Method for constructing signaling network of pharmacological action mechanism and construction device using same

Abstract

The present invention relates to a method for constructing a signaling network of a pharmacological action mechanism, an information providing system, a computer program, and a construction device using same.

Description

Method for constructing a pharmacological action mechanism signal transduction network and a construction device using the same

[0001] The present invention relates to a method for constructing a pharmacological action mechanism signal transduction network; an information providing system; a computer program; and a construction device using the same.

[0002]

[0003] There are several studies that can elucidate the broad mechanisms of herbal medicine, including molecular docking analysis and drug response transcriptome analysis. However, while molecular docking studies can identify direct drug-target interactions, they cannot identify downstream drug actions. Drug response transcriptome analysis, on the other hand, can identify transcriptional outcomes resulting from drug treatment, but cannot identify upstream drug actions.

[0004] Despite active research to elucidate the mechanisms of action of herbal medicines, docking and transcriptomes can only identify the pharmacological mechanism system at a fixed specific space and time point, and cannot identify the actions of various proteins that occur in the middle of the signal transduction system of the pharmacological mechanism.

[0005] Using docking data alone can confirm the likelihood that a drug interacts with a specific receptor, but it cannot determine whether the drug actually works, or if so, what biological mechanisms it activates or deactivates. Furthermore, using transcriptome data alone can identify the target on which a drug is acting, but cannot determine which receptor it interacts with and how it acts on a specific mechanism.

[0006] Conversely, if the signaling pathway between docking and transcriptome data can be identified, the missing link can be identified, enabling the inferred pharmacological mechanism, enabling advanced drug development strategies. Furthermore, when multiple ingredients are administered together, the drug action and signaling pathways derived from docking data can be used to explore the pharmacological mechanisms of herbal medicine or combination therapies, as the process and outcome are derived from each ingredient.

[0007]

[0008] The present inventors completed the invention by modeling the protein signal transduction system starting from docking, and by constructing a pharmacological action mechanism signal transduction diffusion network derived from protein data interacting with single or complex components and drug-induced transcriptome data as input.

[0009]

[0010] The purpose of the present invention is to provide a method for constructing a pharmacological action mechanism signal transmission diffusion network.

[0011] Another object of the present invention is to provide an information providing system.

[0012] Another object of the present invention is to provide a computer program stored in a recording medium to execute the above method.

[0013] Another object of the present invention is to provide a device in which the computer program is stored.

[0014] Another object of the present invention is to provide a method for searching for intermediate proteins in a signal transmission diffusion network of a pharmacological action mechanism of a disease.

[0015]

[0016] By utilizing the pharmacological action mechanism signaling network of the present invention, the signaling system between docking and transcriptome data can be identified to determine the pharmacological mechanism, enabling the establishment of advanced drug development strategies. Furthermore, this network can be used to explore the pharmacological mechanisms of herbal medicine or combination therapies.

[0017]

[0018] Figure 1 illustrates a method for constructing a pharmacological action mechanism signaling network using large-scale molecular docking data and a transcriptome derived from Poria cocos.

[0019] FIG. 2 illustrates a pharmacological action mechanism signal transduction network using molecular docking data using a Poria terpene component and a Poria induced transcriptome according to one embodiment of the present invention.

[0020]

[0021] This is explained in detail as follows. Meanwhile, each description and embodiment disclosed in the present invention can also be applied to each other description and embodiment. In other words, all combinations of the various elements disclosed in the present invention fall within the scope of the present invention. Furthermore, the scope of the present invention should not be considered limited by the specific descriptions described below.

[0022] Furthermore, those skilled in the art will recognize or be able to ascertain, using no more than routine experimentation, numerous equivalents to the specific embodiments of the invention described herein. Furthermore, such equivalents are intended to be encompassed by the present invention.

[0023] The present invention has a technical feature in that it is a method capable of identifying all proteins present in the intermediate stages of a signal transduction system by constructing and / or modeling a protein signal transduction system through molecular docking analysis and drug response transcriptome analysis.

[0024]

[0025] One aspect of the present invention for achieving the above object is to provide a method for constructing a pharmacological action mechanism signal transduction network.

