS-DIAL automated metabolomics software, and secondary mass spectrometry fragment matching was performed using the GNPS molecular network platform, identifying a total of 157 chemical components. An oral bioavailability (OB) ≥ 30% and a drug-likeness DL ≥ 0.18 were set as high exposure potential screening thresholds, from which 21 candidate components were selected. Targeted tracking analysis of drug-containing serum from rats after gavage administration confirmed that 13 of these components could enter the bloodstream, including 8 progenitor components and 5 metabolites, and these were included in the candidate library of active ingredients.

[0066] 1.2 Binding force weight and generation of ethanol regulating factor Thirteen active ingredients were imported into databases such as SwissTargetPrediction for target prediction, resulting in 387 potential targets in the initial screening. A tiered docking strategy was employed, performing molecular docking simulations on only the top 20% of targets (77 in total) and the 13 ingredients, forming 1001 component-target pairs. Using a binding energy ≤ -5.0 kcal / mol as the effective binding threshold, 312 effective binding relationships were obtained, and binding force weighting coefficients were generated based on the absolute value of the binding energy. Simultaneously, ethanol-gene interaction records were exported in batches from the CTD database, and ethanol-target associations from PubMed literature were automatically extracted using text mining tools. After manual verification, an ethanol bidirectional regulatory target library was constructed, containing 11 positive synergistic targets, 23 negative interference targets, and 4 concentration-dependent bidirectional targets. The ethanol concentration of the health wine was 45% vol, with a normalized value A = 0.75. The Hill equation f(A) = A was used. n / (EC50 n +A n (n is taken as 1.5, EC50 is determined according to the literature) Calculate the ethanol concentration-effect function value, and generate the ethanol regulation factor by combining the directional signs of each target point.

[0067] 1.3 Weighted Network Construction and Core Target Identification The above three types of parameters are arranged according to Wij = Wc The product fusion of _i × Wb_ij × Ej generates the edge weights of each component-target pair, and a bidirectional weighted network is constructed with the functional components and targets as nodes. Through weighted degree centrality calculation and ranking, ginsenoside Rg1, ginsenoside Rb1, and gecko hypoxanthine are identified as the core functional components, and their weights are significantly higher than those of other components.

[0068] 1.4 Functional Prediction and Experimental Verification KEGG pathway weighted enrichment analysis was performed on the top 20% of core targets by weighted scores, using weighted scores as input weights. The results showed significant enrichment of the osteoclast differentiation pathway and the FoxO signaling pathway, in addition to the known lung adenocarcinoma pathway. The enrichment results were mapped to the DisGeNET disease database to predict that ginseng and gecko health wine has the potential to improve postmenopausal osteoporosis and provide antioxidant and anti-aging benefits. Validation was performed in ovariectomized osteoporosis mice, where the medium-dose group of ginseng and gecko health wine significantly improved bone mineral density (P<0.05), highly consistent with the network prediction results. This embodiment took 9 working days from sample processing to functional prediction, validating the feasibility and predictive accuracy of the method.

[0069] Specific Implementation Example 2: Efficiency Verification of the Hierarchical Connection Strategy 2.1 Experimental Design To verify the cost-reduction and efficiency-enhancing effects of the tiered docking strategy, two comparative experiments were conducted using the 13 active ingredients and 387 potential targets from Specific Example 1 as the subjects. The traditional docking group performed molecular docking calculations on all 13 × 387 = 5031 component-target pairs. The tiered docking group first performed an initial screening of all component-target pairs based on reverse pharmacophore matching. After obtaining the initial screening scores, molecular docking calculations were performed only on the top 20% of the 77 targets and 13 components, forming 1001 pairs. The binding force weight of component-target pairs not included in the docking was assigned a value of 0. Both experiments were conducted in a 100-core parallel computing environment, with the core target identification overlap rate and the core active ingredient ranking overlap rate as evaluation indicators.

