Method for identifying convergence module nodes in biomolecular network

Abstract

The invention discloses a method for identifying convergence nodes in a biomolecular network. The method comprises the following steps: (1) determining the biomolecular network of which the convergence nodes need to be identified; (2) calculating a collective influence CI value, an overall comprehensive influence index IVI value and a CI-IVI value of each node in the biomolecular network; sorting according to a calculation result, and removing the node with the maximum CI-IVI value; then lambda is calculated; (4) if lambda is smaller than or equal to 1, identification is ended, and the removed nodes are convergence nodes; if lambda is greater than 1, performing the step (5); (5) recalculating to obtain the CI-IVI value of each remaining node of the biomolecular network, sequencing, removing the node with the maximum CI-IVI value, and calculating lambda; (6) if lambda is smaller than or equal to 1, identification is ended, and all removed nodes are a convergence node set; and if lambda is greater than 1, repeating the step (5) until lambda is less than or equal to 1. Compared with a known method for identifying convergence nodes in the biomolecular network, the method provided by the invention has higher effectiveness.

Description

Technical Field

[0001] The present invention belongs to the field of bioinformatics technology. Specifically, the present invention relates to a method for identifying convergent module nodes in complex molecular networks of diseases and drug interventions, such as protein interaction networks, gene expression regulation networks, drug metabolism and drug target networks, etc. Background Art

[0002] With the rise of the omics era and the continuous advancement of high-throughput technology, a large number of biological data sets have been generated. With biological items (such as proteins, genes, metabolites, etc.) as nodes and the relationships between node pairs as edges, a variety of virtual biomolecular networks are formed, such as protein interaction networks, gene expression regulation networks, and drug metabolism networks. Through effective analysis of these data networks, it is helpful to reveal the mechanisms of gene expression regulation, protein interaction, metabolite interaction, etc., which can then be applied in the fields of disease mechanism research and treatment, drug development, etc.

[0003] Studies have shown that there is a modular organizational structure in biomolecular networks. Identifying this modular structure and the key nodes in the module helps to simplify and reduce the dimensions of complex networks, understand biological systems, systematically reveal the complex mechanisms of drug action, and develop new drugs, especially new drugs for complex diseases. The inventor has disclosed a method for identifying key nodes in biomolecular networks in a Chinese invention patent application with publication number CN105956413A (publication date September 21, 2016). The method involves first identifying the key modules of the biomolecular network and then identifying the key nodes in the module.

[0004] As the study of biomolecular networks deepens, researchers have come to realize that there is a concept of convergence in biomolecular networks, which means that there is an upper limit for module nodes. When the module nodes exceed this upper limit, even if the number of nodes continues to increase, their influence will reach saturation.

[0005] Grasping the main contradiction in complex things is the key to solving problems. At present, researchers have carried out the identification of convergent modules or convergent nodes based on the K-core algorithm and CI (Collective influence) algorithm, but these algorithms often ignore the joint propagation factors of module nodes, fail to consider the impact of the relationship between modules in the local, semi-local and global topological dimensions of the network on the convergent module, and lack quantitative analysis of the convergent module nodes. Summary of the invention

[0006] In response to the above technical problems, the present invention provides a new method for identifying convergent module nodes in biomolecular networks. The method can capture all topological dimensions of the network and solve the positional deviation of the global centrality measure, thereby identifying convergent module nodes in the biomolecular network.

[0007] definition:

[0008] In this article, the "biomolecular network" refers to a network that exists in a biological system in different organizational forms, which is composed of nodes representing various types of biomolecules and edges representing the interaction relationship between biomolecules. Common biomolecular networks include gene transcription regulation networks, biological metabolism networks, epigenetic networks, phenotypic networks, signal transduction networks, protein interaction networks, etc.

[0009] In this article, the term "convergence module node" refers to a module node that is critical in both structure and function in the network. For example, it has an important influence on the information transmission of the network, the interaction of nodes, etc., and the impact of deleting this node on the network structure and information transmission exceeds the impact of deleting other nodes in the network.

[0010] In this article, the term "K-core algorithm" is an algorithm for heuristic network convergence node set. Its core idea is to measure the nodes based on their coreness in the network, where nodes located in the core area of ​​the network have higher influence than edge nodes. The specific implementation process is as follows: First, the isolated nodes in the network are regarded as nodes with a coreness of 0 and are removed from the network. Subsequently, a core decomposition operation is performed. In this process, all nodes with a coreness of 1 are first removed from the network, and then nodes with a residual degree less than or equal to 1 are continuously removed until the residual degrees of the remaining nodes in the network are all greater than 1. These removed nodes are collectively referred to as 1-core nodes, and their coreness value is 1. Next, the nodes with a coreness of 2 in the network are processed in the same way. This process continues until all nodes in the network are removed. Under this method, nodes with higher coreness are usually located in the center of the network and have greater influence.

[0011] In this paper, the term “CI” refers to collective influence. The CI value of each node in a molecular biological network can be calculated using formula (1).

[0012] CI i =(d i -1)∑ j∈δB(i，l) (d j -1) (1),

[0013] Among them, d i ,d jRespectively represent the degrees of nodes i and j; δB(i,l) is the boundary of the ball consisting of nodes whose shortest path to node i is of length l, and l=1 is generally set.

[0014] The term "CI algorithm" is an algorithm for converging nodes in a network based on the collective influence CI value. The main idea of ​​the algorithm is to remove the node with the largest CI value in each iteration to quickly collapse the network until λ ≤ 1 calculated by formula (2).