[0026] Specifically, the present invention may be a method for constructing a disease pharmacological action mechanism signal transmission diffusion network, comprising the steps of: collecting components to be used for molecular docking; collecting functional proteins by performing over-representation analysis (ORA) of the components; selecting proteins based on molecular docking among the collected functional proteins; selecting differentially expressed genes (DEGs) among proteins associated with a disease; constructing a protein set; and constructing a network layer based on the protein set by placing molecular docking-based proteins at a start node and DEGs at a final node.

[0027]

[0028] The step of collecting components to be used in the above molecular docking refers to components to be used in the molecular docking analysis method, and the term "molecular docking analysis" of the present invention refers to a method of predicting whether a specific component interacts with a specific protein by calculating a potential energy function and an intermolecular force field based on molecular energy calculated from the molecular structure of a specific component and a protein.

[0029] Specifically, the above molecular docking analysis may include a step of collecting components (compounds or proteins and proteomes, etc.) to be used for molecular docking into a database; a step of converting the components into a structure capable of molecular docking analysis; and a step of calculating molecular docking by considering all cases in which the compound (or protein) and the proteome form a pair.

[0030] The component to be used in the above molecular docking is a component in a three-dimensional form, and specifically, it may be a compound, a natural product, a complex component, or a single component, and the biological entity that interacts with the component may include a protein or any other component known to be usable for molecular docking without limitation.

[0031] There are no specific restrictions on the method for collecting components to be used for molecular docking, but they may be collected from the Pubchem database, ChEBI database, ZINC database, or manually curated. The file format collected from the above databases must be capable of depicting the three-dimensional structure of the components in a common chemical format (InChi, smiles, sdf, pdb) or a utility format (xml, dat).

[0032] The component file to be used for the above molecular docking must be converted into a molecular dynamics and docking format for molecular docking analysis. Specifically, in order to use Autodock Vina software, it can be converted into pdbqt form using OpenBabel software, but is not limited thereto.

[0033] The step of calculating molecular docking by considering all cases in which the above compound and proteome are paired can be calculated using, but is not limited to, Autodock Vina software or the SwissDock platform.

[0034] In a specific embodiment of the present invention, molecular docking analysis was performed based on components contained in Poria cocos. For example, the docking analysis was performed using python and autodock vina software, with the exhaustiveness parameter set to the maximum value of 100, the center coordinates set to [0, 0, 0], and the box size set to the maximum value of [126, 126, 126].

[0035]

[0036] The step of collecting functional proteins by performing over-representation analysis (ORA) of components to be used in the above molecular docking refers to the step of analyzing components to be used in molecular docking using the over-representation analysis method and collecting functional proteins that act on molecular docking.

[0037] The term "overrepresentation analysis" above refers to a method for calculating how statistically distributed specific genes are in a priori defined gene sets. For example, the EnrichR platform calculates a combined score and provides it to users. First, the probability that specific genes of interest are found more frequently than by chance in the a priori defined gene set is calculated using the Fisher exact test. Next, a gene dummy set equal to the number of genes of interest is randomly created, and the z-score of the gene set of interest is calculated using the Fisher exact test using the gene dummy set. Finally, the combined score is selected as the product of the logarithm of the p-value calculated from the Fisher exact test and the z-score. The platform used in the above overrepresentation analysis may utilize the EnrichR platform, and the a priori gene set may utilize the KEGG_2021_Human database, but is not limited thereto.

[0038] The term "functional protein" above means a protein that docks with a component to be used in molecular docking, for example, a compound to be used in molecular docking, and refers to a protein that interacts with a component to be used in molecular docking among a set of a priori proteins of a biological pathway that shows significant results in an "overrepresentation analysis."

[0039] The interaction scores between the components to be used in the above molecular docking and the proteins to be docked thereto can be checked using, but are not limited to, the STRING database.

[0040] At this time, the method for collecting functional proteins can be collected using the STIRNG database, Uniprot database, or GeneCards database, and any database known to contain functional proteins can be collected without limitation.

[0041] Additionally, even if it is not known as a functional protein, it can be used without restriction as long as it can be input together with transcriptome information.

[0042]

[0043] The step of selecting a molecular docking-based protein among the above-mentioned collected functional proteins means a step of selecting a molecular docking-based protein among numerous functional proteins to be docked with a component to be used for molecular docking, and the functional protein does not mean a molecular docking-based protein.

[0044] The above term, "molecular docking-based protein", refers to a protein whose appearance ratio is higher than a certain ratio among the functional proteins that dock with the components to be used for molecular docking, and the step of selecting a molecular docking-based protein among the collected functional proteins is performed based on the appearance ratio.

