KR101963331B1 - Method and system for predicting drug repositioning candidate based on similarity between drug and metabolite - Google Patents

Abstract

A method for predicting drug re-creation candidates and a system thereof are provided. The method includes the steps of: collecting disease-related protein information associated with disease occurrence, information of interaction of a protein with a compound, and metabolite information of a human, Related protein and a human metabolite based on the information, matching the human metabolite having a chemical structure similar to that of each drug for each known drug according to the similarity of the chemical structure, Substituting the disease-associated protein based on the interaction-related information, and extracting the drug and the disease-associated protein whose similarity of the chemical structure satisfies the critical condition, A candidate to regenerate drug-related diseases of extracted proteins And a step of.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and a system for regenerating a drug candidate based on similarity with a metabolite of a human body,

The present invention relates to a method for predicting drug regeneration candidates based on similarity with a human metabolite and a system thereof.

Recently, drug re-creation has emerged as a new paradigm due to the high failure rate in the drug development process. Drug regeneration is a method of developing drugs as drugs that demonstrate new efficacy against drugs that are already on the market or fail to commercialize for reasons other than stability at the clinical stage.

In order to predict candidates for regeneration of drugs, methodologies such as structural similarity with existing drugs and prediction of target binding structure have been developed and contribute to the drug development process.

However, in the methodologies already developed, a space of human innate metabolites has not been considered relatively.

If the structure of the drug is similar to the human metabolite, the human metabolite space can be a great resource for drug development, because it can interact in a similar way to the same target. For example, morphine mimics the endorphin of the human opioid system and has been shown to have similar pharmacological and physiological effects. In this way, although the similarity of human metabolism in drug development is an important characteristic, there is no methodology for predicting drug candidates in consideration of this characteristic, and the search for metabolites of human metabolites is limited.

 Currently, the concept of metabolite-likeness has been proposed as one of the efforts to utilize the properties of human metabolites in drug development.

Metabolite similarity has been proposed as a new drugability filter in the case of drugs with a structure similar to that of human metabolites, in that there is a possibility of using a human body transport system used by drug-like metabolites together.

However, most human metabolites interact with innate targets, including metabolic enzymes, and although new disease-associated target-drug relationships can be inferred directly from the drug's in vivo target-related metabolite similarity And a novel drug prediction methodology using this property has not been proposed.

In this way, there have been many methods of regenerating drugs, but the methodologies already developed have limitations in that it is difficult to regenerate drugs without existing drug information interacting with disease-related target proteins.

The object of the present invention is to deduce the interaction relationship between a disease-related bio-target protein and a drug based on the chemical structure similarity between a disease-related human metabolite and a known drug, A method for predicting a drug candidate, and a system thereof.

According to one aspect of the present invention, a method for predicting drug regeneration candidates is a method for predicting drug regeneration candidates by a system that is operated by at least one processor, wherein disease-related protein information associated with disease occurrence, A step of collecting information on action and metabolism of the human body, a step of extracting information on the interaction relation between the disease-related protein and the human metabolite based on the collected information, The method comprising the steps of: matching a human metabolite having a similar structure to the disease-related protein on the basis of the interaction-related information; And extracting the disease-associated protein, And predicting the extracted protein as a drug regeneration candidate of the related disease.

Wherein the matching step generates a similarity matrix of the drug and the human metabolite based on a similarity score between the chemical fingerprint of each of the structures of the drug and the structure of the human metabolite, The scoring matrix may be generated by replacing the human metabolite with the disease-associated protein and descending.

The similarity degree matrices may be hierarchically clustered using a heat map function, and each similarity score may be displayed in different colors for each section.

Between the replacing step and the predicting step, the drug target protein information collected from the public database, and the interaction relation information between the disease-associated protein and the human metabolite, Generating a receiver operating characteristic curve based on the chemical structure similarity scores of the human metabolite and the drug included in the reference data set, Further comprising the step of extracting a similarity score having the highest Yoden index (Youden's Index) calculated using a receiver operation specific curve, and selecting the similarity score as the reference similarity score, wherein said predicting step includes: And a similarity degree higher than the reference similarity score A drug having a re-creation can be estimated as a candidate.

Wherein the step of selecting the reference data set further comprises the steps of: after the step of selecting the reference data set, a recipient manipulation characteristic curve generated based on a score of the chemical structure similarity of the human metabolite and the drug contained in the reference data set, And comparing the recipient operation characteristic curves generated based on the predicted values calculated using the predictive algorithm of the first embodiment with the predictive accuracy of the reference data set.

Wherein the extracting comprises:

Related information of the disease-related protein and the information on the interaction relation of the metabolite of the human body, and extracts the interaction relationship information determined to be statistically significant.

Wherein the step of extracting comprises the step of determining whether a significance value (P value) according to the significance evaluation satisfies a threshold condition among the disease-related protein information matched to at least two or more and the interaction relation information of the human metabolism information, The interaction relationship information having a priority according to probability is given as a highest priority, and the extracted interaction relationship information can be mapped to the disease-related protein and the human metabolite one to one.

According to another aspect of the present invention, a drug regeneration candidate prediction method is a method of predicting a drug regeneration candidate by a system that is operated by at least one processor. The drug regeneration candidate prediction method predicts drug regeneration candidates by analyzing similarity of chemical structures of known drugs and human metabolites A step of replacing the human metabolite with a disease-associated protein that interacts with the metabolism of the human metabolite, and determining whether the similarity of the chemical structure meets the critical condition And a disease-associated protein matched to the drug, and predicting the extracted drug as a drug regeneration candidate of the extracted disease-associated protein.

The similarity of the chemical structure can be analyzed based on the degree of similarity between the chemical fingerprint of the chemical structure of each of the drug and the human metabolite.

Wherein the predicting step predicting the corresponding drug as a regeneration candidate when the similarity score calculated through the chemical structure similarity analysis between the drug and the disease protein matched with the drug is higher than the reference similarity score, The score is selected using the Receiver Operating Characteristic curve and the Youden's Index based on the chemical structure similarity score of the reference data set including the drug target protein, the metabolites of the metabolites and the relationship between the drugs .