[0015]

[0016] Where N is the total number of nodes, is the average degree of the network, CI i is the collective influence value of node i, and l is generally set to 1. When λ≤1, the removal process will stop, which means that most giant components disappear.

[0017] In this paper, the term "IVI value" refers to the integrated value if influence, which reflects degree centrality (DC), cluster rank (CR), neighborhood connectivity (NC), local H index (H), betweenness centrality (BC) and collective influence (CI). The IVI value of each node in a molecular biological network can be calculated by formula (3).

[0018]

[0019] Among them, DC i , NC i , CR i , B.C. i and CI i They are the scope-normalized degree centrality, scope-normalized local H-index, scope-normalized domain connectivity, scope-normalized cluster rank, scope-normalized betweenness centrality, and scope-normalized collective influence of node i, where CI i Calculated by formula (1).

[0020] The present invention adopts the following technical solution:

[0021] A method for identifying convergence nodes in a biomolecular network comprises the following steps:

[0022] (1) Determine the biomolecular network in which convergence nodes need to be identified;

[0023] (2) Calculate the collective influence CI value of each node in the biomolecular network according to formula (1),

[0024] CI i =(d i -1)Z i∈δB(i，l) (d j -1) (1),

[0025] Among them, d i ,d j denote the degrees of nodes i and j respectively; δB(i,l) is the boundary of the ball consisting of nodes whose shortest path to node i is of length l, where l = 1;

[0026] Then substitute into formula (3) to calculate the overall comprehensive impact index IVI value of each node,

[0027]

[0028] Among them, DC i , NC i , CR i , B.C. i and CI i They are the scope-normalized degree centrality, scope-normalized local H-index, scope-normalized domain connectivity, scope-normalized cluster rank, scope-normalized betweenness centrality, and scope-normalized collective influence of node i;

[0029] Then substitute the IVI value into formula (4) to calculate the CI-IVI value of each node;

[0030] CI-IVI i =(IVI i -1)Σ j∈δB(i，l) (IVI j -1) (4),

[0031] Among them, IVI i and IVI j are the overall comprehensive impact index values ​​of nodes i and j, respectively. δB(i,l) is the sphere boundary composed of nodes whose shortest path to node i is of length l, where l = 1.

[0032] Sort by the calculation results and remove the node with the largest CI-IVI value; then calculate λ according to formula (5):

[0033]

[0034] Where N is the total number of nodes, is the average degree of the network, CI-IVI i Calculated by formula (4);

[0035] (4) If λ≤1, the identification is terminated and the removed node is the convergent node; if λ>1, proceed to step (5);

[0036] (5) recalculating the CI-IVI value of each remaining node in the biomolecular network according to formulas (1), (3) and (4), sorting, removing the node with the largest CI-IVI value, and calculating λ according to formula (5);

[0037] (6) If λ≤1, the identification is terminated and all removed nodes are the convergent node set; if λ>1, repeat step (5) until λ≤1.

[0038] Preferably, the biomolecular network is selected from a gene expression regulatory network, a protein interaction network, a biological metabolism network, an epigenetic network, a phenotypic network, a signal transduction network, etc.; more preferably, it is a gene expression regulatory network.

[0039] The biomolecular network can be any known in the art, such as reported in patents or non-patent literature. It can also be constructed for a specific purpose. For example, the biomolecular network can be the "lung adenocarcinoma metastasis prognosis differential gene regulatory network" constructed in the embodiment.

[0040] Typically, the biomolecular network contains ≥3 nodes and ≥3 edges. For example, the "lung adenocarcinoma metastasis prognosis differential gene regulatory network" constructed in the examples herein contains 90 nodes and 157 edges.

[0041] The method for identifying convergent nodes in a biomolecular network described in the present invention innovatively integrates the collective influence CI value and the overall comprehensive influence index IVI value. The basic flow chart is shown in Figure 1 As shown, the basic idea can be summarized as follows:

[0042] (1) Calculate the node's overall comprehensive impact index IVI, and further calculate the node's CI-IVI value based on the node's overall comprehensive impact index IVI.

[0043] (2) Delete the node with the largest CI-IVI value in each iteration to quickly collapse the network. That is, the node with the largest CI-IVI value will be removed, and then the new CI-IVI value of each node remaining in the network will be calculated; and the λ value will be updated. When λ≤1, it indicates that most of the giant components in the removal process disappear. Finally, a list of converged nodes is obtained.

[0044] Complex networks are composed of nodes and the relationships (edges) between them. In the network, the status of module nodes varies. In fact, convergence module nodes have different functions for the network. When the convergence module nodes are removed, the network will collapse critically. Therefore, the identification of convergence module nodes is very important for understanding the mechanism of action of the entire network. Therefore, the study of convergence module nodes lays the foundation for exploring the main pharmacological mechanism of synergistic effects of combined application of drugs for complex diseases; it lays the foundation for further analyzing the relationship between drugs, guiding combined drug use and disease treatment, and provides a scientific basis for guiding disease treatment and drug development.