[0070] 2.2 Results Analysis Experimental results show that the traditional docking group requires 5031 molecular docking pairs, with a computation time of 36 hours; the hierarchical docking group in this embodiment only requires 1001 docking pairs, with a computation time of 7.2 hours, reducing the number of docking tasks by 80.1% and the computation time by 80.0%. Regarding core target identification, the core targets identified by the hierarchical docking group overlapped with those identified by the traditional docking group by 93.5%; regarding the ranking of core functional components, the top 5 core functional components of the two groups overlapped by 91.7%.

[0071] 2.3 Conclusion The results show that the hierarchical docking strategy achieves 80% saving of computational resources with minimal information loss (less than 7% loss in core target identification and less than 9% loss in core component ranking), reducing the molecular docking computation time for a single health wine variety from 36 hours in the traditional method to 7.2 hours. This significantly lowers the implementation threshold and computational cost of high-throughput network pharmacology methods, verifying the effectiveness of this strategy in reducing costs and increasing efficiency while ensuring analytical accuracy.

[0072] Specific Example 3: Verification of the Segmented Weighted Effect of the Ethanol Concentration-Dependent Bidirectional Target NLRP3 3.1 Experimental Design To verify the accuracy of the concentration-dependent bidirectional target modeling method in this embodiment, NLRP3 (NOD-like receptor heat protein domain-associated protein 3) was selected as the verification object. According to literature reports, NLRP3 exhibits an inhibitory state (anti-inflammatory effect) at low ethanol concentrations (≤0.2% v / v, approximately 30 mM) and an activated state (pro-inflammatory effect) at high ethanol concentrations (≥0.5% v / v, approximately 80 mM), with a critical concentration of approximately 50 mM. Two groups of health wine samples were set up: low-alcohol health wine (ethanol concentration 20% vol, normalized value A = 0.33) and high-alcohol health wine (ethanol concentration 45% vol, normalized value A = 0.75). The Hill equation f(A) = A was used. n / (EC50 n +A n Concentration-effect modeling was performed, with EC50 set to 0.4 (corresponding to a blood alcohol concentration of 48 mM, consistent with the reported critical value of 50 mM) and Hill coefficient n set to 1.5. NLRP3 belongs to a concentration-dependent bidirectional target library, with the direction sign D_low=+1 (inhibition of NLRP3 is a protective effect) set for the low concentration region and D_high=-1 (activation of NLRP3 is a damaging effect) set for the high concentration region, and the concentration inflection point Ac=0.4.

[0073] 3.2 Results Analysis Low-alcohol health wine (20% vol): A = 0.33, substituting into Hill's equation: f(A) = 0.33¹· 5 / (0.4¹· 5 +0.33¹· 5 =0.42. According to the segmentation rule A≤Ac, we use: E=1+D_low×f(A)=1+1×0.42=1.42. The ethanol regulatory factor E=1.42 amplifies the weight of the NLRP3 related edges by 42%, and subsequent pathway enrichment analysis shows significant enrichment of the anti-inflammatory pathway. High-proof health wine (45% vol): A=0.75, substituting into the Hill equation, we get f(A)=0.75¹· 5 / (0.4¹· 5 +0.75¹· 5 =0.89. According to the segmentation rule A>Ac, E=1+D_high×f(A)=1-1×0.89=0.11. The ethanol regulatory factor E=0.11 inhibited the NLRP3 related edge weights by 89%, and subsequent pathway enrichment analysis showed that the pro-inflammatory pathways were not significantly enriched.

[0074] 3.3 Model Comparison Four sets of comparative models were set up: a traditional unweighted network model (ignoring the ethanol effect), a single negative library model (with ethanol pre-defined as a negative factor, E=1-f(A)), a single positive library model (with ethanol pre-defined as a positive factor, E=1+f(A)), and a linear model (E=1±kA). The results show that only the Hill equation + piecewise weighted model used in this invention can correctly reflect the true concentration-dependent bidirectional effect of NLRP3: enrichment of anti-inflammatory pathways at low concentrations and no significant enrichment of pro-inflammatory pathways at high concentrations. Other models either failed to distinguish concentration differences or gave opposite predictions.

