Method for screening active ingredients for improving insulin resistance of coptis chinensis flower stalks and regulating and controlling pathways

Abstract

The invention relates to the technical field of traditional Chinese medicine network pharmacology and virtual screening analysis, in particular to a method for analyzing active ingredients and specific signal pathways of coptis flower stalk extract for improving insulin resistance based on virtual screening. According to the method, the screening model suitable for the GSK3B target inhibitor is constructed, and the inventor further optimizes the model, so that the accuracy of screening the GSK3B target inhibitor by the model is greatly improved. The method provides a new thought for explaining a molecular mechanism of the coptis chinensis flower stalk for treating insulin resistance, and lays a foundation for further research and development of an innovative preparation of the coptis chinensis flower stalk extract for treating insulin resistance.

Description

Technical Field

[0001] The present invention relates to the technical field of traditional Chinese medicine network pharmacology and virtual screening analysis, and particularly to a method for analyzing active ingredients and specific regulatory pathways of a Coptis chinensis flower stalk extract for improving insulin resistance based on virtual screening. Background Art

[0002] The inflorescence of Coptis chinensis (Coptis chinensis Franch.) in the Ranunculaceae family is formed by the flowering of the plant. Flowering consumes some nutrients, and to increase yield, the inflorescence is often removed and discarded, resulting in considerable waste. Studies have shown that Coptis chinensis has lipid-lowering and blood sugar-lowering properties, as well as antioxidant effects. However, its primary active ingredients and mechanisms of action remain unclear. Therefore, increased research and analysis of the effects of Coptis chinensis on the treatment of insulin resistance could provide important guidance and evidence for the development of drugs using Coptis chinensis for the treatment of insulin resistance. Research into the mechanisms of Traditional Chinese Medicine (TCM) in treating diseases often hinders pinpointing its active ingredients and targets due to the numerous active ingredients, numerous targets, and complex mechanisms of action.

[0003] Therefore, there is an urgent need for a more convenient and accurate method to screen the active ingredients and corresponding regulatory pathways of Coptis chinensis stalks that improve insulin resistance. Summary of the Invention

[0004] The present invention aims to solve one of the technical problems in the related art at least to a certain extent.

[0005] To this end, the first aspect of the present invention provides a method for screening the regulatory pathway of the active ingredient of Coptis chinensis flower stalk for improving insulin resistance, specifically comprising:

[0006] 1) obtaining active ingredients from the flower stalks of Coptis chinensis;

[0007] 2) screening the active target of the active ingredient according to database 1;

[0008] 3) Screening disease targets for insulin resistance based on database 2;

[0009] 4) drawing a Venn diagram of the active targets and the disease targets to obtain intersection targets;

[0010] 5) Based on the intersection targets, a network diagram of active ingredients of Coptis chinensis stalks - intersection targets - insulin resistance - regulatory pathways was drawn using Cytoscape;

[0011] 6) Based on the intersection targets, the intersection targets are analyzed using the STRING database to obtain core target genes;

[0012] 7) Based on the core target genes, enrichment analysis was performed to obtain the regulatory pathway of the active ingredients of Coptis chinensis stalks that improve insulin resistance.

[0013] The present invention uses HPLC-MS / MS, network pharmacology technology and virtual screening technology to separate the effective active ingredients, potential targets and pathways of Coptis chinensis flower stalk extract in improving insulin resistance, and analyzes its "multi-component-multi-target-multi-pathway" effect and mechanism of improving insulin resistance. The active ingredients of Coptis chinensis flower stalk and its potential targets were obtained for the first time using TCMSP, SwissTargetPrediction, ChEMBL and STITCH analysis platforms; virtual target genes related to insulin resistance were obtained through DisGeNET, GeneCards, OMIM and CTD databases, and the key targets of Coptis chinensis flower stalk for treating insulin resistance were obtained through interactive processing; Cytoscape3.9.1 was used to construct a "component-target-disease" network diagram of Coptis chinensis flower stalk extract and insulin resistance to show how Coptis chinensis flower stalk extract exerts its efficacy through "multi-target, multi-pathway, multi-level" integrated regulation. DAVID (Version 2021) was used to perform GO enrichment and KEGG pathway annotation analysis on the obtained intersection target information.

[0014] According to a specific embodiment of the present invention, in step 1), the method for obtaining the active ingredient in the flower stalk of Coptis chinensis comprises:

[0015] A1. Obtain the chemical components of the extract of Coptis chinensis stalk by HPLC-MS / MS;

[0016] A2. Screen the chemical components with a peak area of ​​not less than 0.17, a bioavailability of not less than 30%, and a drug-like property of not less than 0.18 in the HPLC-MS / MS mass spectrum as the active ingredients in the Coptis chinensis flower stalk.

