A method for evaluating the effect of multi-target combination on the efficacy of drug treatment

Abstract

The present invention provides a method for evaluating the effect of a multi-target combination on the therapeutic effect during drug treatment. The method effectively associates the targets in the drug treatment process with their therapeutic effect by calculating topological descriptors of nodes in a complex network and constructing a statistical model, thereby quantitatively evaluating the effect of a multi-target combination (module) or a single target on the therapeutic effect during drug treatment of a disease.

Description

Technical Field

[0001] The present invention belongs to the field of bioinformatics and specifically relates to a method for evaluating the effect of a multi-target combination on the therapeutic effect during drug treatment. Background Art

[0002] It is known that there are many interactions between genes, proteins, metabolites, etc. Using graph theory methods, these metabolites, genes, proteins and other molecules can be regarded as nodes, and the interactions between them can be regarded as edges, thus forming an interaction network. Clusters of molecules with relatively close connections in the network are called modules.

[0003] As evidence supports the multi-target effects of drugs on diseases, the study of modules as multi-target combinations of drugs is increasingly being used in multiple fields, including drug development and pharmacological mechanism elucidation. Drugs act on diseases through multiple targets, multiple levels, and multiple pathways. However, quantitatively evaluating the impact of multi-target combinations / single targets on the efficacy of drugs in disease processes remains an unresolved issue. Furthermore, in biological networks, targets are interconnected and influence each other. Therefore, evaluating the impact of multi-target combinations / single targets on the efficacy of drugs in disease processes at the biological network level can quantitatively explain the impact of multi-target combinations on efficacy. Summary of the Invention

[0004] To address the above technical issues, the present invention provides a method for evaluating the impact of targets on the therapeutic efficacy during drug treatment. The method performs the evaluation based on the calculation of topological descriptors of nodes in the biomolecular interaction network formed by the targets and the construction of a statistical model.

[0005] The technical solutions of the present invention are as follows.

[0006] The present invention provides a method for evaluating the effect of a multi-target combination on the therapeutic effect of a drug in a disease process, the method comprising the following steps:

[0007] (1) Obtain biological samples without drug intervention and with drug intervention, respectively, use m biomarkers as nodes and the content of each biomarker in the biological samples as edges, and use the Pearson correlation coefficient to establish the biomarker interaction network without drug intervention and with drug intervention respectively;

[0008] (2) Using the number of nodes ≥3 as the standard, the interaction networks without drug intervention and with drug intervention were subjected to module identification to obtain modules and scattered nodes that did not form modules respectively;

[0009] (3) For the modules and scattered nodes without drug intervention and with drug intervention obtained in step (2), the topological attributes of each node of the module and each scattered node are calculated as topological descriptors;

[0010] (4) The Kolmogorov-Smirnov test was used to compare the differences in the topological descriptors of the same node before and after drug intervention. This was done by comparing the values of multiple topological descriptors of each node and each scattered node in the component modules before and after drug intervention, and finally obtaining a P value. Nodes with significant differences were then selected as drug intervention-specific nodes based on the P value.

[0011] (5) For the modules and scattered nodes obtained after drug intervention in step (2), the weight factor of each module and each scattered node on drug efficacy is calculated using the following multiple linear regression equation:

[0012] h(x)=α0x0+α1x1+α2x2+.....+α n x n

[0013] Where h(x) is the efficacy of the drug, x is the total content of nodes in each module and / or the content of scattered nodes, α is the weight factor of each module or scattered node, and n is the number of modules and scattered nodes;

[0014] (6) Using the following weighted feature significance (WFS) method, the impact score of the module or scattered node on the efficacy is calculated based on the number of modules and scattered nodes after drug intervention obtained in step (2), the number of specific nodes obtained in step (4), and the weight factor on the efficacy obtained in step (5):

[0015]

[0016] where p x is the P value calculated in step (4) for each node and each scattered node in the module of the group that received drug intervention compared with the group that did not receive drug intervention, C is the number of nodes in each module or 1 in the case of scattered nodes, M is the number of specific nodes in each module or 1 in the case of scattered nodes, N C-M is the number of nodes in each module minus the number of specific nodes, and α is the weight factor of each module or scattered node calculated in step (5).

[0017] The method provided by the present invention can be performed on traditional Chinese medicine, chemical medicine or the active ingredients of medicine. That is, the medicine is traditional Chinese medicine, chemical medicine or the active ingredients of medicine.

[0018] Preferably, in step (1) of the method provided by the present invention:

[0019] The biological sample is serum or plasma, or tissue.

[0020] The biomarkers are metabolites and / or proteins detected in serum.

[0021] The number of biomarkers is ≥3.