[0075] 3.4 Conclusion The results show that the method in this embodiment, by constructing a concentration-dependent bidirectional target library and introducing continuous modeling using the Hill equation, has for the first time achieved accurate quantification of the concentration-dependent bidirectional effect of ethanol. This enables the network model to accurately distinguish the real biological characteristics of low-concentration protection and high-concentration damage, solving the core technical problem that existing technologies cannot distinguish the functional differences of health wines under different drinking doses. This provides a reliable mathematical basis for the study of the precise dose-effect relationship of health wines.

[0076] Specific Implementation Example 4: Efficiency Verification of Automated Network Analysis Scripts 4.1 Experimental Design To verify the cost reduction and efficiency improvement effect of the automated network analysis script in this embodiment, a comparative experiment was conducted using the weighted network of ginseng and gecko health wine constructed in Specific Embodiment 1 as the object, with two groups of experiments: manual operation and automated script. The manual operation group used Cytoscape software to manually import node and edge files, sequentially completing the weighted degree centrality calculation, weighted betweenness centrality calculation, and core node sorting and output. The automated script group used Python to write an automated script based on the NetworkX library, achieving one-click execution of the entire process from data loading to result output.

[0077]

[0078] As shown in Table 2, the automated network analysis script based on Python + NetworkX reduces manual operation time from 3-5 days (based on an 8-hour workday) to less than 10 minutes, improving efficiency by over 96%. The script automates the entire process, including batch import of node files, one-click calculation of multicentricity indicators, and automatic sorting of core nodes, completely resolving the workload bottleneck in network analysis within high-throughput network pharmacology methods. This script can run on a regular laptop without requiring high-performance computing resources, further lowering the implementation threshold of this embodiment and providing technical support for industrialization.

[0079] Specific Implementation Example 5: Verification of Method Universality—Analysis of Commercially Available Ganoderma and Goji Berry Health Wine 5.1 Experimental Objective To verify the applicability and robustness of the method in this embodiment to different types of health wines, a commercially available brand of Ganoderma lucidum and wolfberry health wine was selected as the verification object. The efficacy components were analyzed and functions were predicted according to the complete process of this invention, and the results were compared with literature reports.

[0080] 5.2 Analysis Process Component analysis: Samples of Ganoderma lucidum and wolfberry health wine were analyzed using UPLC-Q-TOF-MS / MS technology. After processing with MS-DIAL automated software and matching with the GNPS molecular network platform, a total of 142 chemical components were identified. A screening threshold of OB≥30% and DL≥0.18 was set, resulting in 26 candidate components with high exposure potential. Targeted tracking of rat serum containing the drug confirmed that 15 of these components could enter the bloodstream (11 original components and 4 metabolites), and they were included in the candidate library of active ingredients.

[0081] Target prediction and docking: Fifteen active ingredients were imported into SwissTargetPrediction for target prediction, resulting in 412 potential targets in the initial screening. A tiered docking strategy was employed, and molecular docking simulations were performed on the top 20% of the initial screening targets (82 targets) and the 15 ingredients, forming 1230 component-target pairs. Using a binding energy threshold of ≤-5.0 kcal / mol, 389 effective binding relationships were obtained, generating binding force weighting coefficients.

[0082] Ethanol regulatory factors: Based on the bidirectional regulatory target library of ethanol (11 positive synergistic, 23 negative interference, and 4 bidirectional) constructed in Specific Example 1, the ethanol concentration of the health wine was measured to be 38% vol, with a normalized value A = 0.63. The ethanol regulatory factors of each target were calculated using the Hill equation (EC50 was determined based on the literature of each target, and n was taken as 1.5).