[0017] According to a specific embodiment of the present invention, in step 2), the database 1 includes at least one of TCMSP, SwissTargetPrediction, ChEMBL, and STITCH;

[0018] In step 3), the database 2 includes at least one of DisGeNET, GeneCards, OMIM, and CTD;

[0019] The method for screening disease targets includes: selecting a target in the database 2 having a Relevancescore value of not less than 10 in the GeneCards database as the disease target;

[0020] The active targets and the disease targets were imported into the Wayne Online platform to obtain their intersection targets, which were then imported into the STRING database, and the core target genes were selected with a medium confidence value greater than 0.9.

[0021] According to a specific embodiment of the present invention, before step 5), it further includes: using KEGG enrichment analysis to screen the top 20 pathways with P values ​​less than 0.05 to construct the intersection-target-insulin resistance-regulatory pathway network diagram of the active ingredients of Coptis chinensis stalks.

[0022] According to a specific embodiment of the present invention, the core target genes are found based on the intersection of the disease target and the compound target. The intersection targets are analyzed using the STRING database to obtain the degree values ​​of the intersection targets. Then, the protein interaction network diagram is drawn using Cytoscape software to make the core targets more intuitive.

[0023] According to a specific embodiment of the present invention, in step 7), the enrichment analysis includes GO enrichment and KEGG pathway annotation analysis.

[0024] The second aspect of the present invention provides a method for constructing a screening model for GSK3B target inhibitors, comprising:

[0025] a. Obtain a dataset containing high-throughput detection results of GSK3B target inhibitors from Database 3;

[0026] b. Preprocessing the data set, including:

[0027] Preprocessing 1: IC 50 Compounds with concentrations below 500 nM were considered inhibitors, and the IC 50 Compounds above 500 nM were non-inhibitors;

[0028] Preprocessing 2: Remove duplicates of the same compound in the dataset and compounds that are judged to be both inhibitors and non-inhibitors;

[0029] c. Randomly split the preprocessed dataset into training, validation, and test sets, and construct a screening model for GSK3B target inhibitors based on the deep learning module in the directed message passing network model;

[0030] d. training the screening model to obtain the final screening model, wherein the training includes hyperparameter optimization and threshold screening,

[0031] in,

[0032] The hyperparameter optimization includes a first optimization and a second optimization, wherein the first optimization is automatic hyperparameter optimization, and the second optimization is manual optimization based on parameter 1 obtained by the automatic hyperparameter optimization, and parameter 2 obtained by the second optimization is the optimal parameter of the screening model.

[0033] The parameter 2 is "depth":4,"dropout":0.1,"ffn_hidden_size":800,"ffn_num_layers":3,"hidden_size":800.

[0034] According to a specific embodiment of the present invention, when training a GSK3B target inhibitor screening model, the inventors performed a secondary optimization on the parameters after automatic parameter optimization. Compared with the AUC of 0.834 before optimization, the AUC of the optimized model was 0.921. The criteria for those skilled in the art to judge the quality of the prediction model based on the AUC value are as follows: when AUC = 1, it is judged to be a perfect prediction model; when AUC = [0.85-0.95], the prediction model is very effective; when AUC = [0.7-0.85], the model is judged to be average; when AUC = [0.5-0.7], the model is judged to be less effective; when AUC = 0.5, it is a meaningless model. This application greatly improves the accuracy of model prediction by optimizing the model parameters, and the optimized model has an accuracy rate of up to 100% in predicting the active ingredients of the main active ingredients of Coptis chinensis as GSK3B target inhibitors.

[0035] The threshold is an evaluation indicator that satisfies the precision and recall values ​​of the screening model to be greater than 0.8 at the same time by evaluating the precision, recall rate, harmonic mean of the precision and recall rate, and the Matthews correlation coefficient, and the evaluation indicator is not less than 0.7.

[0036] The inventors used deep learning to construct a screening model for GSK3B core target inhibitors, predicted the main components of the Coptis chinensis flower stalk extract, and further verified it through molecular docking. The results found that seven components in the Coptis chinensis flower stalk are related to 152 insulin resistance targets, including GSK3B. KEGG enrichment analysis mainly involved the PI3K / AKT signaling pathway and the ErbB signaling pathway, with GSK3B as the key target. By constructing a GSK3B inhibitor screening model and verifying it with molecular docking technology, the main active ingredients were found to be berberine, epiberberine, and coptisine. Therefore, the method of the present invention provides new ideas for elucidating the molecular mechanism of Coptis chinensis flower stalk in treating insulin resistance and lays the foundation for further development of innovative preparations for the treatment of insulin resistance with Coptis chinensis flower stalk extract.

[0037] According to a specific embodiment of the present invention, the inventors further optimized the training parameters when constructing a GSK3B target inhibitor screening model. Specifically, they adjusted "depth" to 4, "dropout" ratio to 0.1, "ffn_hidden_size" to 800, "ffn_num_layers" to 3, and "hidden_size" to 800. The model constructed by the present invention had an accuracy rate of up to 100% when screening GSK3B target inhibitors in the active ingredients of Coptis chinensis.

[0038] According to a specific embodiment of the present invention, in step a, the database 3 is CHEMBL.