The substituting step may use the interaction relation information in which the priority of the significance value (P value) satisfying the threshold condition is the highest among the interaction relation information of at least two disease-related proteins matched with each other have.

According to another aspect of the present invention, a drug regeneration candidate prediction system is operated by at least one processor and is a system for predicting drug regeneration candidates, comprising: a disease-associated protein Related metabolite information extracting unit for extracting information on the interaction relationship between the disease-related protein and the human metabolite based on information, information of interaction between the protein and the compound, and human metabolite information, A similarity generating unit for matching a human metabolite having a chemical structure similar to that of each drug and replacing the human metabolic material matched for each drug with the disease related protein based on the interaction relationship information, Drug and disease associations that meet critical conditions And a drug regeneration candidate predicting unit for extracting the protein and for predicting the extracted drug as a drug regeneration candidate of the related disease of the extracted protein.

The disease-related metabolite information extracting unit performs a significance evaluation based on the number of interaction relations with the disease-related protein information and the interaction relation information of the human metabolite information, and determines the priority of the significance value (P value) Wherein the degree of similarity between the drug and the disease-related protein is calculated by calculating the degree of similarity between the chemical fingerprint of the structure of the drug and the metabolite of the human metabolism, Wherein the predictor uses the scoring matrix to predict the drug and disease associated protein having the similarity score higher than the reference similarity score to the drug regeneration candidate and outputs the predicted information to the screen can do.

The drug re-creation candidate predicting system is a method for predicting drug re-creation candidates based on drug target protein information collected from a public database and a chemical structure similarity score between a drug target protein, a human metabolite, and a drug contained in a reference data set selected using the interaction- A Receiver Operating Characteristic curve is generated based on the recipient operating characteristic curve, a Youden's Index according to the receiver operation characteristic curve is calculated, and a similarity score having the highest Yoden index is selected as the reference similarity score, And may further include a reference similarity score selecting unit provided to the predicting unit.

The drug regeneration candidate prediction system includes a drug regeneration candidate database storing drug regeneration candidate information including the relationship between the drug regeneration candidate prediction unit predicted by the regeneration candidate drug and the disease related protein, And a user interface unit for managing access to the drug regeneration candidate database to a terminal connected thereto via a communication network.

According to the embodiment of the present invention, by predicting the interaction with the disease-related target using the metabolism similarity of the human body with the clinical-approved drug already having data on the toxicity and adverse effects on the human body, can do.

In addition, even if there is no known drug information that interacts with a particular disease-associated target protein, it is possible to predict drug regeneration candidates among existing drugs for a particular target protein based on the similarity with the human metabolic relationship that interacts with disease- . Therefore, it is expected to increase the drug regeneration space so as to regenerate the drug to the diseases which have conventionally been difficult to regenerate the drug.

In addition, it can be used as a method of predicting new drug candidates by expanding into lead space and chemical space, not limited to predicting drug candidates, by using similarity of metabolite in human body metabolism have.

In addition, it can be applied and used in predicting new side effects of drugs as well as regenerating drugs or predicting new drug candidates.

In addition, predicting drug candidates based on similarity in human metabolism increases the likelihood of predicting pharmacokinetic and pharmacodynamic superior drug candidates.

1 is a schematic block diagram of a drug regeneration candidate prediction system according to an embodiment of the present invention.
2 is a flowchart of a method for predicting drug regeneration candidates according to an embodiment of the present invention.
FIG. 3 is a diagram illustrating an example of a drug regeneration candidate prediction result according to an embodiment of the present invention.
4 is a schematic block diagram of a drug regeneration candidate prediction system according to another embodiment of the present invention.
FIG. 5 is a flowchart illustrating a process of extracting disease-related human metabolites according to an embodiment of the present invention.
6 is a flowchart illustrating a process of generating a scoring matrix according to an embodiment of the present invention.
FIG. 7 shows a heat map of a similarity matrix according to an embodiment of the present invention.
FIG. 8 is a flowchart illustrating a process of selecting a reference similarity score according to an embodiment of the present invention.
9 is an illustration of a set of reference data in accordance with one embodiment of the present invention.
FIG. 10 is a graph comparing recipient operating characteristic curves of an embodiment of the present invention and a conventional Swiss TargetPrediction (STP) algorithm.
11 is a graph comparing recipient operation characteristic curves of an embodiment of the present invention and a conventional TargetNet (TN) algorithm.
FIG. 12 is a diagram comparing recipient operation characteristic curves of an embodiment of the present invention and a site-directed docking program (Libdock).
FIG. 13 is a graph illustrating a recipient operating characteristic curve according to an embodiment of the present invention.
14 is a graph showing a Yodden index according to an embodiment of the present invention.
15 is a flowchart illustrating a drug regeneration candidate predicting process according to an embodiment of the present invention.
16 is a table showing a drug regeneration candidate related to a metabolic antagonistic substance according to an embodiment of the present invention.
Figure 17 shows a fragment of the metabolic pathway including enzymes and metabolites associated with Gossypollosis in accordance with one embodiment of the present invention.
FIG. 18 is a table showing a list of candidates supported by evidence through literature review according to an embodiment of the present invention.
19 is a block diagram illustrating a hardware configuration of a drug regeneration candidate prediction system according to another embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

Throughout the specification, when an element is referred to as "comprising ", it means that it can include other elements as well, without excluding other elements unless specifically stated otherwise.

Also, the terms " part, "" ... "," module ", and the like described in the specification mean a unit for processing at least one function or operation and may be implemented by hardware or software or a combination of hardware and software .

FIG. 1 is a schematic block diagram of a drug regeneration candidate prediction system according to an embodiment of the present invention, FIG. 2 is a flowchart of a drug regeneration candidate prediction method according to an embodiment of the present invention, FIG. FIG. 4 is a schematic block diagram of a drug regeneration candidate prediction system according to another embodiment of the present invention. FIG. 4 is a schematic block diagram of a drug regeneration candidate prediction system according to an embodiment of the present invention.

