CN115116561A - Construction method and application of drug-target protein-schizophrenia interaction network - Google Patents

Abstract

The invention belongs to the technical field of biomedicine, and discloses a construction method and application of a drug-target protein-schizophrenia interaction network. The invention screens out new drugs which can be used for relieving or even treating schizophrenia from the drug-target protein-disease interaction network by constructing the drug-target protein-disease interaction network and collecting candidate genes of schizophrenia. The construction method of the drug-target protein-schizophrenia interaction network provides a basis for repositioning related drugs for preventing or treating schizophrenia and detecting new clinical application of the existing drugs by the constructed interaction network. Can accelerate the drug development process and save a great deal of manpower and financial resources.

Description

A method for constructing a drug-target protein-schizophrenia interaction network and its application

technical field

The invention relates to the technical field of biomedicine, in particular to a construction method and application of a drug-target protein-schizophrenia interaction network.

Background technique

Schizophrenia is a serious mental illness with a lifetime prevalence of approximately 0.4% worldwide. It can have devastating effects on patients and their caregivers, and impose significant costs on the healthcare system. Schizophrenia is a heterogeneous disorder with positive symptoms (delusions, hallucinations, thought disturbances); negative symptoms (anhedonia, depression, social withdrawal, poor thinking), and cognitive dysfunction. With the deepening of understanding and clinical research on schizophrenia, schizophrenia has been gradually divided into different types, such as early-onset and late-onset schizophrenia according to the age of onset of schizophrenia, according to signs and symptoms, Symptoms of disease course, pathophysiological correlates, risk and etiological factors, and response to treatment are divided into deficient and nondeficient schizophrenia.

The etiology of schizophrenia remains poorly understood, and over the past few years, many different antipsychotics have been developed and tested for their efficacy and safety in reducing positive symptoms and maintaining stability in schizophrenia. Although a variety of antipsychotic drugs are available to reduce the symptoms of schizophrenia, response rates to these drugs are lower than expected, they act slowly, and often have serious adverse side effects.

Current antipsychotic treatments rely primarily on targeting brain dopamine D2 receptors, but newer drugs that act through glutamate receptors, glycine transporters, or alpha-7-nicotinic acetylcholine receptors are being developed. So far, however, none of these new approaches have resulted in a therapeutic breakthrough. Despite the undisputed efficacy of antipsychotics in treating the positive symptoms of schizophrenia, for most patients, the drugs are ineffective for all symptoms, including negative and cognitive symptoms, and severe side effects persist. While subsequent development of new drugs has reduced exercise side effects, other safety and tolerability concerns have arisen. Since the advent of antipsychotic treatment in the early 1950s, subsequent progress has been modest. However, there is still a lack of screening drugs for treatment according to the different symptoms and incidence patterns of different types of schizophrenia. Types of effective drugs are one of the most important research questions in the field of schizophrenia.

SUMMARY OF THE INVENTION

The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for constructing a drug-target protein-schizophrenia interaction network and its application.

To achieve the above object, the technical scheme adopted by the present invention is as follows:

In a first aspect, the present invention provides a method for constructing a drug-target protein-schizophrenia interaction network, comprising the following steps:

(1) Collect drug-target protein interaction relationships through drug databases, and screen out human targets and drugs that meet FDA standards;

(2) Mapping the disease to schizophrenia-related genes through the disease database, and integrating the relationship between disease-causing genes;

(3) Using Python to integrate the drug-target protein interaction relationship and the disease-pathogenic gene relationship to construct a drug-protein-schizophrenia interaction network.

The drug-protein-schizophrenia interaction network of the present invention provides a general approach to systematically understand how drugs treat diseases, systematically identify proteins related to treatment, and predict which schizophrenia-related genes will alter drug efficacy or cause For serious adverse effects of drug therapy, the multidimensional interactome can be easily expanded to add other node types associated with disease.