[0045] In this regard, the present invention proposes the innovative concept of identifying convergent module nodes for complex networks of diseases and drug interventions, and specifically provides a new method for identifying convergent module nodes from biomolecular networks by integrating CI and IVI for this concept. In this specific identification process, all topological dimensions of the network can be captured, the positional deviation of the centrality metric can be solved, and the convergence module nodes can be quantitatively identified based on the idea of ​​complex network percolation. The method of the present invention comprehensively considers multiple important network centrality metric results, including degree centrality, cluster registration, neighborhood connectivity, local H-index, betweenness centrality, and collective influence. Each centrality metric captures different topological dimensions of the network, including local, semi-local, and global topological dimensions. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

[0047] Figure 1 A flow chart of the convergence module node identification method of the present invention is shown.

[0048] Figure 2 The constructed gene regulatory network for differential prognosis of lung adenocarcinoma metastasis is shown.

[0049] Figure 3 The results of identifying convergent module nodes of lung adenocarcinoma metastasis prognosis differential gene regulatory network based on K-core method are shown. The nodes with different colors from other nodes in the figure are identified convergent module nodes.

[0050] Figure 4 The results of identifying convergent module nodes of lung adenocarcinoma metastasis prognosis differential gene regulatory network based on CI method are shown. The nodes with different colors from other nodes in the figure are identified convergent module nodes.

[0051] Figure 5 The results of identifying convergent module nodes of the lung adenocarcinoma metastasis prognosis differential gene regulatory network based on the CI-IVI method of the present invention are shown. The nodes with different colors from other nodes in the figure are identified convergent module nodes.

[0052] Figure 6The results of the changes in the scale G(q) of the largest connected component of the network and the number C(q) of connected components after deleting each convergence module node from the differential gene regulatory network of lung adenocarcinoma metastasis prognosis based on different methods are shown. The horizontal axis q of Figure A and Figure B represents the deletion ratio of the node, the vertical axis G(q) of Figure A represents the scale of the giant connected component, which is the proportion of the giant connected component in the network after deleting nq number of nodes based on a certain convergence method in the initial network G scale, and the vertical axis C(q) of Figure B is the number of connected graphs in the network after deleting nq number of nodes based on a certain convergence method. DETAILED DESCRIPTION

[0053] The present invention is described below with reference to specific examples. It will be appreciated by those skilled in the art that these examples are only used to illustrate the present invention and are not intended to limit the scope of the present invention in any way.

[0054] Example 1 Construction of gene regulatory network for metastatic prognostic differences in lung adenocarcinoma

[0055] 1.1 Differential genes associated with lung adenocarcinoma metastasis

[0056] A total of 344 (M0) and 25 (M1) lung adenocarcinoma RNAseq data and corresponding clinical information were obtained from The Cancer Genome Atlas (TCGA). 1444 differentially expressed genes between metastatic and non-metastatic lung adenocarcinoma were calculated and obtained.

[0057] 1.2 Prognostic genes among differentially expressed genes associated with lung adenocarcinoma metastasis

[0058] Multivariate Cox survival analysis was performed based on differential gene expression and OS of patients. 423 genes associated with patient survival outcomes were identified (P-value<0.05).

[0059] 1.3 Construction of gene regulatory network for differential prognosis of lung adenocarcinoma metastasis

[0060] The lung adenocarcinoma metastasis differential regulation network was constructed by mapping 423 prognostic differential genes to the human regulatory disease network. After analysis, the network contained 171 nodes and 157 edges. In order to maximize the node perturbation in the disease network, isolated nodes and unconnected parts in the network were removed ( Figure 2 The final lung adenocarcinoma metastasis prognostic difference network consists of 90 nodes and 157 edges, as shown in Table 1 and Figure 2 shown.

[0061] Table 1 Topological parameters of gene regulatory network for differential prognosis of lung adenocarcinoma metastasis

[0062]

[0063] For details on the construction of the gene regulatory network for differential prognosis of lung adenocarcinoma metastasis, please refer to: Liu Xi, Research on the optimization of traditional Chinese medicine formula and pharmacodynamic mechanism for inhibiting lung adenocarcinoma metastasis based on network target convergence [D], China Academy of Chinese Medical Sciences, 2024.

[0064] Example 2 Example 1 Identification of convergence module nodes of the gene regulatory network for differential prognosis of lung adenocarcinoma metastasis constructed

[0065] 2.1 Identifying convergence module nodes based on the K-core method

[0066] The isolated nodes in the network are regarded as nodes with a core degree of 0 and are removed from the network. Subsequently, a core decomposition operation is performed. In this process, all nodes with a core degree of 1 are first removed from the network, and then the nodes with a residual degree less than or equal to 1 are continuously removed until the residual degrees of the remaining nodes in the network are all greater than 1. These removed nodes are collectively referred to as 1-core nodes, and their core degree value is 1. Next, the nodes with a core degree of 2 in the network are processed in the same way. This process continues until all nodes in the network are removed. Therefore, the maximum core degree of the network is 4, and the nodes (genes) included are shown in Table 2. These node sets are the convergent node sets identified by the K and method, see for details. Figure 3 .

[0067] Table 2 Convergence node set of K-kernel method

[0068]

[0069] 2.2 CI method to identify convergence module nodes

[0070] The collective influence CI value of each node in the lung adenocarcinoma metastasis prognosis differential gene regulatory network is calculated according to formula (1):

[0071] CI i =(d i -1)∑ j∈δB(i，l) (d j -1) (1)

[0072] Among them, d i ,d j denote the degrees of nodes i and j respectively; δB(i,l) is the boundary of the ball consisting of nodes whose shortest path to node i is of length l, where l=1.