[0045] The term "appearance rate" above refers to the statistically significant number of times a protein involved in a specific disease pathway repeatedly appears. A larger percentage is better, and may vary depending on the input size. Specifically, the appearance rate may be, but is not limited to, 20% to 100%, 30% to 100%, or 50% to 100%.

[0046] For example, in a docking analysis of 125 components contained in Poria cocos and the homologous proteins of a priori gene sets involved in the Alzheimer's biological pathway, 14 proteins that interacted with more than 25 components were selected as functional proteins.

[0047] The above molecular docking-based protein refers to a protein that is located at the very first position, or in other words, the starting point, starting node, input 1, or storage module x (x is 4), when constructing the diffusion network of the present invention.

[0048] We aim to build a diffusion network by connecting network layers according to a certain rule, starting from the above molecular docking-based protein as the starting node until reaching the final node, the transcriptome (DEGs).

[0049]

[0050] The step of selecting differentially expressed genes (DEGs) among the proteins associated with the above disease refers to the step of selecting proteins to be located at the last node, endpoint, input 2, or storage module y (y is 5 or more) of the diffusion network through drug response transcriptome analysis.

[0051] In the present invention, "drug response transcriptome analysis" refers to a method of confirming drug response by calculating the number of sequencing reads of mRNA expressed in a cell or animal after treating the cell or animal with a specific drug. Here, "transcriptome" is a compound word formed by combining "transcription," meaning "transcription," and the suffix "-ome," meaning "whole," and refers to the entirety of transcription products or the sum total of intracellular transcription products in a specific developmental stage or physiological environment. A characteristic of the pharmacological mechanism signal transduction model constructed in the present invention is that the information input to the final layer is the transcriptome.

[0052] The above transcripts are not particularly limited, but can be collected from the GEO database. Disease information is not particularly required for transcript analysis, but information such as disease and pathway can be added when checking only under specific conditions.

[0053]

[0054] The "differentially expressed genes (DEGs)" of the present invention refer to a candidate group of genes whose expression values ​​are measured and statistically processed and whose expression is significantly different between a control group and a comparison group. Specifically, the candidate group of differentially expressed genes may have an adjusted p-value less than 0.05 and an expression value of log2(0.5) to -log2(0.5), but is not limited thereto.

[0055]

[0056] The step of constructing the above protein set refers to a protein set or protein group for selecting proteins located at an intermediate step between the start node and the last node, that is, an intermediate step in the process of connecting network layers from the start node to the last node, after setting the starting node as a molecular docking-based protein and the last node as transcripts (DEGs).

[0057] Here, the protein set is not particularly limited, but can be constructed specifically through a string database, and here, it can be, but is not limited to, a human protein set.

[0058]

[0059] The final step is to place molecular docking-based proteins in the starting node and DEGs (transcripts) in the final node, and to build a network layer (lay) based on the protein set. After placing molecular docking-based proteins in the starting node and transcripts in the final node, based on the previously built protein set, starting from the starting node, proteins with molecular docking-based proteins and protein-protein interaction scores higher than a certain score are selected from the protein set, and these nodes are connected to the starting node and a layer, and then sequentially selecting proteins with a high protein-protein interaction score with the secondarily selected proteins, excluding proteins excluded from the previous node or proteins that have already appeared more than once, and connecting them to the previous node and a layer until reaching the transcripts, thereby building a network.

[0060] Specifically, the network layer construction may include: (a) a step of selecting second layer proteins from proteins included in the first layer using a protein-protein interaction score (PPI combined score); (b) a step of connecting network layers by connecting the first layer and second layer interactions selected in step (a); (c) a step of selecting third and fourth layer proteins in the same manner as step (a), excluding proteins that appear more than once, and connecting network layers in the same manner as step (b); (d) a step of connecting interactions between fourth and fifth layer proteins using PPI scores as edges; (e) a step of deleting proteins (nodes) that are not interconnected in the fourth and fifth layers; and (f) a step of deleting proteins that are not connected in the third layer, the second layer, and the first layer.