[0083] Network Construction and Analysis: A bidirectional weighted network was constructed by productizing the three types of parameters, and core targets were identified by weighted degree centrality ranking. Weighted pathway enrichment analysis showed significant enrichment of the Toll-like receptor signaling pathway, the NOD-like receptor signaling pathway, and the PI3K-Akt signaling pathway.

[0084] Functional prediction: The enrichment results were mapped to the DisGeNET disease database to predict the potential immunomodulatory and hepatoprotective functions of Ganoderma lucidum and wolfberry health wine.

[0085] 5.3 Result Validation The prediction results were compared and verified with existing literature reports, as shown in Table 3:

[0086] Literature search results show that the main active components of Ganoderma lucidum, Ganoderma lucidum polysaccharides and Ganoderma lucidum triterpenes, have clear immunomodulatory and hepatoprotective effects, while the main components of Lycium barbarum, Lycium barbarum polysaccharides and betaine, also have immune-enhancing and hepatoprotective functions. The prediction results of this embodiment are in high agreement with known literature reports, verifying the accuracy and reliability of the method.

[0087] 5.4 Efficiency Statistics In this embodiment, the total time from sample processing to functional prediction was 7 working days, which is roughly the same as the 9 working days in Specific Embodiment 1. The reason for the slightly shorter time is that the bidirectional regulatory target library of ethanol has already been constructed and there is no need to build the library again.

[0088] 5.5 Conclusion As shown in Table 3, the method of this embodiment demonstrates good applicability and robustness to different varieties of health tonics (ginseng and gecko health tonic, and Ganoderma lucidum and wolfberry health tonic), with an analysis cycle consistently within 7-9 working days. The functional prediction results are highly consistent with literature reports. This verifies that the method of this embodiment is not dependent on specific health tonic varieties and has the value of being promoted and applied to various health tonic products, providing a general technical platform for the standardization research and precision development of the health tonic industry.

[0089] Example 2: Combination Figure 8 This embodiment provides a network pharmacological analysis system 800 for health wine, including a parameter acquisition module 801 for acquiring multi-dimensional parameters of the active ingredients in the health wine. The multi-dimensional parameters include exposure-related parameters characterizing the exposure level of the active ingredients in vivo, binding-related parameters characterizing the binding strength between the active ingredients and the target, and solvent effect-related parameters characterizing the regulatory effect of the solvent ethanol on the target. A network construction module 802 is used to construct an active ingredient-target weighted network based on the multi-dimensional parameters. An analysis and prediction module 803 is used to predict the potential functions of the health wine or analyze its active ingredients through analysis of the weighted network.

[0090] The health wine network pharmacology analysis system 800 also includes a health wine function prediction model 900. The model is stored in a computer-readable storage medium in the form of a data structure. The model 900 includes an efficacy component node feature library 901, which stores the exposure-related parameters of each efficacy component; a target node feature library 902, which stores the solvent effect-related parameters of each target, which are dynamically determined based on the ethanol concentration-effect continuous function; and an edge weight matrix 903, which stores the binding-related parameters of each efficacy component-target pair. The weights of the edges connecting efficacy component nodes and target nodes in the model are generated by the exposure-related parameters of the corresponding efficacy component, the binding-related parameters of the corresponding component-target pair, and the solvent effect-related parameters of the corresponding target according to a preset fusion rule.

[0091] It should be understood that the health wine network pharmacology analysis system 800 in this embodiment achieves full-process automation of health wine efficacy analysis and standardized data structure storage through the collaborative design of functional modules and core models. The parameter acquisition module 801, network construction module 802, and analysis and prediction module 803 in system 800 work together to form a complete technical chain from raw data input to functional prediction output, upgrading health wine analysis from traditional manual and fragmented operations to a standardized and process-oriented automatic processing mode, significantly improving analysis efficiency and operational convenience.