[0073] Remove the node with the largest CI value (iteratively) to quickly collapse the network, and calculate the λ value by formula (2);

[0074]

[0075] Where N is the total number of nodes, is the average degree of the network, CIi is the collective influence value of node i, and l=1 is generally set.

[0076] After the first removal, λ>1, and the CI values ​​of all remaining nodes in the network are calculated. The node with the largest CI value is removed for the second time, and the λ value is calculated. In this iterative cycle, when the 11th node is removed, λ≤1, and the removal is stopped. Therefore, the 11 removed nodes (genes) are convergent module nodes, which constitute the convergent module node set identified by the CI method. See Table 3 and Figure 4 .

[0077] Table 3 Convergence node set of CI method

[0078]

[0079] 2.3 CI-IVI method to identify convergence module nodes

[0080] The collective influence CI value of each node in the lung adenocarcinoma metastasis prognosis differential gene regulatory network is calculated according to formula (1). Then the calculated CI value is substituted into formula (3) to calculate the overall comprehensive impact index IVI value of each node,

[0081]

[0082] Among them, DC i , NC i , CR i , B.C. i and CI i They are the scope-normalized degree centrality, scope-normalized local H-index, scope-normalized domain connectivity, scope-normalized cluster rank, scope-normalized betweenness centrality, and scope-normalized collective influence of node i;

[0083] Then substitute the IVI value into formula (4) to calculate the CI-IVI value of each node;

[0084] CI-IVI i =(IVI i -1)Z j∈δB(i，l) (IVI j -1) (4),

[0085] Among them, IVI i and IVI j are the overall comprehensive impact index values ​​of node i and node j respectively, δB(i,l) is the sphere boundary composed of nodes whose shortest path length is l, and l=1.

[0086] Sort by the calculation results and remove the node with the largest CI-IVI value (first iteration); then calculate λ according to formula (5):

[0087]

[0088] Where N is the total number of nodes, is the average degree of the network, CI-IVI i Calculated by formula (4).

[0089] The calculated result λ>1, and the CI-IVI value of each remaining node in the network is recalculated again according to formulas (1), (3) and (4), sorted, and the node with the largest CI-IVI value is removed (second iteration), and the λ value is calculated according to formula (5). This iterative cycle is repeated. When the 16th node is removed, λ≤1, and the removal is stopped. Therefore, the 16 removed nodes (genes) are convergent module nodes, which constitute the convergent module node set identified by the CI-IVI method. See Table 4 and Figure 5 .

[0090] Table 4 Convergence node set of CI-IVI method

[0091]

[0092] 2.4 Performance comparison of different methods for identifying convergent node sets

[0093] The performance of the three methods for identifying convergent module nodes was compared by comparing the network node deletion ratio (q), the size of the largest connected component of the network (G(q)) and the number of connected components of the network (C(q)) after deleting each convergent module node from the differential gene regulatory network for metastasis prognosis of lung adenocarcinoma based on different methods.

[0094] like Figure 6 As shown in Figure A, in the network, when q is approximately equal to 0.51, the scale of G(q) in the network is directly reduced from 74 to 4 (the two values ​​are the product of the ordinate value of Figure A and the initial network scale); after the CI method, to achieve the same degree of network damage, the proportion of nodes to be deleted q needs to reach 0.52, while using the K-core method, the proportion of node deletion needs to be increased to 0.74. It can be seen that the performance of the CI-IVI method is improved compared with the CI method and the K-core method under the condition of the same degree of network damage. Figure 6 Figure B shows that, corresponding to G(q), C(q) increases sharply with the increase of q in the initial stage, and then shows a downward trend after reaching the extreme value. Under this indicator, CI-IVI performs significantly better than CI method and K-core method.

[0095] In addition, Table 5 shows the performance of the convergence module node sets identified by different methods in the network destructive experiment of node immunity. Specifically, when the convergence node set identified by the K-core method is deleted in the network, G(q) is 0.48 and C(q) is 16 in the network; when the convergence node set identified by the CI method is deleted, G(q) is 0.11 and C(q) is 29; and when the convergence node set identified by the CI-IVI method of the present invention is deleted, G(q) is 0.10 and C(q) is 29. In network analysis, when the value of G(q) is small and the value of C(q) is large, it indicates that the network splitting phenomenon is more serious. This phenomenon reflects that the deleted node set plays a more critical role in the network, that is, it has a greater influence. Therefore, from the perspective of evaluating the impact of node immunity strategy on network destructiveness, the CI-IVI method shows higher effectiveness than the K-core and CI methods.

[0096] Table 5 Performance of different convergence node sets in the network destructive experiment of node immunity

[0097]

[0098] In summary, among all the methods, the CI-IVI method performed the best.

[0099] Example 3 Application and verification of convergence module node identification

[0100] 3.1 Application of the convergence module nodes of the gene regulatory network for differential prognosis of lung adenocarcinoma metastasis identified The convergence module nodes identified in Example 2 were used as targets, and the compound information of the TCMSP database (http: / / old.TCMSP-e.com / TCMSP.PHP), TCMID database (http: / / www.megabionet.org / tcmid / ) and CTD database (http: / / ctdbase.org) was used to establish the association between "target and chemical component". The compounds were screened according to scientific and rigorous condition indicators such as OB>20 and Druglike≥0.18. After integrating the compound information in the three databases and comparing it with the existing TCMSP database, compounds related to traditional Chinese medicine components were successfully screened. In order to more clearly reveal the mutual relationship between targets, compounds and traditional Chinese medicines, the corresponding traditional Chinese medicines were preliminarily evaluated, and the relationship between "core target-compound" and "compound-traditional Chinese medicine" was corresponded to each other, and the corresponding relationship between "convergence module node (target)-compound-traditional Chinese medicine" was established.