[0039] According to a specific embodiment of the present invention, in step c, the training set, validation set and test set account for 70-90%, 5-15%, and 5-15% of the preprocessed data set, respectively.

[0040] In step d, the automatic hyperparameter optimization method includes at least one of a random search method, a Bayesian optimization method, and a gradient descent method;

[0041] The number of rounds of automatic hyperparameter optimization is 18-25 rounds;

[0042] The number of rounds of the second optimization is 2-4 rounds;

[0043] The model training cycle is 28-35;

[0044] When the evaluation index of the compound in the model is not less than 0.7, it is determined to be a potential inhibitor of the GSK3B target.

[0045] The third aspect of the present invention provides a GSK3B target inhibitor screening model constructed by the method described in the second aspect.

[0046] The fourth aspect of the present invention provides the use of the GSK3B target inhibitor screening model described in the third aspect in screening drugs related to insulin resistance and / or screening compounds having GSK3B target inhibitors.

[0047] The holistic systemic characteristics of "network pharmacology" align with the multi-component and multi-target nature of traditional Chinese medicine (TCM) formulas. Therefore, network pharmacology can be used to identify the chemical components and primary targets of TCMs. Deep learning is a subset of machine learning based on data representation learning; molecular docking technology uses computer simulations to study intermolecular interactions to predict binding patterns and affinities. The inventors of this application combined these three techniques to verify the primary components and targets, and identified key targets and active ingredients in Coptis chinensis (Coptis chinensis) involved in improving insulin resistance.

[0048] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments with reference to the accompanying drawings, in which:

[0050] Figure 1 This is the total ion chromatogram of the HPLC-MS / MS of the Coptis chinensis flower stalk extract in Example 1 of the present invention;

[0051] Figure 2 A in the figure is the Venn diagram of the extract of Coptis chinensis flower stalk in Example 2 of the present invention, Figure 2 B in the figure is the network diagram of active ingredients of Coptis chinensis flower stalks-intersection targets-insulin resistance-regulatory pathway in Example 2 of the present invention;

[0052] Figure 3 This is a protein interaction network diagram of the Coptis chinensis flower stalk extract in Example 2 of the present invention;

[0053] Figure 4 GO enrichment analysis of the target components in Example 2 of the present invention - biological process, cellular component, molecular function;

[0054] Figure 5 KEGG enrichment analysis of the target component in Example 2 of the present invention;

[0055] Figure 6 Figure 3 shows the ROC curves of the four hyperparameter optimizations in Example 3 of the present invention. Figure A shows the ROC curve corresponding to the final parameters of the automatic hyperparameter optimization, and Figure B shows the ROC curve corresponding to the optimal parameters obtained by three manual hyperparameter optimizations.

[0056] Figure 7 This is a diagram of the molecular docking results in Example 3 of the present invention. DETAILED DESCRIPTION

[0057] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be understood as limiting the present invention.

[0058] It should be noted that the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, features defined as "first" or "second" may explicitly or implicitly include one or more of such features. Furthermore, in the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0059] The endpoints of the ranges and any values ​​disclosed herein are not limited to the precise ranges or values, and these ranges or values ​​should be understood to include values ​​close to these ranges or values. For numerical ranges, the endpoints of each range, the endpoints of each range and individual point values, and the individual point values ​​can be combined with each other to obtain one or more new numerical ranges, which should be considered to be specifically disclosed herein.

[0060] In order to make the present invention more easily understood, certain technical and scientific terms are specifically defined below. Unless otherwise clearly defined elsewhere in this document, all other technical and scientific terms used herein have the meaning commonly understood by those skilled in the art to which the present invention belongs.

[0061] In this document, the terms “include” or “comprising” are open expressions, that is, including the contents specified in the present invention, but not excluding other contents.

[0062] As used herein, the terms "optionally," "optional," or "optionally" generally mean that the subsequently described event or circumstance may but need not occur, and that the description includes instances where the event or circumstance occurs and instances where it does not.

[0063] In this article, glycogen synthase kinase-3 beta (GSK3B) is a threonine / serine kinase that is ubiquitous in mammalian eukaryotic cells. In addition to regulating glycogen synthase, it also participates in regulating processes such as cell differentiation, proliferation, survival, and apoptosis. In the insulin signaling pathway, GSK3B is a key regulator of glycogen synthesis and a negative regulator of the insulin signaling pathway. Inactivation of GSK3B alleviates the inhibitory effect on glycogen synthase and promotes glycogen synthesis.

[0064] The present invention aims to provide a method for analyzing the active ingredients of Coptis chinensis flower stalk extract for improving insulin resistance based on virtual screening. It effectively combines complex network analysis technology and big data technology to analyze and calculate the active ingredients of Coptis chinensis flower stalk that improve insulin resistance, solving the current problems of complex chemical composition and unclear effective ingredients of Coptis chinensis flower stalk.