[0092] Crucially, the health wine function prediction model 900 integrated within System 800 is stored in a computer-readable storage medium in a standardized data structure, serving as the core data layer of the entire system and being accessed by various functional modules. This model 900, through the organic construction of an efficacy component node feature library 901, a target node feature library 902, and an edge weight matrix 903, provides structured storage and management of three core data categories: exposure-related parameters, solvent effect-related parameters, and binding-related parameters. The multi-dimensional parameters generated by the parameter acquisition module 801 are stored in real-time in the corresponding feature library of Model 900; the network construction module 802 directly reads the three types of parameters from Model 900, generates edge weights according to preset fusion rules, and constructs a weighted network; the analysis and prediction module 803 performs topology analysis and function prediction based on the constructed network. This closed-loop data flow design of "module-model-module" enables efficient sharing and orderly access of the three types of parameters within the system, avoiding redundant data processing and transmission.

[0093] It is worth noting that the data structure design of Model 900 enables dynamic updates—when the ethanol concentration data of the health wine changes, the solvent effect-related parameters in the target node feature library 902 can be adjusted in real time based on the ethanol concentration-effect continuous function, without the need to rebuild the entire model. The network construction module 802 directly reads the latest parameters from the updated Model 900 during each analysis, enabling the system to quickly generate suitable weighted networks and prediction results for health wines with different ethanol concentrations, solving the technical challenge of traditional methods being unable to flexibly respond to changes in ethanol concentration. Simultaneously, the standardized data structure of Model 900 allows for persistent storage and cross-platform sharing, enabling one-time construction and multiple reuses, avoiding resource waste caused by repeated modeling, and providing sustainable technical support for batch analysis and long-term research of health wines.

[0094] It should be noted that the schemes or principles involved in the health wine network pharmacology analysis system 800 and the health wine function prediction model 900 in this embodiment are the same as those in Embodiment 1. The same or similar contents will not be described in detail.

[0095] Example 3: This invention also provides a terminal device, which may include a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the above-described functionality. Figure 1-7 The various processes of the network pharmacological analysis method embodiment for health wine shown are all able to achieve the same technical effect, and will not be described again here to avoid repetition.

[0096] Combination Figure 9 As shown, this embodiment also discloses a specific implementation of a computer-readable storage medium 90. This computer-readable storage medium 90 can be configured wholly or partially within a physical computer, server, cluster server, or data center.

[0097] In this embodiment, the computer-readable storage medium 90 stores computer program instructions 91, which are read and executed by a processor 92 to perform the steps in the network pharmacological analysis method for health wine as disclosed in Embodiment 1.

[0098] Optionally, the computer-readable storage medium 90 can be configured as a server, and the server operates on a physical device used to build a private cloud, hybrid cloud, or public cloud. The computer-readable storage medium 90 can also be configured as random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0099] The computer-readable storage medium 90 is used to store a program, and the processor 92, upon receiving an execution instruction, executes the network pharmacological analysis method for health wine disclosed in Embodiment 1.

[0100] Meanwhile, the processor 92 disclosed in this embodiment may be an integrated circuit chip with signal processing capabilities. The processor 92 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor.

[0101] The technical solution of the computer-readable storage medium 90 disclosed in this embodiment that is the same as that in Embodiment 1 and / or Embodiment 2 is described in Embodiment 1 and / or Embodiment 2, and will not be repeated here.

[0102] The detailed descriptions listed above are merely specific descriptions of feasible embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. All equivalent embodiments or modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention.

[0103] 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 specific 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 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.

[0104] 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 network pharmacological analysis of health wine, characterized in that, include: The multidimensional parameters of the functional components in health wine are obtained. The multidimensional parameters include exposure-related parameters characterizing the exposure level of the functional components in vivo, binding-related parameters characterizing the binding strength between the functional components and the target, and solvent effect-related parameters characterizing the regulatory effect of the solvent ethanol on the target. A weighted network of efficacy components and targets is constructed based on the aforementioned multi-dimensional parameters; By analyzing the weighted network, the potential functions of health wine can be predicted or its effective components can be identified.