Referring to FIG. 1, a drug regeneration candidate prediction system 100 includes a disease-related metabolism information extracting unit 101, a similarity-generating unit 103, a reference similarity-degree selecting unit 105, And an operation unit 107. Operations of these structures are as shown in Fig.

Referring to FIG. 2, the disease-associated metabolite information extracting unit 101 extracts information on interaction between the disease-associated protein and the human metabolite (S101). The disease-associated metabolite information extracting unit 101 extracts disease-related metabolites based on the information collected from the disease protein database 200, the protein-compound interaction database 300, and the human metabolism database 400, And extracts information on interaction between metabolites of human body. The specific operation will be described later with reference to Fig.

The similarity degree generation unit 103 generates a similarity degree matrix according to the score of the chemical structure similarity between the approved drug and the disease-related metabolite already used in clinical use (S103). Based on the interaction relationship information extracted in step S101, the similarity generation unit 103 generates a scoring matrix between the drug and the disease-related protein through the disease-associated protein from the similarity matrix generated in step S103 (S105). That is, the disease-related human metabolite in the similarity matrix generated in step S103 is replaced with the disease-associated protein based on the interaction-related information extracted in step S101.

The drug regeneration candidate predicting unit 107 predicts drug regeneration candidates for each disease by applying the reference similarity score selected by the reference similarity degree selecting unit 105 to the scoring matrix generated in step S105 (S107).

Referring to FIG. 3, the target enzyme of the disease Mesothlioma is GART, the related metabolite is 10-formyltetrahydrofolate, and the top-similarity drug (Leucovorin) having the highest structural similarity score (0.97) . Thus, Leucovorin is presented as a drug candidate, and the original indication of this drug may be presented as a drug re-creation candidate for a new association, Osteosarcoma or Mesothlioma (New Indication) .

In addition, the target enzyme of the disease Leukemia is POLA / B, the related metabolite is dCTP, and the drug with the highest structural similarity score (0.81) is Decitabine. Therefore, Decitabine has been proposed as a drug regeneration candidate, and its associated disease can be suggested as a drug regeneration candidate for Myelodysplastic syndrome or a disease called Leukemia.

Thus, the drug regeneration candidate prediction system 100 deduces the interaction relationship between the disease-associated target protein and the drug based on the chemical structure similarity between the disease-related human metabolite and the existing drug, The candidate can be predicted.

Referring to FIG. 4, the drug regeneration candidate predicting unit 107 may be connected to the display device 700. The drug regeneration candidate predicting unit 107 can output the drug regeneration candidate predicted on the screen of the display device 700. [

Further, the drug regeneration candidate predicting unit 107 can store the drug regeneration candidate predicted for each disease in the database 109, and construct the drug regeneration candidate database 109. [ The medicament regeneration candidate prediction system 100 may further include a user interface unit 111.

The user interface unit 111 is connected to the terminal 900 connected through the communication network 800. The user interface unit 111 may be configured to provide access to the database 109 and to exchange various information by being connected to the terminal 900 through the communication network 800. For example, the user interface unit 111 may provide an environment in which the terminal 900 can search for a drug regeneration candidate for each disease. The communication network 800 may be a network such as a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), the Internet, 4G mobile communication network, Wi-Fi, WiBro, etc. The communication method does not depend on wired or wireless, and any communication method may be used. The terminal 900 includes a memory unit such as a smart phone, a personal computer (PC), a tablet PC, a personal digital assistant (PDA), and a web pad, .

Now, the detailed operation of the drug regeneration candidate prediction system 100 will be described.

FIG. 5 is a flowchart showing a process of extracting disease-related metabolites according to an embodiment of the present invention, and shows the operation of the disease-associated metabolite information extraction unit 101.

Referring to FIG. 5, the disease-associated metabolite information extracting unit 101 extracts interaction-related information between a disease-associated protein and a human metabolite. Here, the interaction relationship information refers to human metabolite information interacting with a disease-associated protein. For example, a large amount of uric acid accumulates in the body due to the metabolism of purines (adenosine, guanosine) in the body of Xanthine oxidase, which is one of the proteins, causing gout, and the relationship between Xanthine oxidase, It can be information about the interaction relationship between metabolites of human body.

First, the disease-associated metabolite information extracting unit 101 collects disease-related protein information from the disease protein database 200 (S201). Herein, the disease-related protein information refers to protein information that is responsible for or causes the onset of a specific disease. At this time, the extracting unit can use an open database such as 'DisGeNet'.

The disease-associated metabolite information extracting unit 101 collects protein-compound interaction information from the protein-compound interaction database 300 (S203). Here, the protein-compound interaction information refers to compound information that is reactive with a specific protein, and may include, for example, information on proteins and low molecular compounds involved in cell regeneration. At this time, the disease-related metabolite information extracting unit 101 can use a public database such as 'STITCH', 'KEGG', and 'BRENDA'.

The disease-associated metabolite information extracting unit 101 collects human metabolite information from the human metabolite database 400 (S205). Here, a metabolite refers to a chemical compound that participates in metabolism or is produced by it. Metabolic reactions are catalyzed by enzymes. Metabolism is generally the function of enzymatic functional groups (enzymes), and the chemical bonds of the substrate are changed one by one to become metabolites. At this time, the disease-related metabolite information extracting unit 101 can use a public database such as 'Recon2' and 'HMDB'.

At this time, the enzyme refers to a protein that does not change itself in various chemical reactions but speeds up the reaction. Substrate refers to a specific reactive molecule or group of molecules catalyzed by the enzyme.

The disease-associated metabolite information extracting unit 101 extracts disease-related protein-compound interaction information from the protein-compound interaction information using the disease-related protein information (S207).

The disease-associated metabolite information extracting unit 101 extracts disease-related interaction information related to the human metabolism from the disease-related protein-compound interaction information (S209) using human metabolite information (S209) The interaction relationship between the substances is mapped (S211).