[0061] The term "protein-protein interaction score" is used interchangeably with the mutual score in the present invention, and refers to a direct or indirect contact with high specificity established between two or more protein molecules as a result of a biochemical event mediated by an interaction including distance in the genome, gene fusion, co-occurrence in the genome, co-expression, experimental evidence, association in a selected database, and co-mention in a paper abstract. It refers to a physical contact with high specificity established between two or more protein molecules as a result of a biochemical event mediated by an interaction including electrostatic forces, hydrogen bonding, and hydrophobic effects. It is a score of protein-protein interactions and can be confirmed through the STRING database, but is not limited thereto.

[0062] The protein-protein interaction scores of steps (a) and (d) above may be, but are not limited to, 400 to 990 points provided as a basic score in the STRING database, specifically 900 to 990 points provided as the highest confidence score in the database.

[0063] The above steps (c) to (f) can be repeated multiple times, and there is no particular limitation on the number of times, which can mean repeating the process until the last node where the transcript is located.

[0064] In the present invention, the network is constructed by selecting the next layer protein based on protein-protein interaction scores, connecting the layer selected in the previous step with the current layer interaction to construct the network, and repeating the steps of excluding proteins that appear more than once. The number of layers can be customized as desired by the user. Specifically, the network can be constructed by continuously repeating steps (a) through (c).

[0065] Alternatively, a network can be constructed by repeatedly selecting the next protein based on protein-protein interaction scores, excluding proteins that appear more than once, connecting the interactions to form edges, and deleting unconnected proteins. Specifically, the number of layers can be expanded by continuously repeating steps (c) through (f).

[0066] In the present invention, the repetition may be performed 5 or more times, 10 or more times, 15 or more times, 20 or more times, 25 or more times, 30 or more times, 40 or more times, 50 or more times, 70 or more times, 100 or more times, 200 or more times, 300 or more times, 500 or more times, or 1000 or more times, but is not limited thereto.

[0067]

[0068] The above method may additionally include a step of statistical analysis using a permutation test to remove selection bias.

[0069] The above term, "selection bias," refers to a biased result, and it refers to removing selection bias through statistical analysis to ensure reliability in building a diffusion network.

[0070] Specifically, a random gene set was created by randomly extracting some, specifically 50 to 100, from the selected druggable proteome, and then a permutation list was constructed by grouping the components to be used for molecular docking into sets, and then the permutation list was subjected to overrepresentation analysis in the same way as the proteins derived from docking, and then the appearance rate of each valid pathway derived from docking was counted, and the same process was repeated for the permutation set to extract a valid random pathway (e.g., 125 random gene sets * 1000 permutations = 125,000 ORA scores). After that, the appearance rate of the valid pathway before selection bias was compared with the appearance rate of the permutation set, and the pathway whose appearance rate of the valid pathway was within the top 5% (p<0.05) was selected as the final statistically significant pathway.

[0071]

[0072] In one specific embodiment of the present invention, Poria cocos was used as an example to establish a pharmacological action mechanism signal transduction network and to confirm the operability of the signal transduction network, but the present invention is not limited to Poria cocos.

[0073] Specifically, by applying input data and output data of interest to the present invention, it is possible to identify a protein signaling pathway related to the protein.

[0074] In addition, in one embodiment of the present invention, the mechanism of action of Alzheimer's disease is analyzed as an example, but is not limited thereto, and if the overexpression analysis of the present invention is used, a disease related to input proteins and / or output proteins can be found.

[0075]

[0076] Another aspect of the present invention is to provide an information providing system.

[0077] Specifically, the present invention comprises: a storage module 1 for storing components to be used in molecular docking; a storage module 2 for storing functional proteins selected by overexpression analysis of components of the storage module 1; a storage module x for selecting and storing molecular docking-based proteins based on appearance frequency among functional proteins of the storage module 2; a storage module y for selecting and storing differentially expressed genes among proteins associated with a disease; a storage module 3 for storing a protein set; a storage module x+1 for selecting and storing proteins in the storage module x having a protein-protein interaction score of 400 to 990 points with the proteins of the storage module x; a construction module for connecting network layers by connecting interactions between proteins of the storage module x and proteins of the storage module x+1; An information providing system comprising: a repetition module 1 that builds a network layer as an edge between storage modules sequentially located a certain number of times or more while excluding proteins that appear more than once from a storage module 3 that stores a protein set, and storing the selected proteins in a storage module that increases the storage module x and the storage module x+1 by one, until reaching storage module y; a removal module that removes proteins that are not network-connected with storage module y in the relationship between storage module y and storage module y-1 starting from storage module y; and a repetition module 2 that repeats the removal module while decreasing the storage module by one until all proteins that are not network-connected up to storage module x are removed, excluding proteins that are network-connected, wherein x is 4 and y is 5 or more.