[0022] The Pearson correlation coefficient is calculated as follows:

[0023]

[0024] Among them, x i and y i Represents two different nodes, N represents the total number of nodes, r represents the correlation coefficient of the two nodes, and when r is greater than 0 and less than 1, it means x i and y i There is a positive correlation. When r is greater than -1 and less than 0, it means x i and y i There is a negative correlation, and when r = 0, x i and y i Totally irrelevant.

[0025] Preferably, in step (2) of the method provided by the present invention:

[0026] The MCODE cluster plug-in in Cytoscape 3.7.2 software was used to identify modules in the interaction network before and after drug intervention, with the parameters set as k-core = 2, Max Depth = 100, Node Score cutoff = 0.2, and Degree Cutoff = 2.

[0027] Preferably, in step (3) of the method provided by the present invention:

[0028] The topological properties are selected from one, more or all of characteristic path length, betweenness centrality, closeness centrality, clustering coefficient, degree, eccentricity, neighborhood connectivity, topological coefficient, number of directed edges, radiality and gravity centrality.

[0029] Preferably, in step (4) of the method provided by the present invention:

[0030] Nodes with a P value < 0.1 were selected as specific nodes for drug intervention.

[0031] Preferably, in step (5) of the method provided by the present invention:

[0032] Before calculating the weight factor, the content of each node or each scattered node constituting the module is normalized as follows:

[0033]

[0034] Where max is the maximum value of the sample data, min is the minimum value of the sample data, i is the content value of the biomarker measured in the tissue, and i* is the content value of the biomarker after normalization.

[0035] The method provided by the present invention can be repeated using data obtained at different time periods after drug intervention, and the validity of the model can be verified by comparing the WFS score results calculated at different times.

[0036] The present invention provides a method for evaluating the effects of targets on the therapeutic efficacy of drug treatments in biomolecular networks. By calculating topological descriptors of nodes in complex networks and constructing statistical models, the method effectively associates targets with the therapeutic efficacy of the drug treatment, thereby quantitatively evaluating the effects of multiple target combinations (modules) or single targets on the therapeutic efficacy of the drug treatment. Compared with existing methods, the evaluation method provided by the present invention is more comprehensive, integrated, and quantitative, demonstrating methodological innovation. It provides a scientific basis for guiding the prediction of target effects on the therapeutic efficacy of drug treatments and for identifying targets in new drug development. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, in which:

[0038] Figure 1 target interaction networks for different groups;

[0039] Figure 2 Module division results for target interaction networks of different groups. DETAILED DESCRIPTION

[0040] The purpose of this invention is to evaluate the impact of targets on the efficacy of drug treatment in the process of disease treatment from the complex network after drug intervention, so as to provide a scientific basis for predicting the impact of targets on efficacy and determining targets in the process of new drug development.

[0041] The present invention is described below with reference to specific examples. Those skilled in the art will appreciate that these examples are only used to illustrate the present invention and are not intended to limit the scope of the present invention in any way.

[0042] The experimental methods in the following examples are conventional methods unless otherwise specified. The raw materials, reagents, etc. used in the following examples are commercially available products unless otherwise specified.

[0043] Example 1

[0044] The method of the present invention is specifically described by taking the metabolite interaction network of different drugs improving type 2 diabetic peripheral neuropathy in db / db mice as an example.

[0045] Experimental data: This involves target data and efficacy data. The target data are the concentration data of 15 target metabolites (see Table 1) in the serum of mice with type 2 diabetic peripheral neuropathy improved by different drugs. The efficacy data are derived from the nerve conduction velocity data (motor nerve conduction velocity (MCV) and sensory nerve conduction velocity (SCV)) in db / db mice with type 2 diabetic peripheral neuropathy improved by different drugs.

[0046] The method of the present invention was first performed using the target data and efficacy data from the 12th week of administration, and then verified using the target data and efficacy data from the 6th week of administration.

[0047] The mice with type 2 diabetic peripheral neuropathy model were divided into 7 groups, namely, normal group, model group, Mudan granule group, epalrestat group, Liuwei Luobi granule low-dose group, Liuwei Luobi granule medium-dose group and Liuwei Luobi granule high-dose group.

[0048] Animal Grouping and Dosing: 120 SPF-grade male 12-week-old db / db mice (spontaneous type 2 diabetic mice) were selected and randomly divided according to blood glucose levels and body weight into six dose groups, each containing 20 animals: a model group, a Mudan granule group (2.7 g / kg), an epalrestat group (19.5 mg / kg), and low-, medium-, and high-dose Liuwei Luobi granule groups (4.4, 8.7, and 17.4 g crude drug / kg). Twenty db / m mice of the same age were selected as the normal control group. Each dose group was given the corresponding drug by gavage at 20 mL / kg once daily for 12 consecutive weeks. The normal and model groups were given an equal volume of purified water by gavage.