At this time, if the disease-associated metabolite information extracting unit 101 simply maps the collected information, one human metabolite interaction information is not extracted into one disease-associated protein but the interaction relation with various human metabolites Is extracted. Since the present invention predicts a new disease-associated protein-drug interaction relationship through disease-associated human metabolites, a process of prioritizing the most relevant disease-related human metabolites is needed. To prioritize meaningful disease-related human metabolites, disease-related human metabolite relationships can be defined based on the number of interaction relationships with human metabolites and specific disease-associated proteins.

According to one embodiment, the disease-associated metabolite information extracting unit 101 performs a significance evaluation based on the number of interaction relations between the human metabolite and specific disease-associated proteins (S213). The disease-associated metabolite information extracting unit 101 can generate the frequency table using the number of interaction relations between the human metabolite and specific disease-associated proteins. The frequency table can be implemented as 2 × 2, with the frequency of the number of relationships between a specific human metabolite and disease-associated proteins that interact with the substance. For example, suppose that the number of proteins associated with diseases is N, the number of human metabolites is M, and a particular human metabolite B interacts with a disease-associated protein a. Let's assume that a particular metabolite B interacts with a protein b that is associated with a disease other than disease A. In addition, when the number of the metabolites M-1 except for a specific metabolite B interacts with the disease A-related protein is c and the number of the protein interacting with the disease other than the disease A is d, Frequency tables are generated as shown in Table 1.

Disease-associated protein Disease-associated protein other than disease A system Certain human metabolites B a b a + b Other human metabolites c d c + d system a + c b + d a + b + c + d

The disease-associated metabolite information extracting unit 101 evaluates whether the interaction relation between a specific human metabolite and specific disease-related proteins is statistically significant, and the significance evaluation can utilize a fisher exact test . If the significance probability (p-value) calculated through the significance evaluation has a value less than the threshold value of 0.05, the interaction relation is evaluated as meaningful.

The disease-related metabolite information extracting unit 101 prioritizes the interaction relationship information evaluated as being important based on the significance probability, and extracts the interaction relation information having the highest priority as the disease-related human metabolite information S215). In this way, information on the interaction relationship between one human metabolite and one specific disease-related protein, that is, disease-related human metabolite information, is extracted.

FIG. 6 is a flowchart illustrating a process of generating a scoring matrix according to an embodiment of the present invention, and FIG. 7 illustrates a heat map of a similarity matrix according to an embodiment of the present invention. 6 and 7 illustrate the operation of the similarity generation unit 103. As shown in FIG.

First, referring to FIG. 6, the similarity generation unit 103 collects drug information from the disclosed drug database 500. At this time, the drug information may be 1,861 FDA (Food and Drug Administration) approved drugs.

The similarity generation unit 103 collects structural information files of each metabolite collected from all the collected drugs and the metabolite database 400 from the public database (S301). At this time, for example, 'DrugBank', 'Recon2', etc. may be used as the public database.

The similarity generation unit 103 may collect the structure information file of the drug from the 'DrugBank' and collect the structure information file of the human metabolism from the 'Recon2'. Here, the structure information file may be a file format representing a compound structure, for example, a file format such as sdf, smiles, and inchi.

According to one embodiment, the similarity generation unit 103 generates a chemical fingerprint of all structures using the sdf file of each of the drug and the human metabolite collected (S301) (S303). At this time, the similarity generation unit 103 can generate a MACCS key fingerprint type chemical fingerprint using the Python RDKit module.

The similarity generation unit 103 calculates a chemical structural similarity between a drug and a human metabolite using a chemical fingerprint (S305). According to one embodiment, the structural similarity score can be calculated using Tanimoto similarity.

The similarity level generator 103 calculates the similarity degree of each drug with respect to each metabolite of the human body to generate a structural similarity matrix (S307).

The similarity generation unit 103 performs the measurement clustering (S309) using the heat map function, the structure similarity matrix generated in step S307. Then, the structure similarity score is divided into a plurality of sections, and the colors assigned to the divided sections are varied to output in the form of a heat map as shown in FIG. 7 (S311).

At this time, the similarity level generator 103 can generate a structural similarity matrix having 1,861 FDA-approved drugs and 1,110 humanoid metabolism scores. The similarity degree generator 103 hierarchically clusters the generated structure similarity matrices using the heatmap.2 function of R, and divides the Tanimoto similarity score into 10 equal parts and displays them in different colors, thereby improving readability. Here, the heat map is a visualization method that displays information by slightly different colors according to values of a certain continuous variable on a graph represented by x-axis and y-axis or on a two-dimensional map.

The similarity-level generating unit 103 generates a disease-related protein-drug scoring matrix through the disease-associated human metabolism from the drug-metabolism similarity matrix generated in step S307 (S313). Here, as explained in FIG. 5, the disease-related human metabolite information includes information on interaction relationship (one-to-one mapping information) between one human metabolite and one specific disease-associated protein. Therefore, substitution of the human metabolite of the similarity matrix with the disease-related protein according to the interaction-related information and the descending order based on the structural similarity score can generate a scoring matrix between the disease-associated protein and the drug. At this time, the similarity generation unit 103 may provide a scoring matrix in the form of a heat map as shown in FIG.

FIG. 8 is a flowchart illustrating a process of selecting a reference similarity score according to an exemplary embodiment of the present invention. FIG. 9 is a view illustrating an example of a reference data set according to an embodiment of the present invention. FIG. 10 is a block diagram of a conventional SwissTargetPrediction (STP) algorithm. FIG. 11 is a graph comparing recipient operating characteristic curves of a conventional TargetNet (TN) algorithm according to an embodiment of the present invention. FIG. FIG. 13 is a graph showing recipient operating characteristic curves according to an embodiment of the present invention, and FIG. 14 is a graph illustrating a recipient operating characteristic curve of a site-directed docking program according to an embodiment of the present invention. Fig.