The drug-target protein-schizophrenia interaction network of the present invention can start from the different expression patterns of different schizophrenia types, and find the drugs that affect the corresponding targets according to their differential expression genes, so as to play a role in different types of schizophrenia. to the therapeutic effect.

As a preferred embodiment of the method for constructing a drug-target protein-schizophrenia interaction network according to the present invention, in the step (2), candidate genes for schizophrenia are collected and sequenced to obtain defective schizophrenia Differential genes for early-onset schizophrenia and early-onset schizophrenia differential genes; the candidate genes, defective schizophrenia differential genes and early-onset schizophrenia differential genes are pathogenic genes; the pathogenic genes are mapped by Cytoscape and R On the human protein interaction network, the schizophrenia-disease gene interaction relationship was screened out.

As a preferred embodiment of the method for constructing a drug-target protein-schizophrenia interaction network according to the present invention, the schizophrenia candidate genes include ABL2, ALDOA, ARHGAP1, ABR, SLC25A6, ARNT, ACACA, APAF1, STS , ACVR2A, APBA2, ASCL1, ADRA1A, BIRC3, SERPINC1, JAG1, APOE, RERE, AGT, FAS, ATP2A2, AK4, AQP6, ATP2B4, AKT1, AR, ATRX, ALDH1A1, ARL4D and KIF1A; the deficient schizophrenia Differential genes include ABR, ALAD, ARHGAP1, ACLY, ALAS1, ARNT, ACO2, ABCD1, ARRB1, ACTG1, ALDH1A1, GET3, ACTN4, ALDH3B1, ATIC, ACVR1B, ALOX15B, ATOX1, ACVR2B, APLP2, TNFRSF17, ADCY7, APP, ZFP36L1 , AP2A1, APRT, ZFP36L2, AKT1, ARF3, ARHGAP1; the early-onset schizophrenia differential genes include ABCF1, AQP3, POLR3D, ACVR18, ARHGAPS, BNIP1, PLIN2, ARHGDIA, KLF9, ADORA2A, ARHGDIB, C3AR1, PARP1, ATF3, TMEM258, AGER, ATP5F1D, CFAP410, AHR, ATPSF1E, CBFA2T3, ALDOA, ATP6V1A, SLC25A6, ATP6V1C1, CCNE1, FAS, BMI1, CCNH.

In the second aspect, the present invention applies the drug-target protein-schizophrenia interaction network obtained by the construction method in screening drugs for preventing or treating schizophrenia.

As a preferred embodiment of the application of the present invention, weights and restart probabilities are assigned according to the node types in the drug-target protein-schizophrenia interaction network, and random walks are restarted from the disease and drug nodes. Frequency of nodes Calculation of the effects of drugs and diseases on nodes in the multidimensional interaction group And the Euclidean distance between each pair is calculated separately, and the predicted drug ranking for a given schizophrenia is performed by the distance.

As a preferred embodiment of the application described in the present invention, the weight is assigned to the node as W={'indication':3.541889556309463,'protein':4.396695660380823,'drug':3.2071696595616364}, and the walker continues to walk at a given step instead of Probability of restarting α = 0.8595436247434408.

As a preferred embodiment of the application described in the present invention, the diffusion curves of drugs and diseases and the schizophrenia-related disease diffusion curve SIM are calculated by L2 norm, respectively.

The diffusion curves of the present invention provide predictive power and interpretability in drug-disease treatment modeling, allowing identification of proteins associated with the treatment of schizophrenia.

As a preferred embodiment of the application of the present invention, the drugs for treating schizophrenia include Rilonacept, Tetrabenazine, lsometheptene, Zuclopenthixol, Droperidol, Acetophenazine, lloperidone, Dopamine, Ketanserin, Enzastaurin, Vilazodone, Epicriptine, Dihydrexidine, Flupentixol, Taurine , Benzphetamine, Piceatannol, Citalopram, Butababital, and Cinnarizine.