[0078] The above terms, “molecular docking analysis”, “components to be used for molecular docking”, “overrepresentation analysis”, “appearance frequency”, “differentially expressed genes”, “protein set”, “transcriptome”, “protein-protein interaction”, “network layer”, “protein removal” and “iteration” are as described above.

[0079] In the above storage module x, x is a module that stores a protein selected from among 4 functional proteins based on molecular docking, and means the first starting node in constructing the diffusion network of the present invention.

[0080] The above storage module x+1 means storage modules that increase one by one sequentially starting from storage module x. For example, it can mean that storage module x starts from 4, the next storage module is storage module 5, the next storage module is storage module 6, the next storage module is storage module 7, and so on, sequentially increasing.

[0081] The above storage module x+1 does not increase continuously and sequentially, but stops increasing when the number becomes the same as that of the storage module y in which the transcript DEGs are stored.

[0082] The above storage module y is a storage module that selects and stores transcripts, which are differentially expressed genes among proteins associated with a disease. Specifically, it may be a number that increases by one from a number that is one greater than the storage module x, and more specifically, y may be a natural number greater than or equal to 5.

[0083] The above repetition module 1 may be configured to build a network by connecting proteins of storage module x+1, in which proteins having a protein-protein interaction score between 400 and 990 points are stored, without selecting proteins that appear more than once among the proteins stored in storage module 3, starting from the protein of storage module x, and the protein of storage module y as the last node, and repeating while increasing the number of storage modules by one until the layer is connected to the protein of storage module y (x=4, y>=5).

[0084] For example, if x is 4 and y is 5, a molecular docking-based protein is stored in storage module 4 with storage module 4 as the starting node, and a differentially expressed gene is stored in storage module 5, so that an information provision system can be provided that connects a network with storage module 5, which is the final node, as an edge with storage module 4 as the starting node.

[0085] For example, if x is 4 and y is 6, a molecular docking-based protein is stored in storage module 4 with storage module 4 as the starting node, and a differentially expressed gene is stored in storage module 6, and a network is first connected by edges between storage modules 4 and 5 based on the protein-protein interaction score with storage module 4 as the starting node, and then a network is also connected by edges between storage modules 5 and 6, and since x+1 is equal to 6, which is the y value, that is, the last node, step 1 is terminated, and conversely, an information providing system that connects a network can be provided by sequentially removing proteins that are not connected to the network by edges starting from storage module 6 until reaching storage module 4, which is the starting node.

[0086] In addition, the removal module may remove proteins in storage modules that are not connected to the network by decreasing the number of storage modules one by one, starting with storage module y as the last node, and repeating this until reaching storage module x, which is the starting node (x=4, y>=5).

[0087]

[0088] Another aspect of the present invention is to provide a computer program stored in a recording medium to execute the above method.

[0089] In addition, another aspect of the present invention provides a device in which the computer program is stored.

[0090] The method according to one embodiment of the present invention described above may be implemented in the form of a computer program instruction that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., alone or in combination. The program instructions recorded on the medium may be those specially designed and configured for the present invention or may be those known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute program instructions such as ROMs, RAMs, and flash memories. Examples of program instructions include not only machine language codes generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter, etc. The above hardware device may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

[0091]

[0092] Hereinafter, the present invention will be described in more detail through examples and experimental examples. However, these examples and experimental examples are intended to exemplify the present invention, and the scope of the present invention is not limited to these examples and experimental examples.

[0093]

[0094] In order to construct a pharmacological action mechanism signal transduction network of the present invention and to confirm the operability of the signal transduction network, as an example, it was confirmed whether a pharmacological action mechanism signal transduction diffusion network was constructed based on Poria cocos.

[0095]

[0096] Experimental Example 1. Molecular Docking Analysis

[0097] First, we collected the components contained in Poria cocos through various known literature and databases and performed molecular docking analysis on terpene components.