At this time, 8, 9, 10, 11, and 12 indicate the operation of the reference similarity degree selection unit 105.

First, referring to FIG. 8, the reference similarity degree selection unit 105 extracts drug target protein information from an open drug target protein database 600. The drug target protein-human metabolite-drug is extracted based on the disease-associated metabolite information and drug target protein information provided by the disease-associated metabolite information extracting unit 101. And a gold standard positive is selected from the drug target protein-human metabolite-drug (S401). At this time, the drug database 500 and the drug target protein database 600 are one open database, and a 'Drugbank' may be used.

At this time, the reference similarity degree selection unit 105 can select the drug target protein-human metabolite-drug whose structural similarity score satisfies the reference value as a reference data set. Here, the reference value may be set to 0.5 or more.

These reference data sets include the drug target protein-human metabolite, the drug, their structural similarity score and the significance (P-value), and can be prioritized and sorted according to the structural similarity score. At this time, the structural similarity score is provided from the disease-related protein-drug scoring matrix calculated in step S313 of FIG. And the significance probability is a value derived from the significance evaluation of the disease-related human metabolite calculated through the step S213 of FIG.

According to one embodiment of the present invention, a metabolic antimetabolite family of drugs may be used as the reference data set. Here, the drugs of the metabolic antagonistic substance series are drugs that act as a substrate analog of a specific enzyme to inhibit the activity of the enzyme, thereby exhibiting a drug effect. These metabolic antagonists are good examples of drugs that are compatible with drugs that have a high structural similarity to human metabolites.

Therefore, the metabolism analogy-based disease-related protein-drug interaction prediction is performed using the relationship information of the target enzyme-substrate-metabolic antagonist obtained from metabolic antagonist drugs already in clinical use as reference data sets.

However, the present invention is not limited to a metabolite-antagonistic substance-based drug, and the composition and method of the present invention can be applied to various drugs.

The reference similarity degree selection unit 105 collects from the public database or collects the target information of each of the metabolite antagonistic drugs and the drug from the database entered by the user. The reference similarity selection unit 105 can collect enzyme information mediating the substrate and the reaction from the 'Recon2', 'KEGG human pathway', and 'BRENDA' databases.

According to one embodiment, the reference similarity degree selection unit 105 considers only the enzymes in the target of the metabolic antagonist drug, and when there is a structural similarity reference value between the drug and the target enzyme substrate, that is, a chemical structure similarity score of 0.5 or less, Respectively. When the reference similarity-predicting unit 105 is mapped to two or more substrates mediated by a target enzyme of a metabolic antagonist drug, the substrate 105 having the highest chemical structure similarity with the metabolite antagonist-based drug in the substrate of the reaction mediated by the target enzyme .

For example, Gemcitabine (metabolic antagonist) - TYMS (target enzyme) - dUMP (substrate) - 0.82 (similarity) relationship is mapped and Gemcitabine (metabolic antagonist) - TYMS (target enzyme) - Methylene - When 0.62 (degree of similarity) is also mapped, both can be selected as the reference data set considering only the first relation (similarity 0.82), which is higher than 0.5 but higher in similarity.

In this manner, the reference similarity degree selection unit 105 can select a reference data set having a relationship between 18 metabolic antagonistic substances, 11 target enzymes, and 15 substrates, and the selected reference data set is as shown in FIG.

9, the reference data set includes a target enzyme, a substrate to which a target enzyme mediates a reaction, and an antimetabolite, and their structural similarity score and significance probability (P-value).

At this time, the structural similarity score is provided from the disease-associated protein-drug scoring matrix of FIGS. The probability of significance is a value derived from the significance evaluation of the disease-related human metabolite calculated in FIG. 5, and is mapped to the metabolic antagonist. Metabolism antagonists are similar in structure to human metabolites, so a significance assessment is mapped to a derived value indicating whether the interaction relationship between a particular human metabolite and a particular disease-associated protein is statistically significant.

Referring again to FIG. 8, the reference similarity degree selection unit 105 performs a performance comparison to verify the prediction accuracy of the selected reference data set (S403). At this time, the prediction accuracy of the reference data set is verified based on the Receiver Operating Characteristic curve.

Here, the reference similarity degree selection unit 105 can use the ROCR library of the R programming language to generate the receiver operation characteristic curve. ROCR is a program that draws a ROC curve by freely selecting one performance measure for the X and Y axes.

The reference similarity degree selection unit 105 generates the recipient operation characteristic curve of the present invention based on the similarity score included in the reference data set among the similarity degree matrices of Fig.

The reference similarity degree selection unit 105 uses three already known (or disclosed) prediction algorithms, i.e., SwissTargetPrediction, TargetNet, and Libdock (Site-Directed Docking Program) Lt; RTI ID = 0.0 > a < / RTI >

Here, the SwissTargetPrediction (STP) algorithm predicts molecular-target interactions using a combination of second- and third-dimensional chemical structure similarity scores. The STP algorithm is a web-based tool that predicts and provides up to 15 possible interactions for query molecules. Therefore, the reference similarity degree selection unit 105 queries the STP tool for 1,861 FDA-approved drugs to extract a predicted target candidate, and aligns the molecule-target in descending order based on the probability score provided in the STP tool. Then, the reference similarity degree selection unit 105 generates a receiver operation characteristic curve based on this probability score, and is compared with the present invention as shown in FIG. Since the STP tool predicts only 13 relationships among the 26 reference data set relationships, only 13 relationships are used as the reference data sets, and the comparison with the present invention is proceeded.

Referring to FIG. 10, the receiver operation characteristic curve of the receiver operation characteristic curve (Metabolite-likeness) and SwissTargetPrediction of the present invention is shown.