[0049] Testing of sciatic nerve conduction velocity and sensory conduction velocity in mice: At the 6th and 12th weeks of drug administration, 3 (6th week) and 10 (12th week) mice were randomly selected from each group, respectively. After anesthesia, the animals were fixed in a prone position. The room temperature was strictly controlled at 20±0.5°C, and the mouse body temperature was maintained at 37°C. The motor conduction velocity and sensory nerve conduction velocity of the right sciatic nerve of the mice were measured using a BL-420S biofunctional experimental system. The specific method was as follows: the skin between the biceps femoris and semimembranosus muscles of the right lower limb was opened, and blunt dissection was performed along the two muscles to expose and separate the right sciatic nerve. The exposed nerve was gently hung on an electrode hook containing two sets of stimulating and recording electrodes. Paraffin oil preheated at 37°C was dripped onto the surface of the nerve to maintain the nerve in a moist and warm environment. The stimulation intensity was 1.5 times the stimulation threshold. Measurement method: (1) MCV: Place one end of the stimulating electrode at the proximal end of the nerve and one end of the recording electrode at the distal end of the nerve. After stimulation, record the evoked action potential. The time from the application of stimulation to the appearance of the evoked potential is the latency. The distance between the starting point of the stimulating electrode and the starting point of the recording electrode is the distance. Repeat the measurement 3-5 times and calculate the average value. (2) SCV: Swap the positions of the stimulating electrode and the recording electrode so that they are opposite to the position of the MCV measurement.

[0050] Detection of metabolite content in mouse serum: At the 6th and 12th weeks of administration, after the MCV and SCV of mice were determined, serum was collected from the mice. The specific procedures were as follows: the mice were fasted but not watered for 12 hours before sample collection. The animals were anesthetized with an intramuscular injection of 9.75 mg / kg xylazine hydrochloride, and 1-1.5 ml of blood was collected from the orbital venous plexus of the mice. The blood was allowed to stand at room temperature (20°C-25°C) for approximately 30 minutes (no more than 2 hours), and then centrifuged at a centrifugal force of approximately 3000g for 10 minutes. Approximately 200-300 μL of the upper transparent liquid was carefully aspirated into a 1.5 mL centrifuge tube, and the centrifuge tube was capped and checked for fit. The tube was placed in a sample box, labeled with sample information, and stored in a -80°C refrigerator. The content of each metabolite in the mouse serum was then detected using targeted metabolomics technology.

[0051] Step 1: Obtain biological samples without drug intervention and with drug intervention, respectively, use m biomarkers as nodes and the content of each biomarker in the biological sample as edges, and use the Pearson correlation coefficient to establish the biomarker interaction network without drug intervention and with drug intervention respectively; see Figure 1 ;

[0052] Step 2: Using the number of nodes ≥ 3 as the standard, the interaction networks without drug intervention and with drug intervention were identified as modules, and the modules and scattered nodes that did not form modules were obtained respectively; see Figure 2 ;

[0053] Step 3: For the modules and scattered nodes without drug intervention and with drug intervention obtained in step (2), the topological attributes of each node of the module and each scattered node are calculated as topological descriptors; see Table 1;

[0054] Step 4: The Kolmogorov-Smirnov test was used to compare the topological descriptors of the same node between the pre- and post-drug intervention phases. This was done by comparing the values of multiple topological descriptors for each node and each scattered node in the pre- and post-drug intervention phases, ultimately yielding a P value. Nodes with significant differences were then selected as drug intervention-specific nodes based on the P value; see Table 2.

[0055] Step 5: For the modules and scattered nodes obtained after drug intervention in step (2), the weight factor of each module and each scattered node on drug efficacy is calculated using the following multiple linear regression equation:

[0056] h(x)=α0x0+α1x1+α2x2+.....+α n x n

[0057] Where h(x) is the efficacy of the drug, x is the total content of nodes in each module and / or the content of scattered nodes, α is the weight factor of each module or scattered node, and n is the number of modules and scattered nodes; see Table 3;

[0058] Step 6: Use the following weighted feature significance (WFS) method to calculate the impact score of the module or scattered node on the efficacy based on the number of nodes, number of specific nodes, and weight factors of the module or scattered node after drug intervention obtained in step (2):

[0059]

[0060] where p x is the P value calculated in step (4) for each node and each scattered node in the module of the group that received drug intervention compared with the group that did not receive drug intervention, C is the number of nodes in each module or 1 in the case of scattered nodes, M is the number of specific nodes in each module or 1 in the case of scattered nodes, N C-M is the number of nodes in each module minus the number of specific nodes, and α is the weight factor of each module or scattered node calculated in step (5). See Tables 4 and 5.

[0061] Through the above analysis examples, it can be concluded that although the present invention only uses a method of integrating multiple parameters to screen specific targets for the metabolite interaction network, and uses the WFS model to quantitatively evaluate the effect of the target on the therapeutic effect after the drug acts on the disease, it shows that the method of evaluating the effect of the target on the therapeutic effect in the biological network is effective and feasible.