[0098] Specifically, the information on the ingredients contained in Poria cocos was collected through [Chemical Constituents and Pharmacological Properties of Poria cocos Mycology, cultivation, traditional uses, phytochemistry and pharmacology of Wolfiporia cocos (Schwein.) Ryvarden et Gilb.: A review Phytochemistry and Biological Activities of Poria Phytochemistry and Pharmacological Activities of Wolfiporia cocos (FA Wolf) Ryvarden & Gilb], the TCMSP database, and the TM-MC database. Then, 125 ingredients in Poria cocos, for which the 3D molecular structure of the collected ingredients could be confirmed in the PubChem database, were selected as compounds with the potential to act as drugs (druggable compounds or compounds with druggability), i.e., candidate compounds. The above terms may be used interchangeably.

[0099] Next, we obtained human protein structure information from AlphaFold (AF) and extracted 869 proteins included in the proteome list that could be targets of HPA drugs. Then, we selected a proteome (druggable proteome) that could be targets of candidate drugs for molecular docking.

[0100] The compounds with potential drug interactions and the proteomes that could serve as drug targets were each converted to pdbqt format, a structure suitable for molecular docking analysis, using OpenBabel software. Then, considering all possible pairings between the selected compounds and the proteomes, the binding affinity scores for molecular docking were calculated using Autodock vina software (125 compounds * 869 proteomes = 108,625 pairs). By calculating the binding affinity score of the above molecular docking, the binding possibility of a component and a specific protein can be numerically predicted. Although there is no specific number, using the p-value calculation method, scores belonging to the top 5% (-9.328 kcal / mol) and top 10% (-8.919 kcal / mol) can be viewed as significant scores with a high binding possibility. At this time, the molecular docking analysis was performed using Python and autodock vina software with the exhaustiveness parameter set to the maximum value of 100, the center coordinates to [0, 0, 0], and the box size to the maximum value of [126, 126, 126]. Under these conditions, the results are most reproducible, and the widest range can be detected and calculated.

[0101]

[0102] Experimental Example 2. Drug-induced transcriptome analysis of Poria extract

[0103] In the above Experimental Example 1, proteins that were analyzed to interact with 125 components of Poria cocos through docking analysis were subjected to over-representation analysis (ORA) using the EnrichR platform and the 'KEGG_2021_Human' gene set provided by Python's gseapy library, and the results with the top 10 interaction scores (combined scores) were selected as the ORA results for the specific components of Poria cocos.

[0104] Meanwhile, to eliminate selection bias, we additionally performed relative statistical analysis using the permutation test. Specifically, we created a random gene set by randomly extracting 100 proteomes that could be targets of 869 candidate drugs. Next, 125 random gene sets were grouped to create a single 1,000-permutation list. The permutation list was subjected to ORA in the same way as the proteins derived from docking. The number of appearances of each valid pathway derived from docking was counted, and the same process was repeated for 1,000 permutation sets to extract valid random pathways (125 random gene sets * 1,000 permutations = 125,000 ORA scores). Afterwards, the appearance rate of the valid path was compared with the appearance rate of 1,000 permutation sets, and 54 paths with an appearance rate of the valid path within the top 5% (p<0.05) were selected as the final statistically significant paths.

[0105]

[0106] Example 1. Construction of a pharmacological action mechanism signaling network using docking analysis data and transcriptomes.

[0107] In order to construct an interaction network that connects upstream and downstream data using two input data: proteins (input 1) that interact with a single drug or each compound component, differentially expressed genes (DEGs, input 2) derived from transcriptome data, and output data, the network was used to model the diffusion process starting from a specific protein, and the detailed mechanism pathway was explored using this.

[0108]

[0109] First, by connecting the molecular docking analysis results of Experimental Examples 1 and 2 with the transcriptome analysis results, a pharmacological action mechanism signaling network (hereinafter referred to as a "diffusion network") was constructed to predict disease action pathways and explain the expected action pathways. The diffusion network selected the Kegg Alzheimer's disease pathway, which was analyzed as a significant pathway in both the molecular docking analysis and transcriptome analysis results, as an example.

[0110] Next, using the results of molecular docking analysis and the results of differentially expressed genes (DEGs) of the transcriptome as the starting and ending points, a new layer was created using proteins that exceeded a certain score based on the PPI interaction score provided by the string database. Specifically, the network consisted of a total of five layers, and the first and fifth layers were composed of proteins and DEGs based on the molecular docking analysis results, respectively. In the protein set excluding the proteins included in the first and fifth layers, those with a protein-protein interaction (PPI) combined score of 950 or higher with the proteins included in the first layer were selected as the proteins in the second layer. Next, a network was constructed by connecting interactions with a PPI score of 950 or higher in the first and second layers, and the third and fourth layers were constructed in the same way as the second layer, but all proteins that had already appeared in the network were excluded from the layer construction. The basic form of the diffusion network was completed by connecting the interactions with PPI scores of 950 or higher between proteins belonging to each protein set in the fourth and fifth layers as edges (Fig. 1A).