In addition, the TargetNet (TN) algorithm predicts molecular-target interactions based on the structure-activity relationship (SAR) model. The TN algorithm is also a web-based tool. The reference similarity selection unit 105 extracts predictive target candidates by querying the TN tool with 1,861 FDA-approved drugs, and calculates the numerator-target in descending order based on the probability score provided by the TN tool . Then, the reference similarity degree selection unit 105 generates a receiver operation characteristic curve on the basis of this probability score, and is compared with the present invention as shown in FIG. In the TN tool, like the STP tool, only 13 relationships among the 26 reference data set relationships were predicted. Therefore, comparison with the present invention was made using only 13 relationships as reference data sets. The relationship of the 13 reference data sets used to generate the recipient manipulation characteristic curve of the STP tool and the 13 reference data set relationships used to generate the recipient manipulation characteristic curve of the TN tool have different relationships.

Referring to FIG. 11, the recipient manipulation characteristic curve of the present invention and the recipient manipulation characteristic curve of TargetNet are shown.

Libdock is also an algorithm of molecular docking. The reference similarity degree selection unit 105 performs the molecular docking experiment using the Accelrys Discovery Studio 3.1 (DS) program. The standard affinity selection unit 105 performs docking experiments on dihydrofolate reductase (DHFR) and thymidylate synthase (TYMS) enzymes using 1,861 FDA-approved drugs. The reference similarity degree selection unit 105 collects the X-ray crystal structure complex file of DHFR and the substrate folate in a protein data bank (PDB ID: 1DHF). In addition, the reference similarity degree selection unit 105 collects an X-ray crystal structure complex file of TYMS and dUMP as a substrate in a protein data bank (PDB ID: 1HVY). The reference similarity selection unit 105 proceeds with DS by preparing and minimizing protein structure, adding hydrogen atoms, removing water molecules, and neutralizing pH environment. The active site of each protein is defined as a radius 10 Å around the substrate binding site. The reference similarity degree selection unit 105 acquires the Libdock score calculated through the DS libdock algorithm module and arranges the libdock scores in descending order of the libdock scores of only one drug. Then, the reference similarity degree selection unit 105 generates a receiver operation characteristic curve based on the libdock score, and is compared with the present invention as shown in FIG. In the Libdock algorithm, only 10 relationships among the 26 reference data set relationships were compared with the present invention.

Referring to FIG. 12, the recipient manipulation characteristic curve of the present invention and the recipient manipulation characteristic curve of Libdock are shown.

Referring to FIGS. 10, 11 and 12, the X axis of each receiver operation characteristic curve represents the specificity, and the Y axis represents the sensitivity.

The reference similarity degree selection unit 105 calculates the area under the curve (AUC) for each receiver operation characteristic curve to compare the performance of the receiver operation characteristic curve.

10, the AUC of the present invention is 0.914, and the AUC of STP is 0.658. 11, the AUC of the present invention is 0.991 and the AUC of TN is 0.862. In Figure 12, the AUC of the present invention is 0.989 and the AUC of libdock is 0.721.

Therefore, the AUC of the present invention has a larger value than the AUC of each of STP, TN, and libdock. Therefore, it can be seen that the reference data set prediction of the present invention is superior to STP, TN, and libdock.

Referring again to FIG. 8, the reference similarity degree selection unit 105 generates a receiver operation characteristic curve as shown in FIG. 13 for the 26 reference data set relationships (S405). Then, a Wedenen's index is calculated based on the recipient operating characteristic curve (S407), and a graph is generated as shown in FIG. The Yodden index is calculated as shown in Equation (1).

는 판단 기준을 x값으로 할때의 판단 결과의 정확도를 의미한다.

는 판단 기준을 x값으로 할때의 판단 결과의 민감도를 의미한다. 이때, x는 기준 데이터 세트의 화학적 구조 유사도 점수를 의미한다. here,

Means the accuracy of the determination result when the determination criterion is set as the x value.

Means the sensitivity of the determination result when the determination criterion is x. Here, x represents the chemical structure similarity score of the reference data set.

) 정보를 포함하는 X축 값과 민감도(

)를 나타내는 Y축 값을 이용하여 요덴 지표를 계산한다.The reference similarity degree selection unit 105 determines the accuracy degree of the similarity score of the reference data set in the recipient operation characteristic curve of the present invention in Fig.

) X axis values and sensitivity (

) Is used to calculate the Yodden index.

The reference similarity degree selection section 105 generates a graph as shown in FIG. 14 based on the Yodden index calculated by the above equation (1). Referring to FIG. 14, when the structural similarity score is 0.654, the Yield index shows the highest value of 0.979. Therefore, the reference similarity degree selection unit 105 selects 0.654 as the threshold similarity score. This critical similarity score is judged to be the reference similarity score that best classifies the reference data set.

FIG. 15 is a flowchart illustrating a drug regeneration candidate predicting process according to an embodiment of the present invention, and FIG. 16 is a table showing drug regeneration candidates related to a metabolic antagonistic substance according to an embodiment of the present invention. At this time, FIGS. 15 and 16 show the operation of the drug regeneration candidate predicting unit 107. FIG.

Referring to FIG. 15, the drug regeneration candidate predicting unit 107 applies a reference similarity score to a scoring matrix between a disease-related protein and a drug (S501), and predicts drug regeneration candidates for each disease.

The medicament regeneration candidate predicting unit 107 extracts medicines to which a similarity score higher than the reference similarity score is mapped (S503), and predicts the medicament regeneration candidate for the disease (S505). In this case, the higher the similarity score than the standard similarity score, the more likely the candidates to regenerate the drug. However, since it is to predict candidates for new regeneration, drugs equivalent to human metabolites among FDA-approved drugs may be excluded.

Referring to FIG. 16, the highest score among the drug regeneration candidates for a total of 11 disease-related enzymes related to the metabolic antagonist contained in the reference data set is shown. The target enzyme, the substrate, the substrate, the candidate drug, the new target disease (Indicated Disease), and the structural similarity score of these target enzymes (Target Enzyme) Respectively.

At this time, in the case of the target enzyme XDH, there was no new candidate drug having a similarity score of 0.654 or more as the standard similarity score.

Thus, in order to find out whether the 10 candidate regenerating drug candidates predicted according to the embodiment of the present invention are actually highly likely candidate drugs, a literature survey was carried out to find out that seven drug-disease out of the 10 predicted drug- It was supported by a literature survey that the relationship was already known.