As a preferred embodiment of the application of the present invention, the drugs for treating early-onset schizophrenia include Veliparib, Talazoparib, Niraparib, Fingolimod, E-2012, Rucaparib, Olaparib, Mexiletine, Nadroparin, Aldesleukin, Denileukin diftitox, Asparagine , Vinblastine, Auranofin, Rivanicline, Mesalazine, Dupilumab, Aspirin, Muromonab and Ribavirin.

As a preferred embodiment of the application of the present invention, the drugs for treating deficient schizophrenia include Tasonermin, A-674563, Castanospermine, Belantamab mafodotin, Bexarotene, lpilimumab, Thyrotropin alfa, Reversine, Puromycin, Anisomycin, Botulinum toxintype A , Glycolic acid, Pazopanib, Aminolevulinic acid, Alfacalcidol, Dimethylfumarate, Pexidartinib, Wortmannin, Bezafibrate, and Cediranib.

Compared with the prior art, the beneficial effects of the present invention are:

The invention constructs a drug-target protein-disease interaction network and collects candidate genes for schizophrenia, and uses a multidimensional interaction group to screen out new drugs that may be used for relieving or even treating schizophrenia. The method for constructing a drug-target protein-schizophrenia interaction network of the invention provides a basis for repositioning schizophrenia-related preventive or therapeutic drugs, and the constructed interaction network detects new clinical uses of existing drugs. It can speed up the drug development process and save a lot of manpower and financial resources.

Description of drawings

Figure 1 is a flow chart of the construction of a drug-target protein-disease interaction network;

Figure 2 is a network diagram of drug-protein-disease interaction;

Figure 3 is a schematic diagram of the diffusion curve of drugs and diseases and the SIM schematic diagram of the disease diffusion curve.

Detailed ways

Protein interactions play an important role in the cellular molecular signaling pathway network. The molecular mechanism of disease is studied through the protein interaction network of disease-causing genes. From the perspective of systems biology, disease risk genes in schizophrenia may play a role in A common molecular network, such a common molecular action network may involve multiple signaling pathways to perform related cellular functions, it is very important to understand the pathogenic mechanism of schizophrenia and identify therapeutic drugs through the schizophrenia network. However, with the deepening of understanding of pharmacology, the "multi-target, multi-drug" model has been widely accepted to replace the "one target, one drug" model. Drugs often target multiple proteins, not just one. Furthermore, in addition to the primary therapeutic target, drugs may also interact with other proteins, known as off-target effects. And a disease usually has more than one causative gene, but includes a few major genes and many minor genes.

Therefore, identifying drug-target interactions (DTIs) is an important prerequisite for related fields such as pharmacology, drug repositioning, drug discovery, side effect prediction, and drug resistance.

In order to better illustrate the purpose, technical solutions and advantages of the present invention, the present invention will be further described below with reference to specific embodiments. Those skilled in the art should understand that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

The test methods used in the examples are conventional methods unless otherwise specified; the materials, reagents, etc. used can be obtained from commercial sources unless otherwise specified.

Example 1: A method for constructing a drug-target protein-disease interaction network

Schizophrenia is a complex multifactorial disease. It seems impossible to treat all symptoms of the disease with a single target drug. A more ideal treatment should be based on the different expression patterns of different schizophrenia types. Different types of drugs and drug combinations. In view of the huge cost of time, money and resources in the development of new drugs. By exploiting known interaction data and unpaired small molecule compounds stored in various databases, new therapeutic effects and uses of existing or obsolete drugs for disease, i.e. drug repositioning, can be discovered. The process is shown in Figure 1.

Specific steps are as follows:

(1) Data collection

Drug-target protein interaction relationships were collected through drug databases (DrugBank, Drug Repurposing Hub), and human targets and FDA-compliant drugs (4,622 nodes, 11,959 edges) were screened, including 2,268 drugs and 2,354 target proteins.