[0101] The 16 convergence module nodes (targets) identified by the CI-IVI method correspond to a total of 1136 compounds in the CTD, TCMSP and TCMID databases. According to the name, 36 compounds can be corresponding to them in the TCMSP database. Through the pubmed ID correspondence, 9909 IDs can be obtained, corresponding to 541 compounds. According to the restrictions of OB≥20, Druglike≥0.18 and TCM frequency≥8, the following TCM formula was screened: Lobelia, Atractylodes, Amomum villosum, Atractylodes macrocephala, Lysimachia chinensis, Chrysanthemum, named "Bai Lian Fang". For details, see: Liu Xi, Research on the optimization and efficacy mechanism of TCM formula for inhibiting lung adenocarcinoma metastasis based on network target convergence [D], China Academy of Chinese Medical Sciences, 2024.

[0102] The six convergence module nodes (targets) identified by the K-core method correspond to a total of 722 compounds in the CTD, TCMSP and TCMID databases. According to the name, 28 compounds can be matched in the TCMSP database. Through the pubmed ID correspondence, 6735 IDs can be obtained, corresponding to 367 compounds. According to the restrictions of OB≥30, Druglike≥0.18 and TCM frequency greater than or equal to 6, the following TCM formula was screened: Lobelia, Chrysanthemum, Licorice, Lysimachia, Elsholtzia, Inula, Ginkgo, named "Yinlian Fang". For details, see: Liu Xi, Research on the optimization and efficacy mechanism of TCM formula for inhibiting lung adenocarcinoma metastasis based on network target convergence [D], China Academy of Chinese Medical Sciences, 2024.

[0103] 3.2 Zebrafish pharmacodynamics experimental verification

[0104] 3.2.1 Materials

[0105] 3.2.1.1 Experimental drugs:

[0106] ① Chinese medicine composition: Extract powder 1 (brown yellow, Bailian Fang Chinese medicine composition): content is 2.45g / g (raw drug amount / extract powder), extract powder 2 (brown yellow, Yinlian Fang Chinese medicine composition): content is 3.40g / g (raw drug amount / extract powder).

[0107] The above-mentioned extract powder preparation method is:

[0108] Extract powder 1: 15g of Lobelia, 15g of Lysimachia chinensis, 10g of Chrysanthemum, 10g of Atractylodes macrocephala, 9g of Atractylodes lancea, and 6g of Amomum villosum; mix each of the medicinal materials, add 5 to 10 times the weight of water, heat to boiling, then boil for 25 minutes, and filter while hot; add 5 to 10 times the weight of water to the residue, heat to boiling, boil for 20 to 30 minutes, filter while hot, combine the two medicinal liquids, and then concentrate to a relative density of 1.3 to 1.4 at 80°C.

[0109] Extract powder 2: 15g of Lobelia, 15g of Lysimachia chinensis, 10g of Chrysanthemum, 3g of Licorice, 6g of Elsholtzia ciliata, 6g of Inula japonica, and 10g of Ginkgo; mix each medicinal material, add 5 to 10 times the weight of water, heat to boiling, then boil for 25 minutes, and filter while hot; add 5 to 10 times the weight of water to the residue, heat to boiling, boil for 20 to 30 minutes, filter while hot, combine the two medicinal liquids, and then concentrate to a relative density of 1.3 to 1.4 at 80°C.

[0110] ② Paclitaxel, white powder, Shanghai Aladdin Biochemical Technology Co., Ltd., batch number J2126744, stored at 4°C. Prepared into 20.0 mM stock solution with DMSO, diluted as needed, and stored at -20°C.

[0111] 1.2 Experimental animals

[0112] ① Wild-type AB strain zebrafish were bred in natural pairs, with a total of 300 fish, 30 fish in each experimental group, and the age was 3 dpf, for the determination of the maximum detection concentration (MTC) of extract powder 1.

[0113] ② Wild-type AB strain zebrafish were bred in natural pairs, with a total of 1,200 fish, 30 fish in each experimental group, and the age was 2 dpf, for evaluation of the inhibitory effect of extract powder 1 on non-small cell lung cancer metastasis.

[0114] The fish were kept in 28°C fish farming water (water quality: 200 mg instant sea salt was added to 1 L of reverse osmosis water, conductivity was 480-510 μS / cm; pH was 6.9-7.2; hardness was 53.7-71.6 mg / LCaCO3), and the experimental animal use license number was: SYXK (Zhejiang) 2022-0004. The feeding and management met the requirements of the international AAALAC certification.

[0115] 3.2.1.3 Experimental instruments:

[0116] ①Dissecting microscope (SZX7, OLYMPUS, Japan);

[0117] ②CCD camera (VertA1, Shanghai Tusen Vision Technology Co., Ltd., China);

[0118] ③Microinjection instrument (IM300, NARISHIGE, Japan);

[0119] ④Electric focusing and continuous zoom fluorescence microscope (AZ100, Nikon, Japan);

[0120] ⑤ Precision electronic balance (CP214, OHAUS, America);

[0121] ⑥6-well plate (Nest Biotech, Shanghai, China).