[0065] The technical solution of the present invention specifically includes the following steps:

[0066] S1. Preparation of Coptis chinensis flower stalk extract and analysis of main chemical components:

[0067] The flower stalks of Coptis chinensis were extracted with 50-80% ethanol and freeze-dried to obtain the flower stalk extract. The main chemical components of the flower stalk extract of Coptis chinensis were obtained by HPLC-MS / MS.

[0068] S2. Identify drug targets:

[0069] The active ingredients of Coptis chinensis flower stalk extract that met the requirements were screened in the TCMSP database, and the relevant target genes were obtained in the TCMSP, SwissTarget Prediction, ChEMBL and STITCH databases;

[0070] S3. Identify disease targets:

[0071] DisGeNET, GeneCards, OMIM, and CTD databases were used to obtain virtual target genes related to insulin resistance, which were then intersected with the target genes of the main components of Coptis chinensis obtained by S2. The resulting intersection genes were the drug-disease intersection target genes.

[0072] S4. Constructing a network of Coptis chinensis stalks and diseases, and screening core nodes:

[0073] The qualified main components and target genes of Coptis chinensis stalks obtained in S2 and the drug-disease intersection target genes obtained in S3 were entered into Cytoscape 3.9.1 software. A "component-target-disease" network diagram for Coptis chinensis stalks and insulin resistance was constructed using Cytoscape 3.9.1. The intersection genes obtained in S3 were analyzed in the STRING database, and the results were imported into Cytoscape 3.9.1 software. A protein interaction network diagram was constructed based on the degree parameter, and core nodes were selected for subsequent analysis.

[0074] S5. Enrichment analysis of core nodes:

[0075] Using the DAVID biomolecular function annotation system, GO enrichment and KEGG pathway annotation analysis were performed on core nodes to predict target functional distribution. The enrichment degree of core pathways was analyzed based on enrichment factors, and the main regulatory pathways of Coptis chinensis peduncle in improving insulin resistance were analyzed.

[0076] S6. Constructing a core target inhibitor screening model:

[0077] Using the previously screened GSK3B as a keyword, we searched for and downloaded datasets containing high-throughput assays for GSK3B inhibitors from the CHEMBL database. We then set a threshold to classify the compounds in the dataset as inhibitors or non-inhibitors. After removing duplicates from the dataset, we set model training parameters and trained the Chemprop model. We also set model evaluation metrics to identify a model that met the screening requirements.

[0078] S7. Screening of GSK3B target inhibitors from the main components of Coptis chinensis flower stalks:

[0079] The constructed target inhibitor screening model was used to predict the target inhibitory activity of the main components of Coptis chinensis stalks, set thresholds, and obtain potential active ingredients;

[0080] S8. Use molecular docking technology to verify the inhibitory effect of potential active ingredients:

[0081] The PubChem database was used to obtain SDF files for potential active ingredients. Protein information for GSK3B was retrieved from the RCSBPDB database using the keyword GSK3B and downloaded. The obtained protein receptors and compounds were processed using AutoDock Tools 1.5.7. Molecular docking was performed using Autodock Vina 1.1.2 software, with threshold settings, and visualization using Pymol 2.5.4.

[0082] In the method S2 and S3, the screening criteria are as follows: threshold: peak area ≥ 0.17, OB ≥ 30%, DL ≥ 0.18, medium confidence > 0.9, Relevance score ≥ 10, P < 0.05.

[0083] In the method described, in S2 and S3, the topological parameters of all nodes in the network graph are calculated using Cytoscape software, the degre parameter is selected as the screening condition for the core nodes, and the core nodes are selected for subsequent analysis.

[0084] In S6, the threshold value is, inhibitor: IC 50 <500nM, non-inhibitor: IC 50 >500nM, the model training parameters are 20 rounds of hyperparameter optimization and 30 training cycles, and the model evaluation indicators are precision, recall rate, F1 score and Matthews correlation coefficient.

[0085] In S7, the thresholds are: potential inhibitors: score ≥ 0.7, potential non-inhibitors: score < 0.7.

[0086] In S8, the threshold value is, for confirmed inhibitors: binding free energy < -8.0 kcal / mol.

[0087] The present invention has the following beneficial effects:

[0088] This invention combines authoritative medical databases from home and abroad to systematize, streamline, and standardize research methods for Coptis chinensis flower stalks. It also reduces the cost of analyzing and studying Coptis chinensis flower stalks and expands the application of advanced computer network technologies in the medical field. This invention aims to establish novel network pharmacology and virtual screening research methods, which are of particular significance for breakthroughs in the research of Coptis chinensis flower stalks and other traditional Chinese medicines, providing a foundation for supporting and guiding the development and industrialization of innovative drugs in my country.