On the other hand, in order to determine whether the above-described standard similarity scores can predict a new drug candidate that can treat a specific disease other than the selected reference data set, experiments for predicting drug candidates of Gaucher disease, .

Ghoshha disease is an autosomal recessive hereditary disease caused by the accumulation of glucosylceramide due to lack of activity of glucocerebrosidase enzyme or its quantity. Currently, Enzyme Replacement Therapy is mainly used to treat this disease. However, the treatment cost is close to 300 million won per year, and various unexplained reasons have caused the treatment effect to vary widely. In addition, when enzyme replacement therapy is not possible, Substrate Reduction Therapy is underway, and there are currently only two FDA-approved drugs, miglustat and eliglustat. If there is no treatment effect, there is no treatment option.

In the experimental example of the present invention, a metabolite substance having a substrate relationship with Glucocerebrosidase, Ceramide glucosyltransferase and the like was used as an enzyme related to Gossypetic Disease in order to predict a drug regeneration candidate for treatment of a Gochae disease.

Figure 17 shows a fragment of the metabolic pathway including enzymes and metabolites associated with Gossypollosis in accordance with one embodiment of the present invention.

Referring to FIG. 17, Lactosylceramide has a Ceramide-Glucose-Galactose structure. Lactosylceramide is converted to glucosylceramide by the action of beta-galactosidase, which produces a metabolite called galactose.

Glucosylceramide is composed of Ceramide-Glucose structure. Glucocerebrosidase produces the metabolite called Glucose. Glucocerebrosidase enzymes are directly linked to Gossypia.

Galactosylceramide is composed of Ceramide-Galactose structure, which is transformed into Ceramide by the action of Galactosylceramidase, which generates a metabolite called Galactose. Ceramide is converted to glucosylceramide by enzymatic action of ceramide glucosyltransferase, which generates a metabolite called Glucose. Miglustat, a traditional ghosharma drug, blocks the activity of ceramide glucosyltransferase enzymes and thus prevents the conversion of ceramide to glucosylceramide. Because accumulation of glucosylceramide due to the inactivity of glucocerebrosidase enzyme is the cause of cholesteatoma, it can treat choleric disease by preventing the change of ceramide to glucosylceramide.

Therefore, the enzyme proteins associated with Gossypium are Glucocerebrosidase and Ceramide glucosyltransferase. Glucosylceramide and Ceramide are the metabolites associated with Gossyphei. Therefore, the chemical structure of the drugs with Glucosylceramide and Ceramide is compared and the method Likewise, we predicted drug candidates. At this time, a total of 36 drug regeneration candidates were obtained, except miglustat, which is a conventional Goshhee drug.

Figure 18 is a list of candidates supported by evidence through literature review in accordance with one embodiment of the present invention. Among drug candidates predicted to be used in the treatment of Gochishha disease, candidates supported by the literature survey Lists.

Referring to FIG. 18, it can be seen that half of the total of 36 new candidate drugs for remedy of GCS disease are antibiotics (Aminoglycosides). The other half were classified as anti-hypertension, immunosuppressant, and anti-diabetic.

We conducted a literature review to verify the efficacy of these drug candidates. Most antibiotics among amendment candidates were aminocyclitol - related antibiotics related to aminoglycoside. Recently, aminocyclitol derivatives have been reported to be effective against chorioallans.

Miglustat, which is currently used in the treatment of ghoshha disease, was originally developed as a Nojirimycin family of antibiotics.

Recently, there have been reports that antihypertensive agents and anti-immunity agents may be effective in the treatment of Gha-Shi disease.

As such, it can be seen that candidates for drug regeneration are likely candidates for the new drug of Kocher's disease. Thus, it shows the possibility of suggesting drug regeneration candidates using the similarity of human metabolites interacting with disease-related proteins.

In particular, it appears that there will be great advantages in regenerating new drugs of genetic rare diseases associated with enzymatic dysfunction, such as Gossypian, so that the embodiments of the present invention can be used in the pharmaceutical industry as a whole, including the development of human metabolite- It is expected to be a helpful tool to help.

Meanwhile, FIG. 19 is a block diagram illustrating a hardware configuration of a drug regeneration candidate prediction system according to another embodiment of the present invention.

19, the drug regeneration candidate prediction system 1000 includes at least one storage unit 1001, at least one output unit 1003, at least one input unit 1005, at least one communication unit 1007, And at least one processor (1009). The medicament regeneration candidate prediction system 1000 includes various software such as an operating system, middleware, and programs that operate in combination with hardware. The hardware and software of the drug regeneration candidate prediction system 1000 have the configuration and the capability to execute the present invention.

The storage unit 1001 stores a program for implementing the configuration and method described with reference to FIG. 1 to FIG. The storage unit 1001 includes any medium that stores or transmits data in a form readable by a device such as a computer. The storage unit 1001 may include a read only memory (ROM), a random access memory (RAM) Storage media, flash memory devices, and other electrical, optical, or acoustic signal transmission media, and the like. The program stored in the storage unit 1001 includes instructions that implement the operation of the drug regeneration candidate prediction system described with reference to FIGS. The processor 1009 loads the program and performs operations of the drug regeneration candidate prediction system 1000 described in FIGS. 1 to 19 described in the present invention.

The output unit 1003 outputs information according to the operation of the processor 1009. At this time, it can be outputted on the screen, printed matter, or output in various ways which can be recognized by the user. The input unit 1005 collects information necessary for construction on the Internet or receives it from the user and outputs it to the processor 1009. [

The communication unit 1007 is connected to the communication network 800 and performs a data transmission / reception function.

In this case, each of the components 1001, 1003, 1005, and 1007 is shown mounted in one processor 1009, which means that the components 1001, 1003, 1005, and 1007 operate on a processor basis , A single processor 1009, or may be implemented as a separate server connected to each other via Ethernet or a network.