Disease-causing gene interactions (27,092 nodes, 673,412 edges) were integrated by Cytoscape mapping diseases to their affected genes by effects such as genomic alterations, expression alterations, or post-translational modifications (27,092 nodes, 673,412 edges), which included 21,878 disease, 5,214 causative genes.

(2) Construction of multidimensional interactome

The three interaction relationships were integrated using Python to construct a drug-protein-disease interaction network, as shown in Figure 2.

(3) Calculate drug and disease diffusion curves

Drug or disease spread profiles learn which proteins are most affected by each drug or disease by spreading the effects of each drug and disease across multidimensional interactomes using restart random walks. Each drug or disease diffusion curve is computed by a biased random walk starting from the drug or disease node. At each step, the random walker can restart its walk or jump to adjacent nodes according to the optimized edge weights. After multiple walks, the diffusion curve measures how often each node is visited, thus representing the effect of a drug or disease on that node.

Weights W={w _drug , w _protein , w _indication } are assigned to different types of nodes, which is the relative possibility of jumping from one node type to another. α represents the probability that the walker continues walking at a given step instead of starting over.

First calculate the probability that the walker jumps to different types of nodes. Next, calculate the probability that the walker jumps to different nodes of the same type. Finally, the diffusion distribution is calculated by power iteration.

(4) Drug prediction

For a drug to treat a disease, it must affect a protein that is disrupted by the disease and has a similar biological function. Diffusion curves for drugs and diseases encode the effects of drugs and diseases on proteins.

Comparing the diffusion curves of drugs and diseases predicts which drugs will treat a given disease. For each drug, the L2 norm was used to calculate its SIM (similarity) to the disease diffusion curve, as shown in Figure 3, and the drugs most likely to treat the disease were sorted and summarized.

Example 2: A method for constructing a drug-target protein-schizophrenia interaction network

(1) Data collection

Drug-target protein interaction relationships were collected through drug databases (DrugBank, Drug Repurposing Hub), and human targets and FDA-compliant drugs (4,622 nodes, 11,959 edges) were screened, including 2,268 drugs and 2,354 target proteins.

Disease-causing gene interactions (27,092 nodes, 673,412 edges) were integrated by Cytoscape mapping diseases to their affected genes by effects such as genomic alterations, expression alterations, or post-translational modifications (27,092 nodes, 673,412 edges), which included 21,878 disease, 5,214 causative genes.

We collected 1,720 schizophrenia candidate genes, mapped them on the human protein interaction network (18,276 nodes, 817,086 edges) by Cytoscape and R, and screened out the interaction sub-network of schizophrenia pathogenic genes (1,501 nodes, 9,084 edges). And through sequencing, the differential genes of defective and early-onset schizophrenia were obtained.

The top 30 schizophrenia candidate pathogenic genes (n=1501) are shown in Table 1:

Table 1 Candidate causative genes for schizophrenia

The top 30 early-onset schizophrenia differentially expressed genes (n=1206) are shown in Table 2:

Table 2 Differentially expressed genes in early-onset schizophrenia

The top 30 deficient schizophrenia differentially expressed genes (n=1161) are shown in Table 3:

Table 3 Differentially expressed genes in deficient schizophrenia

(2) Construction of multidimensional interactome

The three interaction relationships were integrated using Python to construct a drug-protein-schizophrenia interaction network. In order to find a drug for treating different types of schizophrenia, schizophrenia and its candidate genes were added to the disease and pathogenic genes in the process of constructing a multi-dimensional interaction group, and the interaction relationship between different types of schizophrenia and its differential genes.

(3) Calculate drug and disease diffusion curves

Assign weights and restart probabilities according to the node type, restart the random walk from the disease and drug nodes, and calculate the influence of drugs and diseases on nodes in the multidimensional interaction group by the frequency of walking nodes.

Weights are given by different types of nodes W={'indication':3.541889556309463,'protein':4.396695660380823,'drug':3.2071696595616364}, α=0.8595436247434408.