[0089] The present invention uses HPLC-MS / MS, network pharmacology technology and virtual screening technology to separate the effective active ingredients, potential targets and pathways of Coptis chinensis flower stalk extract in improving insulin resistance, and analyzes its "multi-component-multi-target-multi-pathway" effect and mechanism of improving insulin resistance. The active ingredients of Coptis chinensis flower stalk and its potential targets were obtained for the first time using TCMSP, SwissTargetPrediction, ChEMBL and STITCH analysis platforms; virtual target genes related to insulin resistance were obtained through DisGeNET, GeneCards, OMIM and CTD databases, and the key targets of Coptis chinensis flower stalk for treating insulin resistance were obtained through interactive processing; Cytoscape3.9.1 was used to construct a "component-target-disease" network diagram of Coptis chinensis flower stalk extract and insulin resistance to show how Coptis chinensis flower stalk extract exerts its efficacy through "multi-target, multi-pathway, multi-level" integrated regulation. DAVID (Version 2021) was used to perform GO enrichment and KEGG pathway annotation analysis on the obtained intersection target information. Deep learning was used to construct a screening model for GSK3B core target inhibitors, predict the main components of the Coptis chinensis flower stalk extract, and further verify them through molecular docking. The results found that seven components in the Coptis chinensis flower stalk are related to 152 insulin resistance targets, including GSK3B, through KEGG enrichment analysis. The main targets were the PI3K / AKT signaling pathway and the ErbB signaling pathway, with GSK3B as the key target. By constructing a GSK3B inhibitor screening model and verifying it with molecular docking technology, it was found that the main active ingredients are berberine, epiberberine, and coptisine. Therefore, the method of the present invention provides new ideas for elucidating the molecular mechanism of Coptis chinensis flower stalk in treating insulin resistance and lays the foundation for the further development of innovative preparations for the treatment of insulin resistance with Coptis chinensis flower stalk extract.

[0090] The scheme of the present disclosure will be explained below in conjunction with the examples. Those skilled in the art will understand that the following examples are only used to illustrate the present disclosure and should not be considered to limit the scope of the present disclosure. Where specific techniques or conditions are not specified in the examples, they are carried out according to the techniques or conditions described in the literature in this area or according to the product instructions. Where the manufacturer of the reagents or instruments is not specified, they are all conventional products that can be obtained commercially.

[0091] Example 1 Analysis of the main chemical components of the extract of Coptis chinensis

[0092] 1. Preparation of Coptis chinensis flower stalk extract

[0093] Take 100g of Coptis chinensis flower stalks, grind them into powder using a grinder, dissolve them in 70% ethanol, extract them for 30 minutes, and heat them under reflux at 80℃ for 3 times. Take the supernatant, distill it under reduced pressure at 55℃, dry the concentrated solution at constant temperature, extract it, and store it at 4℃ for later use.

[0094] 2. Identification of the main chemical components of the extract of Coptis chinensis flower stalk

[0095] Separation was performed using an Agilent Extended-C18 column (4.6 mm x 250 mm, 5 μm) with a 0.1% formic acid aqueous solution (phase A)-acetonitrile (phase B) as the mobile phase for gradient elution (60% A-40% B in the first 5 minutes, 60% A-60% B from 5 to 20 minutes, 60% A-80% B from 20 to 30 minutes, and 60%-100% B from 30 to 48 minutes). Samples were collected using an ESI source in positive ion mode at a flow rate of 0.8 mL / min, a column temperature of 40°C, and an injection volume of 10 μL.

[0096] The ethanol extract of Coptis chinensis flower stalk was qualitatively analyzed by HPLC-MS / MS technology, and its total ion current is shown in the attached figure. Figure 1 The molecular formula of the compound was deduced using XCalibur 3.0 software. The MS / MS fragment peaks and retention times were compared with the literature data on Coptis chinensis flower stalks. The chemical components of Coptis chinensis flower stalks were preliminarily identified. The results are shown in Table 1.

[0097] Table 1

[0098]

[0099]

[0100]

[0101] The main chemical components of Coptis chinensis flower stalks include: norcoclaurine, quercetin, acacetin, sakura alkaloids, magnolia alkaloids, jatrophorrhizine, terrigine, berberine, epiberberine, palmatine, isoquercetin, mongoside, chlorogenic acid, methyl coptisine, larch resin alcohol, oleander flavonoids, rhamnetin, isorhamnetin, coptisine, inula lactone, tectoriusin, javanicin A, 16,17-dehydrocapsaicin, baicalein, calycosin isoflavones, and quercetin 3-O-2(g)-rhamnosyl-rutinoside.

[0102] Example 2 Main regulatory pathways of Coptis chinensis stalks in improving insulin resistance

[0103] 1. Identify drug targets:

[0104] Combining the peak areas of the compounds and with OB ≥ 30% and DL ≥ 0.18, seven main active compounds were screened out from the TCMSP (https: / / old.tcmsp-e.com / tcmsp.php) database, as shown in Table 2, namely: berberine, epiberberine, palmatine, isorhamnetin, coptisine, baicalein and calycosin.

[0105] Table 2

[0106]

[0107] The Swiss Target Prediction (http: / / swisstargetprediction.ch) database, ChEMBL (https: / / www.ebi.ac.uk / chembl) database, TCMSP (https: / / old.tcmsp-e.com / tcmsp.php) and STITCH (http: / / stitch.embl.de) database were used to screen out the relevant target genes of the above 7 compounds.