[0122] 3.2.2 Methods

[0123] (1) Concentration groups

[0124] Experimental group 1 Model control group Experimental group 2 Positive control drug Paclitaxel 1.00 pmol / tail Experimental group 3 Positive control drug Paclitaxel 2.00 pmol / tail Experimental group 4 Positive control drug Paclitaxel 4.00 pmol / tail Experimental group 5 Extract powder 1 31.2 μg / mL

[0125] Experimental group 6 Extract powder 1 62.5μg / mL

[0126] Experimental group 7 Extract powder 1 125μg / mL

[0127] Experimental group 8 Extract powder 2 31.2μg / mL

[0128] Experimental 9 group extract powder 2 62.5μg / mL

[0129] Experiment 10 group extract powder 2 125μg / mL

[0130] (2) Model making

[0131] Human non-small cell lung cancer (NCI-H460 and A549) cells were labeled with green fluorescent dye and transplanted into the zebrafish yolk sac by microinjection, with about 200 cells transplanted per tail, to establish a zebrafish lung cancer tumor cell transplantation model.

[0132] (3) Experimental methods

[0133] ① Determination of MTC

[0134] 330 3dpf wild-type AB strain zebrafish were randomly selected in a six-well plate, and 30 zebrafish were treated in each well (experimental group). The drug concentrations of extract powder 1 were 125, 250, 500, 1000 and 2000 μg / mL, respectively. A normal control group was set up at the same time, and the volume of each well (experimental group) was 3mL. The zebrafish were cultured in a 35℃ incubator until 5dpf, and the death of zebrafish was counted. The MTC of extract powder 1 on normal zebrafish was determined based on the toxic reaction and death of zebrafish.

[0135] ②Evaluate the inhibitory effect of Chinese medicine combination on the growth and metastasis of non-small cell lung cancer (NCI-H460 and A549)

[0136] Human non-small cell lung cancer (NCI-H460 and A549) cells were labeled with green fluorescent dye and transplanted into the yolk sac of wild-type AB zebrafish at 2dpf by microinjection. About 200 cells were transplanted per tail to establish a zebrafish human non-small cell lung cancer transplantation model; zebrafish injected with human non-small cell lung cancer cells were cultured at 35°C until 3dpf. At 3dpf, zebrafish with good consistency of transplanted tumors were selected under a microscope and randomly assigned to 6-well plates, 30 tails per well, and the fish water capacity of each well was 3mL. The extract powder 1 and extract powder 2 were respectively dissolved in water at concentrations of 31.2, 62.5 and 125μg / mL, and the positive control group was paclitaxel 1.00pmol / tail, 2.00pmol / tail and 4.00pmol / tail, and a model control group was set up at the same time. The cells were incubated at 35°C until 5 dpf. Ten zebrafish were randomly selected from each experimental group for observation under a fluorescence microscope, photographed, and saved. Nikon NIS-Elements D 4.30.00 advanced image processing software was used for image analysis to calculate the fluorescence intensity of zebrafish transplanted tumors and the metastasis distance (M) of the tumor cells in the tail vein. The statistical analysis results of fluorescence intensity and metastasis distance were used to evaluate the inhibitory effect of extract powder 1 on the growth and metastasis of human non-small cell lung cancer in zebrafish. The statistical processing results were expressed as mean ± SE, and the inhibitory effect on the growth and metastasis of zebrafish tumors was calculated as follows:

[0137]

[0138] One-way analysis of variance and Dunnett's T test were used for statistical analysis, and p < 0.05 indicated significant differences. Representative experimental graphs are provided.

[0139] (3) Experimental results

[0140] ① Determination of MTC

[0141] Extract powder 1 induced no more than 10% death and obvious toxicity in normal zebrafish at a concentration of 500 μg / mL, and did not induce death and toxicity phenotypes in tumor model zebrafish at a concentration of 250 μg / mL. Therefore, the MTC of the growth inhibition effect of extract powder 1 on zebrafish human metastatic non-small cell lung cancer was determined to be 250 μg / mL, and the MTC of the growth inhibition effect of extract powder 2 on zebrafish human metastatic non-small cell lung cancer was determined to be 125 μg / mL. See Table 6 for details.

[0142] Table 6 "Concentration-lethality" results of Chinese medicine composition after treating normal zebrafish (n=30)

[0143]

[0144] ②Inhibitory effect of Chinese herbal medicine combination on the growth of NCI-H460 transplanted tumor

[0145] The fluorescence intensity of zebrafish NCI-H460 transplanted tumors at paclitaxel concentrations of 1.00 pmol / tail, 2.00 pmol / tail, and 4.00 pmol / tail was 255176±15040, 181161±21683, and 144306±11680 pixels. There was a significant difference (p<0.01) between the high-dose paclitaxel group and the model control group (27962.50±24644 pixels), and the tumor growth inhibition effect was 48.4%, indicating that the positive control drug can inhibit the growth of NCI-H460 transplanted tumors.

[0146] The fluorescence intensity of zebrafish NCI-H460 transplanted tumors in the 31.2, 62.5 and 125 μg / mL concentration groups of extract powder 1 were 275152±28973, 162435±19859, and 171974±18511 pixels, respectively. Among them, the medium and high dose groups were significantly different from the model control group (27962.50±24644 pixels) (respectively, p<0.01 and p<0.01), and the tumor growth inhibition effects were 41.91% and 38.50%, respectively. See Table 7 for details.