[0111] Next, the network skeleton was modified using backpropagation. After that, all proteins (nodes) with at least one interaction connection with the fifth layer were deleted from the fourth layer. Similarly, proteins with no connection to the third layer, the second layer, and the first layer were deleted in that order (e.g., proteins with no connection to the fourth node in the third layer). The network was constructed using Python, and the network was visualized using cytoscape software (Fig. 1B).

[0112]

[0113] Example 2. Establishment of a diffusion network of pharmacological action mechanisms using terpene components of Poria cocos.

[0114] In order to identify the Alzheimer's disease pathological action point in the diffusion network constructed in Example 1, the diffusion network was newly reconstructed using the kegg Alzheimer's disease pathway gene set. At this time, the molecular docking-based proteins used in the network were collected from all proteins predicted to act on the Alzheimer's disease pathway through ORA analysis among the 125 components used for docking, and 14 proteins with an appearance rate of 20% or higher were selected as molecular docking-based proteins (n≥5). The DEGs used in the network used the results of the intermediate concentration that showed a significant effect on the Alzheimer's disease pathway. From the drug transcriptome results of the intermediate concentration, 22 proteins with an adjusted p-value <0.05 and an expression value greater than log2(0.5) or less than -log2(0.5) were selected as DEGs. The protein set for layer construction was constructed by downloading human protein information from the string database.

[0115] First, proteins belonging to the Alzheimer's disease pathway gene set were color-mapped onto the Alzheimer's diffusion network to determine how closely the two pathways overlapped (Figure 2A). This allowed us to identify protein signaling pathways associated with Alzheimer's disease, as well as a select group of candidates likely to be associated with Alzheimer's disease but not yet included in a specific gene set.

[0116] Next, the network was reconstructed using only the proteins included in the Kegg Alzheimer's Gene Set (KAGS) (color-coded proteins above) in the Alzheimer's Diffusion Network (ADN). Specifically, all nodes other than those included in KAGS were removed from the ADN, and nodes with no edges were also removed. Next, the second layer was reorganized into five groups based on connections with the same proteins in the first layer, and the third layer was also divided into five groups based on connections with the groups in the second layer. The groups in the third layer were identified as action points in the Kegg pathway, and groups that act on sub-modules in the pathway were connected by adding sub-module nodes. The reconstructed network was then reorganized for each layer. All visualization work was performed in cytoscape software (Fig. 2B).

[0117] With the detailed network construction as described above, MAP2K2, which activates the AGE-RAGE signaling pathway among the detailed mechanisms of Alzheimer's disease, could be analyzed as a protein activated via PIK3CD-HRAS, and through this, the protein signaling pathway could be confirmed using the known gene set, and the detailed mechanism in complex diseases could be identified.

[0118]

[0119] From the above description, those skilled in the art will understand that the present invention can be implemented in other specific forms without altering its technical spirit or essential characteristics. In this regard, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. The scope of the present invention should be interpreted as encompassing all changes or modifications derived from the meaning and scope of the following claims and their equivalent concepts, rather than the detailed description above.

Claims

1. A method for constructing a disease pharmacological action mechanism signal transmission diffusion network comprising the following steps: A step of collecting components to be used for molecular docking; A step of collecting functional proteins by performing over-representation analysis (ORA) of the above components; A step of selecting a protein based on molecular docking among the collected functional proteins; A step of selecting differentially expressed genes (DEGs) among proteins associated with diseases; Step of building a protein set; and A step of constructing a network layer based on the protein set by placing molecular docking-based proteins at the starting node and DEGs at the last node.

2. A method according to claim 1, wherein the components to be used for molecular docking are collected from the Pubchem database, the ChEBI database, the ZINC database, or manually curated.

3. In paragraph 1, The above overrepresentation analysis is performed using the EnrichR platform and the KEGG_2021_Human gene set.

4. In paragraph 1, A method wherein the above-mentioned functional protein is a protein having a protein-protein interaction score (combined score) of 400 to 990 points.