[0108] 2. Identify disease targets:

[0109] In the OMIM (online Mendelian inheritance in man) (https: / / omim.org / ) database, DisGeNET (http: / / www.disgenet.org / ) database, and GeneCards (https: / / www.genecards.org / ) database, we searched for "insulin resistance" as the keyword and screened the targets of insulin resistance. The compound targets and disease targets were imported into the Venny2.1.0 online Venn diagram website, and a total of 152 intersection targets were obtained ( Figure 2 A in the figure was imported into Cytoscape 3.9.1 software, and a network diagram of “active ingredients of Coptis chinensis flower stalks-target-disease-pathway” was obtained by drawing. The results are as follows Figure 2 In Figure B, the green diamonds represent the flower stalks of Coptis chinensis, the blue hexagons represent the seven main active ingredients, the yellow octagon represents insulin resistance, the orange triangles represent the top 20 pathways with P < 0.05 from KEGG enrichment analysis, and the purple rectangle in the middle represents the potential target genes of Coptis chinensis flower stalks for the treatment of insulin resistance.

[0110] 3. Constructing a network of Coptis chinensis stalks and diseases, and screening core nodes:

[0111] The intersection targets were input into the STRING database for analysis to obtain the degree value ranking of the active ingredients, which were then imported into Cytoscape 3.9.1 software for visualization and drawing to obtain the protein interaction network diagram (see Appendix). Figure 3 . It consists of 152 "nodes" and 560 "edges": the higher the correlation, the redder its color, the larger the circle, the more connected edges, and the larger the font. "Edges" represent the interactions between the three. The results showed that the degree value of glycogen synthase kinase-3&beta (GSK3B) was at the top, indicating that it may be a key target for Coptis chinensis to improve insulin resistance. Each active ingredient corresponds to multiple targets, and each target is connected to multiple ingredients, reflecting the therapeutic mechanism of Coptis chinensis with multiple components and multiple targets. At the same time, the pathways are connected through common targets and are not independent, further indicating that the pathways interact synergistically.

[0112] 4. Enrichment analysis of core nodes

[0113] Using the DAVID (https: / / david.ncifcrf.gov / tools.jsp) biomolecular functional annotation system, we obtained information on targets associated with improving insulin resistance in Coptis chinensis stems and conducted GO enrichment and KEGG pathway annotation analysis. The enrichment analysis module used a topological analysis algorithm to perform GO enrichment and KEGG enrichment analysis on each module based on the module gene relationship table generated by the interval partitioning module. The resulting GO and KEGG analysis results included a list of modules and a list of genes within the modules. The main regulatory pathways associated with improving insulin resistance in Coptis chinensis stems were analyzed based on the biological processes regulated by the chemical targets and their interactions with disease-related targets.

[0114] GO enrichment analysis is attached Figure 4 The figure shows the enrichment results for biological process, cellular component, and molecular function. Biological processes primarily involve protein phosphorylation and negative regulation of apoptosis, while cellular components primarily involve mitochondria, endoplasmic reticulum membranes, and presynaptic membranes. Molecular functions primarily involve protein serine / threonine kinase activity and transmembrane receptor protein tyrosine kinase activity.

[0115] Figure 5 KEGG pathway annotation analysis revealed 174 relevant information pathways related to the target of the target component prediction. Among them, two relevant information pathways were highly frequent, namely the PI3K-Akt signaling pathway and the ErbB signaling pathway, and the key targets of these two pathways were GSK3B.

[0116] Example 3 Model construction and verification

[0117] 1. Construction of a GSK3B target inhibitor screening model:

[0118] 1) Download the dataset CHEMBL262 containing the high-throughput detection results of GSK3B target inhibitors from the CHEMBL database. 50 Compounds with concentrations below 500 nM were considered inhibitors, and IC 50 Compounds with concentrations above 500 nM were judged as non-inhibitors. The dataset was processed according to the following rules: (i) for duplicates with the same biological activity, only one was retained; (ii) for duplicates with different biological activities, all were removed.

[0119] 2) The dataset obtained from the CHEMBL database, filtered in step 1), was randomly split into a training set (80%), a validation set (10%), and a test set (10%), with the inhibitor / non-inhibitor ratio kept constant. A classifier was constructed using the deep learning module in Chemprop, and 20 rounds of hyperparameter optimization were performed. The model was automatically trained based on the optimal parameters for a total of 30 epochs. Hyperparameter optimization methods included automatic and manual hyperparameter optimization. Automatic hyperparameter optimization utilized random search, Bayesian optimization, and gradient-based optimization. The parameters obtained from automatic hyperparameter optimization were "depth": 2, "dropout": 0.3, "ffn_hidden_size": 200, "ffn_num_layers": 2, and "hidden_size": 200. The model was trained with these parameters, achieving an AUC of 0.834. After three manual hyperparameter optimizations, the model's optimal AUC was 0.921. The optimized parameters were then selected to train the GSK3B target inhibitor screening model. The model parameters at this time were "depth":4,"dropout":0.1,"ffn_hidden_size":800,"ffn_num_layers":3,"hidden_size":800. The ROC curves for the final parameters of automatic hyperparameter optimization and the optimal parameters of manual hyperparameter optimization are shown in the attached figure. Figure 6 .