2. The method according to claim 1, characterized in that, The acquisition of exposure-related parameters includes: Qualitative and quantitative analysis of the chemical components in health wine was conducted to obtain the content data of each component; Obtain oral bioavailability data for each component; The content data and the oral bioavailability data are fused to generate in vivo exposure weight coefficients for each component, which are used as the exposure-related parameters.

3. The method according to claim 1, characterized in that, The acquisition of the relevant parameters includes: Predict the potential targets of each active ingredient to obtain component-target pairs; Molecular docking simulations were performed on the component-target pairs to obtain binding energy data; Based on the binding energy data, binding force weighting coefficients for each component-target pair are generated as binding-related parameters.

4. The method according to claim 3, characterized in that, The molecular docking simulation of component-target pairs includes a hierarchical docking strategy: All component-target pairs are initially screened to obtain an initial screening score; Molecular docking simulations were performed only on the component-target pairs that ranked in the top 10%-20% of the initial screening scores to obtain binding energy data; The binding force weighting coefficient of component-target pairs that are not included in the molecular docking simulation is assigned to 0 or a preset minimum value.

5. The method according to claim 1, characterized in that, The acquisition of solvent effect-related parameters includes: A target library for bidirectional regulation of ethanol was constructed, which includes positive synergistic targets, negative interference targets, and concentration-dependent bidirectional targets, and each target was assigned a corresponding directional symbol. An ethanol concentration-effect continuous function is established based on the ethanol concentration data of health wine. The function satisfies f(0)=0, has a range of [0,1], and is monotonically increasing. Based on the ethanol concentration-effect continuity function and the direction sign, an ethanol regulation factor is generated for each target point, which serves as a solvent effect-related parameter.

6. The method according to claim 5, characterized in that, The ethanol concentration-effect continuity function is represented by the Hill equation: f(A) = A n / (EC50 n + A n ), where A is the normalized value of the ethanol concentration of the health wine, EC50 is the normalized value of the half-number effect concentration, and n is the Hill coefficient.

7. The method according to claim 1, characterized in that, The construction of the efficacy component-target weighted network includes: The exposure-related parameters, the combination-related parameters, and the solvent effect-related parameters are multiplied together to generate the edge weights of the efficacy component-target pairs; A bidirectional weighted network is constructed using the functional components and targets as nodes and the edge weights as edge weights.

8. The method according to claim 1, characterized in that, The analysis and prediction of the weighted network includes: Perform topology analysis on the weighted network, calculate the weighted degree centrality of the nodes, and sort and screen core target points according to the weighted degree centrality; The selected core targets were used as weighted inputs for pathway enrichment analysis to obtain weighted enrichment results. The weighted enrichment results are mapped to a disease database to predict potential new functions of health wine.

9. A predictive model for the function of health wine, characterized in that, The model is stored in a computer-readable storage medium in the form of a data structure, including: The functional component node feature library is used to store exposure-related parameters for each functional component; A target node feature library is used to store solvent effect-related parameters for each target, which are dynamically determined based on an ethanol concentration-effect continuous function. The edge weight matrix is ​​used to store the binding-related parameters of each functional component-target pair; The weights of the edges connecting the functional component nodes and the target nodes in the model are generated according to a preset fusion rule by the exposure-related parameters of the corresponding functional components, the binding-related parameters of the corresponding component-target pairs, and the solvent effect-related parameters of the corresponding targets.

10. A network pharmacological analysis system for health wine, characterized in that, include: The parameter acquisition module is used to acquire multi-dimensional parameters of the functional components in the health wine. The multi-dimensional parameters include exposure-related parameters characterizing the exposure level of the functional components in vivo, binding-related parameters characterizing the binding strength between the functional components and the target, and solvent effect-related parameters characterizing the regulatory effect of the solvent ethanol on the target. The network construction module is used to construct an efficacy component-target weighted network based on the multi-dimensional parameters. The analysis and prediction module is used to predict the potential functions of health wine or analyze its effective components by analyzing the weighted network.

Images