The embodiments of the present invention described above are not implemented only by the apparatus and method, but may be implemented through a program for realizing the function corresponding to the configuration of the embodiment of the present invention or a recording medium on which the program is recorded.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, It belongs to the scope of right.

Claims (15)

WHAT IS CLAIMED IS: 1. A method for predicting drug regeneration candidates by a system operating by at least one processor,
Collecting information on disease-related proteins associated with disease outbreaks, information on interaction between proteins and compounds, and human metabolite information,
Extracting interaction relationship information between the disease-associated protein and the human metabolite based on the collected information,
Generating a similarity matrix of the drug and the human metabolite based on the similarity score between the chemical fingerprint of each structure of the drug and the structure of the human metabolite,
Generating a scoring matrix by substituting the disease metabolism material with the disease-associated protein in the similarity matrix based on the interaction relationship information; and
Extracting a drug having a similarity score higher than the reference similarity score among the drugs of the scoring matrix, and predicting the extracted drug as a re-creation candidate
A candidate drug candidate candidate prediction method. delete The method of claim 1,
Wherein the similarity degree matrix includes:
A method of predicting a drug re-creation candidate, the method comprising: hierarchically clustering using a heat map function and displaying similarity scores between respective chemical fingerprints in different colors for each segment. The method of claim 1,
Between the step of generating the scoring matrix and the step of predicting,
Selecting a reference data set including relationship information between a drug target protein, a human metabolite, and a drug using drug target protein information collected from a public database and interaction relationship information between the disease-associated protein and a human metabolite ,
Generating a Receiver Operating Characteristic curve based on a score of similarity between the chemical fingerprint of the human metabolite and the drug contained in the reference data set; and
Extracting the similarity score having the highest Yoden index (Youden's Index) calculated using the receiver operation specific curve, and selecting the similarity score as the reference similarity score
Further comprising the step of: 5. The method of claim 4,
After selecting the reference data set,
A recipient manipulation characteristic curve generated based on the score of similarity between the chemical fingerprints of the human metabolite and the drug contained in the reference data set and the predicted value calculated using at least one prediction algorithm for predicting the protein target interaction of the compound Comparing the recipient operating characteristic curves generated based on the recipient operating characteristic curves to verify the prediction accuracy of the reference data set
Further comprising the step of: The method of claim 1,
Wherein the extracting comprises:
Related protein information and the metabolism information of the human body, a significance evaluation is performed on the basis of the number of interaction relations with respect to the information on interaction relation between the disease-related protein information and the metabolism information of the human body, thereby extracting the interaction relationship information determined to be statistically significant. Way. The method of claim 6,
Wherein the extracting comprises:
(P value) according to the significance evaluation satisfies the threshold condition among the information of the interaction between the disease-related protein information and the human metabolism information at least two of which match each other, and the priority according to the significance probability Extracting the highest priority interaction relationship information,
The extracted interaction relationship information includes,
Wherein the disease-associated protein and the human metabolite are mapped on a one-to-one basis. WHAT IS CLAIMED IS: 1. A method for predicting drug regeneration candidates by a system operating by at least one processor,
A similarity score of the chemical structure between the drug and the human metabolite is calculated using a chemical fingerprint of each chemical structure of each known drug and human metabolite and a drug having similar chemical structure and a human metabolite Matching step,
Replacing the human metabolite with a disease-associated protein that interacts with the human metabolite, and
Related protein having a similarity score higher than the reference similarity score and the disease-associated protein matched to the drug, and predicting the extracted drug as a drug regeneration candidate of the extracted disease-associated protein
A candidate drug candidate candidate prediction method. delete 9. The method of claim 8,
The score of the reference similarity score
A drug selected using a Receiver Operating Characteristic curve and Youden's Index based on the score of similarity between chemical fingerprints of the reference data set including the relationship between the drug target protein, the human metabolite and the drug, A method for predicting re - creation candidates. 9. The method of claim 8,
Wherein the replacing comprises:
A method for predicting drug re-creation candidates using interaction-related information having a priority of a significance value (P value) satisfying a critical condition among the information on the interaction relation of human metabolism with at least two disease-related proteins matched with each other . A system for predicting drug re-creation candidates,
A memory for storing a program for predicting drug regeneration candidates, and
And at least one processor for executing the program,
The program includes:
Extracting interaction relationship information between a disease-associated protein and a human metabolite based on disease-related protein information associated with disease occurrence collected from at least one public database, information on interaction between a protein and a compound, and human metabolite information,
The method comprising: matching a drug and a human metabolite having a chemical structure similar to each other based on a similarity score between a known drug and a chemical fingerprint of a structure of each of the human metabolites; Said disease-associated protein,
And instructions for extracting a drug and disease-related protein having a degree of similarity score higher than a reference similarity score and predicting the extracted drug as a drug regeneration candidate of a related disease of the extracted protein, system. The method of claim 12,
The program includes:
Related protein information and the interaction relation information of the human metabolite information on the basis of the number of interaction relations and determine the statistical significance based on the priority of the significance value (P value) Extracting the action relation information,
Generating a scoring matrix consisting of scores of similarity between the drug and the disease-associated protein,
And a command for predicting the drug and disease-related protein having the similarity score higher than the reference similarity score using the scoring matrix to the drug regeneration candidate and outputting the predicted information to a screen, . The method of claim 13,
The program includes:
Based on the drug target protein information collected from the public database and the similarity score between the chemical fingerprint of the chemical structure of the drug target protein, the human metabolite and the drug contained in the selected reference data set using the interaction relationship information, A curve (Receiver Operating Characteristic curve) is generated,
Calculating a YODEN index according to the receiver operation characteristic curve and selecting the similarity score having the highest YODEN index as the reference similarity score. The method of claim 14,
A drug regeneration candidate database storing drug regeneration candidate information including the predicted drug and disease-related protein relationship information, and
A user interface unit connected to the drug regeneration candidate database and managing access to the drug regeneration candidate database to a terminal connected through a communication network,
Further comprising the step of:

Images