[0120] 3) After model training is complete, the model performance is evaluated using four parameters: precision (Equation 1), recall (Equation 2), F1 score (Equation 3, the harmonic mean of precision and recall), and Matthews correlation coefficient (Equation 4) to ensure that the model performance meets the requirements of the target inhibitor screening model. Model performance evaluation indicators are calculated using the following formula:

[0121]

[0122] TP (True Positive): The actual example is a positive example and is predicted as a positive example; FP (False Positive): The actual example is a negative example, but is predicted as a positive example; TN (True Negative): The actual example is a negative example and is predicted as a negative example; FN (False Negative): The actual example is a positive example, but is predicted as a negative example.

[0123] 2. Model screening for prediction of potential active ingredients:

[0124] The final model was used to predict the test set. When a threshold of 0.7 was used to distinguish between positive and negative results, the model's precision, recall, F1 score, and Matthews correlation coefficient were 0.821, 0.810, 0.815, and 0.664, respectively. At this threshold, the model's precision and recall were relatively balanced, and the model was able to effectively distinguish positive compounds, meeting the requirements of a GSK3B target inhibitor screening model. Therefore, a threshold of 0.7 was selected as the threshold for distinguishing GSK3B inhibitors from non-inhibitors. The model was then used to predict the target inhibitory activity of the main chemical components of Coptis chinensis flower stalks. Compounds with a score of 0.7 or above were considered potential target inhibitors. The results of the deep learning screening are shown in Table 3. Coptis chinensis, berberine, and epiberberine were predicted to be GSK3B target inhibitors.

[0125] Table 3

[0126]

[0127] 3. Verification of model screening for potential active ingredients:

[0128] Molecular docking was used to verify the inhibitory effects of potential active ingredients. SDF files for seven major active ingredients from Coptis chinensis were obtained from the PubChem database. Protein information for GSK3B was searched from the Protein Database (RCSB) and downloaded in PDB format. The obtained protein receptors and compounds were converted to PDBQT format after removing water molecules, adding hydrogen, and calculating charges using AutoDock Tools 1.5.7. Molecular docking was performed using AutodockVina 1.1.2 software. Lower binding energies indicate more stable conformations. The results were analyzed, and the optimal conformations were visualized using Pymol 2.5.4.

[0129] As attached Figure 7, coptisine, berberine and epiberberine are tightly clustered with GSK3B through residues, and there are multiple interactions such as van der Waals forces and hydrogen bonds, and the binding free energy is less than -8.0 kcal / mol. Therefore, coptisine, berberine and epiberberine are confirmed to be effective GSK3B target inhibitors. The binding energies of calycosin, bamipine, wogonin and isorhamnetin with GSK3B are -7.6 kcal / mol, -7.9 kcal / mol, -7.8 kcal / mol and -7.8 kcal / mol, respectively, all greater than -8.0 kcal / mol, and are confirmed to be non-GSK3B inhibitors, which is consistent with the results of deep learning virtual screening. Through verification, it can be seen that the model constructed by the present invention has an accuracy rate of up to 100% in screening GSK3B target inhibitors among the active ingredients of Coptis chinensis flowers.

[0130] In summary, the inventors analyzed the use of Coptis chinensis stems for the treatment of insulin resistance using HPLC-MS / MS, network pharmacology, deep learning, and molecular docking. They identified 29 chemical components, 7 major active ingredients, and 152 targets potentially associated with the use of Coptis chinensis stems for the treatment of insulin resistance, with GSK3B as the primary target. The results suggest that coptisine, epiberberine, and berberine may be the primary active ingredients in Coptis chinensis stems for the treatment of insulin resistance, and that they exert their effects through the GSK3B target.

[0131] The "active ingredient-target-disease-pathway" network diagram and GO and KEGG enrichment analysis results showed that GSK3B is the main potential target of Coptis chinensis.

[0132] Therefore, the inventors speculated that Coptis chinensis may play a role in treating insulin resistance through the GSK3B target.

[0133] Deep learning virtual screening for GSK3B inhibitors revealed that coptisine, berberine, and epiberberine are potential candidate inhibitors. Furthermore, molecular docking results further confirmed that coptisine, berberine, and epiberberine have good binding affinity to the GSK3B inhibitory pocket. Therefore, based on deep learning virtual screening and molecular docking techniques, the inventors discovered that the main active ingredients in Coptis chinensis flower stems that treat insulin resistance are coptisine, berberine, and epiberberine.