[0062] The above description of the specific embodiments of the present invention does not limit the present invention. Those skilled in the art can make various changes or modifications based on the present invention. As long as they do not depart from the spirit of the present invention, they should fall within the scope of the claims attached to the present invention.

[0063]

[0064]

[0065] Table 2 Screening of specific targets

[0066]

[0067]

[0068] Table 3 Weight factors affecting target efficacy

[0069]

[0070] Table 4 WFS impact scores of targets on efficacy

[0071]

[0072]

[0073] Table 5 Verification results of the WFS model

[0074] .

Claims

1. A method for evaluating the effect of a multi-target combination on the efficacy of drug therapy, the method comprising the following steps: (1) Obtain biological samples without drug intervention and with drug intervention, respectively, use m biomarkers as nodes and the content of each biomarker in the biological samples as edges, and use the Pearson correlation coefficient to establish the biomarker interaction network without drug intervention and with drug intervention respectively; (2) Using the number of nodes ≥3 as the standard, the interaction networks without drug intervention and with drug intervention were subjected to module identification to obtain modules and scattered nodes that did not form modules respectively; (3) For the modules and scattered nodes without drug intervention and with drug intervention obtained in step (2), the topological attributes of each node of the module and each scattered node are calculated as topological descriptors; (4) The Kolmogorov-Smirnov test was used to compare the differences in the topological descriptors of the same node before and after drug intervention. This was done by comparing the values of multiple topological descriptors of each node and each scattered node in the component modules before and after drug intervention, and finally obtaining a P value. Nodes with significant differences were then selected as drug intervention-specific nodes based on the P value. (5) For the modules and scattered nodes obtained after drug intervention in step (2), the weight factor of each module and each scattered node on drug efficacy is calculated using the following multiple linear regression equation: h(x)=α0x0+α1x1+α2x2+.......+α n x n Where h(x) is the efficacy of the drug, x is the total content of nodes in each module and / or the content of scattered nodes, α is the weight factor of each module or scattered node, and n is the number of modules and scattered nodes; (6) Using the following weighted feature significance (WFS) method, the impact score of the module or scattered node on the efficacy is calculated based on the number of modules and scattered nodes after drug intervention obtained in step (2), the number of specific nodes obtained in step (4), and the weight factor on the efficacy obtained in step (5): where p x is the P value calculated in step (4) for each node and each scattered node in the module of the group that received drug intervention compared with the group that did not receive drug intervention, C is the number of nodes in each module or 1 in the case of scattered nodes, M is the number of specific nodes in each module or 1 in the case of scattered nodes, N C-M is the number of nodes in each module minus the number of specific nodes, and α is the weight factor of each module or scattered node calculated in step (5).

2. The method according to claim 1, characterized in that In step (1): The biological sample is serum or plasma, or tissue.

3. The method according to claim 1, characterized in that In step (1): The biomarkers are metabolites and / or proteins detected in serum.

4. The method according to claim 1, wherein In step (1): The number of biomarkers is ≥3.

5. The method according to claim 1, characterized in that In step (1): The Pearson correlation coefficient is calculated as follows: Among them, x i and y i Represents two different nodes, N represents the total number of nodes, r represents the correlation coefficient of the two nodes, and when r is greater than 0 and less than 1, it means x i and y i There is a positive correlation. When r is greater than -1 and less than 0, it means x i and y i There is a negative correlation, and when r = 0, x i and y i Totally irrelevant.

6. The method according to claim 1, wherein In step (2): The MCODE cluster plug-in in Cytoscape 3.7.2 software was used to identify modules in the interaction network of drug intervention and non-drug intervention, with the parameters set as k-core = 2, Max Depth = 100, NodeScore cutoff = 0.2, and Degree Cutoff = 2.

7. The method according to claim 1, characterized in that In step (3): The topological descriptors are selected from one, more or all of characteristic path length, betweenness centrality, closeness centrality, clustering coefficient, degree, eccentricity, neighborhood connectivity, topological coefficient, number of directed edges, radiality and gravity centrality.

8. The method according to claim 1, characterized in that In step (4): Nodes with a P value < 0.1 were selected as specific nodes for the drug intervention group.

9. The method according to claim 1, characterized in that In step (5): Before calculating the weight factor, the content of each node or each scattered node constituting the module is normalized as follows: Where max is the maximum value of the sample data, min is the minimum value of the sample data, i is the content value of the biomarker measured in the tissue, and i* is the content value of the biomarker after normalization.

10. The method according to any one of claims 1 to 9, characterized in that The medicine is a traditional Chinese medicine, a chemical medicine or its medicinal ingredients.

Images