[0147] The fluorescence intensity of zebrafish NCI-H460 transplanted tumors in the 31.2, 62.5, and 125 μg / mL concentration groups of extract powder 2 were 228483±32757, 185791±24387, and 155179±12587 pixels, respectively. Among them, the high-dose group was significantly different from the model control group (27962.50±24644 pixels) (p<0.01), and the tumor growth inhibition effect was 44.51%. See Table 7 for details.

[0148] The growth inhibition effect of the medium-dose group of extract powder 1 on zebrafish NCI-H460 transplanted tumors was significantly different from that of the model group (p<0.01); while the tumor growth inhibition effect of the high-dose group of extract powder 2 and paclitaxel was significantly different from that of the model control group. See Table 7 for details.

[0149] Table 7 Inhibitory effect of Chinese medicine composition on the growth of zebrafish NCI-H460 transplanted tumor (n=10)

[0150]

[0151] Note: Compared with the model control group, **p<0.01, ***p≤0.001.

[0152] ③Inhibitory effect of Chinese herbal medicine combination on metastasis of NCI-H460 transplanted tumor

[0153] The metastasis distance of NCl-H460 tumor cells in zebrafish tail vein at paclitaxel concentrations of 1.00 pmol / tail, 2.00 pmol / tail and 4.00 pmol / tail was 119950±10289, 93602±13052 and 80031±9542, and there was no statistical difference (p>0.05) between each concentration group and the model control group (119209±17587 pixels), indicating that the positive control drug at this concentration could not inhibit the metastasis of NCI-H460 transplanted tumors. See Table 8 for details.

[0154] The metastasis distances of zebrafish caudal vein NCl-H460 tumor cells in the 31.2, 62.5 and 125 μg / mL concentration groups of extract powder 1 were 71861±14093, 27380±3142, and 27755±2845, respectively. Among them, the medium and high dose groups were significantly different from the model control group (119209±17587 pixels) (respectively, p<0.001 and p<0.001), and the tumor metastasis inhibition effects were 77.03% and 76.72%, respectively; indicating that the medium and high doses of extract powder 1 have a significant inhibitory effect on tumor metastasis. See Table 8 for details.

[0155] The metastasis distances of NCl-H460 tumor cells in the caudal vein of zebrafish in the 31.2, 62.5, and 125 μg / mL concentration groups of extract powder 2 were 76867±10817, 34809±3662, and 35008±5693, respectively. Among them, the medium and high dose groups were significantly different from the model control group (119209±17587 pixels) (respectively, p<0.001 and p<0.001), and the tumor metastasis inhibition effects were 70.80% and 70.63%, respectively; indicating that the medium and high doses of extract powder 2 also have a significant inhibitory effect on tumor metastasis. However, compared with the same dose level of extract powder 1, the tumor inhibition rate of the medium and high dose groups of extract powder 2 was significantly lower than that of the same dose group of extract powder 1 (p<0.05); indicating that the growth inhibition effect of extract powder 1 on zebrafish NCl-H460 transplanted tumors is significantly stronger than that of extract powder 2. See Table 8 for details.

[0156] Table 8 The inhibitory effect of the Chinese medicine composition on the metastasis of zebrafish NCI-H460 transplanted tumor (n=10)

[0157]

[0158] Note: Compared with the model control group, **p<0.01, ***p≤0.001; compared with the same level dose group of extract powder 1, # p<0.05.

[0159] Therefore, the metastasis inhibition effect of extract powder 1 on zebrafish NCI-H460 transplanted tumor was stronger than that of extract powder 2 and the positive control paclitaxel group.

[0160] ④ The inhibitory effect of Chinese herbal medicine combination on the growth of A549 transplanted tumor

[0161] The fluorescence intensity of zebrafish A549 transplanted tumors at 1.00 pmol / tail, 2.00 pmol / tail and 4.00 pmol / tail concentrations of paclitaxel was 326139±37347, 269894±24063, 249623±25639 pixels, which were significantly different from the model control group (494191±44230 pixels) (respectively, p<0.01 and p<0.001, p<0.001), and the tumor growth inhibition effect was 34.01%, 45.39%, 49.49%, indicating that the positive control drug can significantly inhibit the growth of A549 transplanted tumors. See Table 9 for details.

[0162] The fluorescence intensity of zebrafish A549 transplanted tumor in the 32.5, 62.5 and 125 μg / mL concentration groups of extract powder 1 were 408184±56037, 262577±31503 and 248242±20246 pixels, respectively. Among them, the medium and high dose groups were significantly different from the model control group (494191±44230 pixels) (p<0.01 and p<0.01, respectively), and the tumor growth inhibition effects were 46.87% and 49.77%, respectively. This shows that extract powder 1 has a significant inhibitory effect on the growth of zebrafish A549 transplanted tumors. See Table 9 for details.

[0163] The fluorescence intensity of zebrafish A549 transplanted tumor in 32.5, 62.5 and 125 μg / mL concentration groups of extract powder 2 were 405119±39223, 299785±31821, and 196138±29992 pixels, respectively. Among them, the medium and high dose groups were significantly different from the model control group (494191±44230 pixels) (p<0.01 and p<0.001, respectively), and the tumor growth inhibition effects were 39.34% and 60.31%, respectively.

[0164] Table 9 Inhibitory effect of Chinese medicine composition on the growth of zebrafish A549 transplanted tumor (n=10)

[0165]

[0166] Compared with the model control group, **p<0.01, ***p≤0.001; compared with the same dose group of extract powder 1, # p<0.05.