[0134] It should be noted that the model constructed by the deep learning virtual screening technology proposed in this application is a screening model for all inhibitors of the GSK3B target. This model is not limited to screening active compounds in Coptis chinensis that mainly act on the GSK3B target. For any active ingredients of traditional Chinese medicine included in the existing database, this method can be used to screen out active ingredients related to the GSK3B target, laying the foundation for studying the mechanism of action of drugs.

[0135] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", "some implementation plans" or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0136] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for screening the regulatory pathway of the active ingredient of Coptis chinensis flower stalk for improving insulin resistance, characterized in that: include: 1) obtaining active ingredients from the flower stalks of Coptis chinensis; 2) screening the active target of the active ingredient according to database 1; 3) Screening disease targets for insulin resistance based on database 2; 4) drawing a Venn diagram of the active targets and the disease targets to obtain intersection targets; 5) Based on the intersection targets, a network diagram of active ingredients of Coptis chinensis stalks - intersection targets - insulin resistance - regulatory pathways was drawn using Cytoscape; 6) Based on the intersection targets, the intersection targets are analyzed using the STRING database to obtain core target genes; 7) Based on the core target genes, enrichment analysis was performed to obtain the regulatory pathway of the active ingredients of Coptis chinensis stalks that improve insulin resistance.

2. The method according to claim 1, characterized in that In step 1), the method for obtaining the active ingredient in the flower stalk of Coptis chinensis comprises: A1. Obtain the chemical components of the extract of Coptis chinensis stalk by HPLC-MS / MS; A2. Screen the chemical components with a peak area of ​​not less than 0.17, a bioavailability of not less than 30%, and a drug-like property of not less than 0.18 in the HPLC-MS / MS mass spectrum as the active ingredients in the Coptis chinensis flower stalk.

3. The method according to claim 1, characterized in that In step 2), the database 1 includes at least one of TCMSP, SwissTargetPrediction, ChEMBL, and STITCH; In step 3), the database 2 includes at least one of DisGeNET, GeneCards, OMIM, and CTD; The method for screening disease targets includes selecting a target in the database 2 having a Relevancescore value of not less than 10 in the GeneCards database as the disease target.

4. The method according to claim 1, wherein Before step 5), the method further includes: using KEGG enrichment analysis to screen the top 20 pathways with a P value less than 0.05 to construct a network diagram of the intersection of the active ingredients of the Coptis chinensis flower stems - target - insulin resistance - regulatory pathway, Step 6) further includes importing the intersection targets into the STRING database, selecting core target genes with a medium confidence value greater than 0.9, and drawing a protein interaction network diagram based on the core targets.

5. The method according to claim 1, wherein In step 7), the enrichment analysis includes GO enrichment and KEGG pathway annotation analysis.

6. A method for constructing a screening model for GSK3B target inhibitors, characterized in that: include: a. Obtain a dataset containing high-throughput detection results of GSK3B target inhibitors from Database 3; b. Preprocessing the data set, including: Preprocessing 1: IC 50 Compounds with concentrations below 500 nM were considered inhibitors, and the IC 50 Compounds above 500 nM were non-inhibitors; Preprocessing 2: Remove duplicates of the same compound in the dataset and compounds that are judged to be both inhibitors and non-inhibitors; c. Randomly split the preprocessed dataset into training, validation, and test sets, and construct a screening model for GSK3B target inhibitors based on the deep learning module in the directed message passing network model; d. training the screening model to obtain the final screening model, wherein the training includes hyperparameter optimization and threshold screening, in, The hyperparameter optimization includes a first optimization and a second optimization, wherein the first optimization is automatic hyperparameter optimization, and the second optimization is manual optimization based on parameter 1 obtained by the automatic hyperparameter optimization, and parameter 2 obtained by the second optimization is the optimal parameter of the screening model. The parameter 2 is "depth":4,"dropout":0.1,"ffn_hidden_size":800,"ffn_num_layers":3,"hidden_size":800, The threshold is an evaluation indicator that satisfies the precision and recall values ​​of the screening model to be greater than 0.8 at the same time by evaluating the precision, recall rate, harmonic mean of the precision and recall rate, and the Matthews correlation coefficient, and the evaluation indicator is not less than 0.

7.

7. The method according to claim 6, characterized in that In step a, the database 3 is CHEMBL.

8. The method according to claim 6, characterized in that In step c, the training set, validation set, and test set account for 70-90%, 5-15%, and 5-15% of the preprocessed data set, respectively. In step d, the automatic hyperparameter optimization method includes at least one of a random search method, a Bayesian optimization method, and a gradient descent method; The number of rounds of automatic hyperparameter optimization is 18-25 rounds; The number of rounds of the second optimization is 2-4 rounds; The model training cycle is 28-35; When the evaluation index of the compound in the model is not less than 0.7, it is determined to be a potential inhibitor of the GSK3B target.

9. A GSK3B target inhibitor screening model, characterized in that: Constructed by the method according to any one of claims 6 to 8.

10. Use of the GSK3B target inhibitor screening model according to claim 9 in screening drugs related to insulin resistance and / or screening compounds having GSK3B target inhibitors.

Images