5. In paragraph 1, A method in which a step of selecting a molecular docking-based protein among the collected functional proteins is performed based on an appearance ratio, and a protein is selected as a molecular docking-based protein when the appearance ratio is 20% to 100% or less.

6. In paragraph 1, A method wherein the component to be used in the above molecular docking is a compound, protein, natural product, complex component, or single component.

7. In paragraph 1, A method wherein the above differentially expressed genes are selected when the adjusted p-value is less than 0.05 and the expression value is log2(0.5) to -log2(0.5).

8. In the first paragraph, the molecular docking, A step of collecting compounds or proteins and proteomes as components to be used for molecular docking with a database; A step of converting the above components into a structure capable of molecular docking analysis; and A method comprising: calculating molecular docking by considering all cases in which the compound or protein and the proteome are paired; 9. In the first paragraph, the network layer construction is (a) a step of selecting second layer proteins based on protein-protein interaction score (PPI combined score) from proteins included in the first layer; (b) a step of connecting network layers by connecting the first layer and second layer interactions selected in step (a); (c) a step of selecting third and fourth layer proteins in the same manner as step (a), excluding proteins that appear more than once, and connecting the network layers in the same manner as step (b); (d) a step of connecting the interactions as edges with PPI scores between the fourth and fifth layer proteins; (e) a step of deleting proteins (nodes) that are interconnected by edges in the fourth and fifth layers; and (f) A method for constructing a pharmacological action mechanism signal transmission diffusion network, comprising the steps of: deleting proteins without edge connections in the third layer, the second layer, and the first layer, respectively; 10. In paragraph 9, A method wherein the protein-protein interaction scores of steps (a) and (d) are between 400 and 990 points.

11. In paragraph 9, A method wherein steps (c) to (f) are additionally repeated several times.

12. In paragraph 1, A method, wherein the method further includes a step of statistical analysis using a permutation test method to remove selection bias.

13. Storage module 1 for storing components to be used for molecular docking; Storage module 2 that stores the selected functional protein by analyzing the components of the above storage module 1; Storage module x that selects and stores a protein based on molecular docking among the functional proteins of the above storage module 2; A storage module y that selects and stores differentially expressed genes among proteins associated with a disease; Storage module 3 for storing protein sets; Storage module x+1 that selects and stores proteins of the storage module x and proteins having a protein-protein interaction score of 400 to 990 points in storage module 3; A construction module that connects network layers by connecting the interactions between proteins in storage module x and proteins in storage module x+1; Proteins that appear more than once are excluded from storage module 3 that stores the protein set, and a repetition module 1 that stores the selected proteins in storage modules that increase the storage module x and storage module x+1 by one, and builds a network layer as an edge between storage modules that are sequentially located a certain number of times or more until reaching storage module y; A removal module that starts from storage module y and removes proteins that are not network connected to storage module y in relation to storage module y-1; and An information providing system comprising a repetition module 2 that repeats the removal module while decreasing the storage modules one by one until all non-network-connected proteins up to storage module x are removed, excluding network-connected proteins, An information providing system wherein the above x is 4 and y is 5 or greater.

14. In paragraph 13, The above repetition module 1 does not select proteins that appear more than once among the proteins stored in storage module 3, and builds a network by connecting proteins of storage module x+1, in which proteins having a protein-protein interaction score between 400 and 990 points are stored, starting from proteins of storage module x. The protein of storage module x is used as the starting node and the protein of storage module y is used as the last node, and the number of storage modules is increased by one and repeated until the layer is connected to the protein of storage module y. Information provision system.

15. In paragraph 13, The above removal module removes proteins in the storage modules that are not connected to the network by decreasing the number of storage modules one by one, with the storage module y as the last node. Repeat until reaching the storage module x, which is the starting node. Information provision system.

16. A computer program stored in a computer-readable recording medium to execute a method according to any one of claims 1 to 12.

17. A device storing the computer program of Article 16.

18. A method for searching intermediate proteins in a disease pharmacological action mechanism signal transmission diffusion network comprising the following steps: A step of collecting components to be used for molecular docking; A step of collecting functional proteins by performing over-representation analysis (ORA) of the above components; A step of selecting a protein based on molecular docking among the collected functional proteins; A step of selecting differentially expressed genes (DEGs) among proteins associated with diseases; Step of building a protein set; and A step of constructing a network layer based on the protein set by placing molecular docking-based proteins at the starting node and DEGs at the last node.

Images