First calculate the probability that the walker jumps to different types of nodes. Next, calculate the probability that the walker jumps to different nodes of the same type. Finally, the diffusion distribution is calculated by power iteration.

(4) Drug prediction

The SIM (similarity) of the diffusion curves of drugs and diseases and the diffusion curves of schizophrenia-related diseases were calculated by L2 norm, and the Euclidean distance between them was calculated by comparing the SIM of the diffusion curves of drugs and diseases. The predicted drugs are sorted for each given disease, and the top 20 drugs are screened out.

An integrated analysis of various predictive drugs for schizophrenia was carried out. The closer the drug is, the more schizophrenia types and the protein regulatory mechanisms of related diseases are associated with higher associations, and it is more likely to be used for the treatment of schizophrenia, Drugs are ranked and annotated by the calculated Euclidean distance, resulting in predicted drugs for relieving or even treating schizophrenia.

The top 20 drugs for the treatment of schizophrenia predicted by the multiscale action group, including the drug ID, name, and treatment disease in DrugBank, are shown in Table 4:

Table 4 Predicted top 20 drugs for schizophrenia

The protein interaction network was further replaced with the interaction network between differentially expressed genes of different types of schizophrenia, and drug prediction was performed for different types of schizophrenia. The result is as follows:

The top 20 drugs for the treatment of early-onset schizophrenia predicted by the multiscale action group, including the ID, name, and treatment disease of the drug in DrugBank, are shown in Table 5:

Table 5 Predicted top 20 drugs for early-onset schizophrenia

The top 20 drugs for treatment-deficient schizophrenia predicted by the multiscale action group, including the drug ID, name, and treatment disease in DrugBank, are shown in Table 6:

Table 6. Top 20 drugs predicted to treat deficient schizophrenia

Some of the above-predicted drugs are existing drugs that have been used to treat schizophrenia, which proves the feasibility of the construction method of the drug-target protein-schizophrenia interaction network.

The multidimensional interactome of the present invention provides a general approach to systematically understand how drugs treat diseases, systematically identify proteins associated with treatment, and predict which genes will alter drug efficacy or lead to serious adverse effects of drug treatment, multidimensional interactions Groups can be easily expanded to add other node types related to disease. Diffusion curves provide predictive power and interpretability in drug-disease treatment modeling, allowing identification of proteins relevant to treating a given disease.

The drug-target protein-schizophrenia interaction network of the present invention can start from the different expression patterns of different schizophrenia types, and find the drugs that affect the corresponding targets according to their differential expression genes, so as to play a role in different types of schizophrenia. to the therapeutic effect.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit the protection scope of the present invention. Although the present invention is described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that, The technical solutions of the present invention may be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims (10)