[0167] Therefore, extract powder 1 and extract powder 2 both have an inhibitory effect on the growth of zebrafish A549 transplanted tumor.

[0168] (4) The inhibitory effect of the Chinese medicine composition on the metastasis of A549 transplanted tumors

[0169] The metastasis distance of zebrafish A549 tail vein tumor cells at 1.00pmol / tail, 2.00pmol / tail and 4.00pmol / tail concentrations of paclitaxel was 31673±3467, 31137±3553 and 29941±2953 pixels, which were significantly different from the model control group (70006±8870 pixels) (respectively, p<0.001 and p<0.001, p<0.001), and the tumor metastasis inhibition effect was 54.76%, 55.52% and 57.23%, indicating that the positive control drug can significantly inhibit the metastasis of A549 transplanted tumors. See Table 10 for details.

[0170] The metastasis distances of zebrafish A549 tail vein tumor cells in the 32.5, 62.5 and 125 μg / mL concentration groups of extract powder 1 were 37236±5113, 30906±2519 and 23285±2334 pixels, respectively; among them, the medium and high dose groups were significantly different from the model control group (70006±8870 pixels) (p<0.05 and p<0.001, respectively), and the tumor growth inhibition effects were 55.85% and 66.74%, respectively. This shows that extract powder 1 has a significant inhibitory effect on the metastasis of zebrafish A549 transplanted tumors. See Table 10 for details.

[0171] The metastasis distances of zebrafish A549 tail vein tumor cells in the 32.5, 62.5 and 125 μg / mL concentration groups of extract powder 2 were 36060±4114, 33075±2991 and 26201±3574 pixels, respectively. There were significant differences between the high-dose group and the model control group (70006±8870 pixels) (p<0.05 and p<0.001), and the tumor metastasis inhibition was 52.75% and 62.57%. However, the inhibition of zebrafish A549 transplanted tumor metastasis in each dose group of extract powder 2 was significantly lower than that in the same dose group of extract powder 1 (p<0.05), indicating that the inhibitory effect of extract powder 1 on zebrafish A549 transplanted tumors was significantly stronger than that of extract powder 2. See Table 10 for details.

[0172] Table 10 The inhibitory effect of the Chinese medicine composition on the metastasis of zebrafish A549 transplanted tumor (n=10)

[0173]

[0174] Compared with the model control group, **p<0.01, ***p≤0.001; compared with the same dose group of extract powder 1, # p<0.05.

[0175] Through the above example analysis, it is proved that the method of the present invention can effectively identify the convergent target module of the disease in the biological network, and has advantages over other convergent module node identification methods, providing a scientific basis for guiding disease treatment and drug development.

[0176] The above description of the specific embodiments of the present invention does not limit the present invention. Those skilled in the art may make various changes or modifications based on the present invention. As long as they do not depart from the spirit of the present invention, they should all fall within the scope of the claims attached to the present invention.

Claims

1. A method for identifying convergence nodes in a biomolecular network, comprising the following steps: (1) Determine the biomolecular network in which convergence nodes need to be identified; (2) Calculate the collective influence CI value of each node in the biomolecular network according to formula (1), CI i =(d i -1) j∈δB(i，l) (d j -1) (1), in, d i ,d j denote the degrees of nodes i and j respectively; δB(i,l) is the boundary of the ball consisting of nodes whose shortest path to node i is of length l, where l = 1; Then substitute into formula (3) to calculate the overall comprehensive impact index IVI value of each node, Among them, DC i , NC i , CR i , B.C. i and CI i They are the scope-normalized degree centrality, scope-normalized local H-index, scope-normalized domain connectivity, scope-normalized cluster rank, scope-normalized betweenness centrality, and scope-normalized collective influence of node i; Then substitute the IVI value into formula (4) to calculate the CI-IVI value of each node; CI-IVI i =(IVI i -1)∑ j∈δB(i，l) (IVI j -1) (4), Among them, IVI i and IVI j are the overall comprehensive impact index values ​​of nodes i and j, respectively. δB(i,l) is the sphere boundary composed of nodes whose shortest path to node i is of length l, where l = 1. Sort by the calculation results and remove the node with the largest CI-IVI value; then calculate λ according to formula (5): Where N is the total number of nodes, is the average degree of the network, CI-IVI i Calculated by formula (4); (4) If λ≤1, the identification is terminated and the removed node is the convergent node; if λ>1, proceed to step (5); (5) recalculating the CI-IVI value of each remaining node in the biomolecular network according to formulas (1), (3) and (4), sorting, removing the node with the largest CI-IVI value, and calculating λ according to formula (5); (6) If λ≤1, the identification is terminated and all removed nodes are the convergent node set; if λ>1, repeat step (5) until λ≤1.

2. The method according to claim 1, characterized in that The biomolecular network is selected from a gene expression regulation network, a protein interaction network, a biological metabolism network, an epigenetic network, a phenotypic network, a signal transduction network, etc.; more preferably, it is a gene expression regulation network.

3. The method according to claim 2, characterized in that The biomolecular network contains ≥3 nodes and ≥3 edges.

4. The method according to claim 2, characterized in that: The biomolecular network is a lung adenocarcinoma metastasis prognosis differential gene regulatory network.

5. The method according to claim 4, characterized in that The lung adenocarcinoma metastasis prognosis differential gene regulatory network contains 90 nodes and 157 edges.

Images