1. a construction method of drug-target protein-schizophrenia interaction network, is characterized in that, comprises the following steps: (1) Collect drug-target protein interaction relationships through drug databases, and screen out human targets and drugs that meet FDA standards; (2) Mapping the disease to schizophrenia-related genes through the disease database, and integrating the relationship between disease-causing genes; (3) Using Python to integrate the drug-target protein interaction relationship and the disease-pathogenic gene relationship to construct a drug-protein-schizophrenia interaction network. 2. the construction method of drug-target protein-schizophrenia interaction network according to claim 1, is characterized in that, in described step (2), collect schizophrenia candidate gene, and by sequencing, obtain defect Differential genes for schizophrenia and early-onset schizophrenia; the candidate genes, defective schizophrenia differential genes and early-onset schizophrenia differential genes are pathogenic genes; The disease gene was mapped on the human protein interaction network, and the schizophrenia-disease gene interaction relationship was screened out. 3. the construction method of drug-target protein-schizophrenia interaction network according to claim 2, is characterized in that, described schizophrenia candidate gene comprises ABL2, ALDOA, ARHGAP1, ABR, SLC25A6, ARNT, ACACA, APAF1, STS, ACVR2A, APBA2, ASCL1, ADRA1A, BIRC3, SERPINC1, JAG1, APOE, RERE, AGT, FAS, ATP2A2, AK4, AQP6, ATP2B4, AKT1, AR, ATRX, ALDH1A1, ARL4D, and KIF1A; the deficient Differential genes in schizophrenia include ABR, ALAD, ARHGAP1, ACLY, ALAS1, ARNT, ACO2, ABCD1, ARRB1, ACTG1, ALDH1A1, GET3, ACTN4, ALDH3B1, ATIC, ACVR1B, ALOX15B, ATOX1, ACVR2B, APLP2, TNFRSF17, ADCY7, APP, ZFP36L1, AP2A1, APRT, ZFP36L2, AKT1, ARF3, ARHGAP1; the early-onset schizophrenia differential genes include ABCF1, AQP3, POLR3D, ACVR18, ARHGAPS, BNIP1, PLIN2, ARHGDIA, KLF9, ADORA2A, ARHGDIB, C3AR1 , PARP1, ATF3, TMEM258, AGER, ATP5F1D, CFAP410, AHR, ATPSF1E, CBFA2T3, ALDOA, ATP6V1A, SLC25A6, ATP6V1C1, CCNE1, FAS, BMI1, CCNH. 4. Application of the drug-target protein-schizophrenia interaction network obtained by the construction method according to any one of claims 1 to 3 in screening drugs for preventing or treating schizophrenia. 5. The application according to claim 4, characterized in that, according to the node type in the drug-target protein-schizophrenia interaction network, weights and restart probability are given, and the random walk is restarted from the disease and drug nodes, Calculate the influence of drugs and diseases on nodes in the multidimensional interaction group through the frequency of wandering nodes to calculate drug and disease diffusion curves; respectively calculate the diffusion curves of the drugs and diseases and the schizophrenia-related disease diffusion curve SIM, through all The SIM compares and calculates the Euclidean distance between each pair, by which the predicted drug ranking for a given schizophrenia is performed. 6. The application according to claim 4, wherein the node is given a weight of W={'indication':3.541889556309463,'protein':4.396695660380823,'drug':3.2071696595616364}, and the walker continues at a given step Probability of walking instead of starting over α = 0.8595436247434408. 7 . The application according to claim 4 , wherein the diffusion curves of drugs and diseases and the diffusion curve SIM of schizophrenia-related diseases are calculated respectively by L2 norm. 8 . 8. application according to claim 4 is characterized in that, the described medicine for the treatment of schizophrenia comprises Rilonacept, Tetrabenazine, lsometheptene, Zuclopenthixol, Droperidol, Acetophenazine, lloperidone, Dopamine, Ketanserin, Enzastaurin, Vilazodone, Epicriptine, Dihydrexidine, Flupentixol, Taurine, Benzphetamine, Piceatannol, Citalopram, Butababital, and Cinnarizine. 9. application according to claim 4 is characterized in that, described medicine for the treatment of early-onset schizophrenia comprises Veliparib, Talazoparib, Niraparib, Fingolimod, E-2012, Rucaparib, Olaparib, Mexiletine, Nadroparin, Aldesleukin, Denileukin diftitox, Asparagine, Vinblastine, Auranofin, Rivanicline, Mesalazine, Dupilumab, Aspirin, Muromonab, and Ribavirin. 10. application according to claim 4, is characterized in that, the described medicine for the treatment of deficient schizophrenia comprises Tasonermin, A-674563, Castanospermine, Belantamab mafodotin, Bexarotene, lpilimumab, Thyrotropin alfa, Reversine, Puromycin, Anisomycin, Botulinum toxintype A, Glycolic acid, Pazopanib, Aminolevulinic acid, Alfacalcidol, Dimethylfumarate, Pexidartinib, Wortmannin, Bezafibrate, and Cediranib.

Images