WO2017211059A1 - 一种判别或比较药物作用模块的方法 - Google Patents

一种判别或比较药物作用模块的方法 Download PDF

Info

Publication number
WO2017211059A1
WO2017211059A1 PCT/CN2016/109895 CN2016109895W WO2017211059A1 WO 2017211059 A1 WO2017211059 A1 WO 2017211059A1 CN 2016109895 W CN2016109895 W CN 2016109895W WO 2017211059 A1 WO2017211059 A1 WO 2017211059A1
Authority
WO
WIPO (PCT)
Prior art keywords
module
network
drug
biological
biological network
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2016/109895
Other languages
English (en)
French (fr)
Inventor
王�忠
李兵
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Priority to EP16904507.7A priority Critical patent/EP3467689A4/en
Publication of WO2017211059A1 publication Critical patent/WO2017211059A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B5/00—ICT specially adapted for modelling or simulations in systems biology, e.g. gene-regulatory networks, protein interaction networks or metabolic networks
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B15/00—ICT specially adapted for analysing two-dimensional [2D] or three-dimensional [3D] molecular structures, e.g. structural or functional relations or structure alignment
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
    • G16B40/30—Unsupervised data analysis

Definitions

  • the invention belongs to the field of bioinformatics.
  • the present invention relates to methods for discriminating and comparing drug target modules based on biological networks.
  • the occurrence and development of many complex diseases are related to a series of interacting genes or proteins.
  • the disease phenotype is the manifestation of the interaction of different biological processes in a complex network.
  • the drug through acting on multiple targets in the disease network, produces a synergistic effect on the action of each target, thereby intervening in the occurrence and development of the disease, and finally achieving the therapeutic effect. Therefore, in the field of disease research and drug research, network analysis methods are increasingly being applied to identify biomarkers of complex diseases and targets of drugs.
  • the analysis of biological networks also provides new ideas for better understanding the pathological mechanism of complex diseases, systematically revealing the pharmacological mechanisms of drugs and the development of new drugs.
  • Modularity is an important feature of system biology networks.
  • modular analysis of networks through various computational models and algorithms, deconstruction and analysis of networks at the module level has become a new research direction in network pharmacology and systems biology.
  • Module refers to a tightly connected functional group in a biological network.
  • the related literature also refers to clusters, subnetworks, communities, subsets, motifs, and the like.
  • the module has tightly connected topological features in the biological network, as well as stability and functionality.
  • researchers have developed a large number of module partitioning methods based on the principles of network clustering, heuristic search, seed expansion, network topology, matrix decomposition, etc. Modular research has become a hotspot in biological network research.
  • the invention provides a method for discriminating and comparing a drug action module based on a biological network.
  • the method integrates the topology analysis and statistical analysis of the module to comprehensively identify a function module (also referred to as a “response module”) of the biological network after the drug intervention.
  • the combined biological function analysis can then identify whether it is a target module of the drug; in addition, the method of the present invention can also compare and analyze the difference in response of different drugs by comparing the difference modules of the biological networks after different drug interventions.
  • biological network refers to a network composed of nodes and edges, represented by G(V, E), V represents a collection of nodes in the network, and nodes may be genes, proteins, compounds, metabolites, etc.
  • the target point; E represents the set of edges in the network, and the edge is the interaction relationship between the nodes, for example, it may be a co-expression or interaction relationship of genes, a protein interaction relationship, a transcriptional regulatory relationship, and the like. Interaction relationship.
  • Common biological networks include gene co-expression networks, protein interaction networks, metabolic networks, gene transcriptional regulatory networks, non-coding gene regulatory networks, and various cross-omics regulation and interaction networks.
  • the gene co-expression network is constructed based on the correlation of gene expression data between genes.
  • the graph model is often used to describe the relationship between genes.
  • the nodes in the graph represent genes and the edges represent Co-expression interaction between the two genes.
  • module refers to a closely connected functional group consisting of nodes and edges in a biological network, which has the same meaning as “subset”, “cluster”, “social”, “subnet” in a biological network, “Phantom” and so on.
  • module of action of a drug refers to a module that exhibits a significant change in topology parameters as compared to a corresponding module in a biological network prior to the action of a drug, which can be used interchangeably with the term "response module for a drug.”
  • target module of a drug refers to a module that functionally enriches a module of action of a drug to determine that it does have a relevant function.
  • drug difference module refers to a module that compares the corresponding modules in a biological network obtained after the action of the two drugs, and the topology parameters change significantly.
  • the term "conservative module of a drug” as used herein refers to a module that compares the corresponding modules in a biological network obtained after the action of the two drugs, and the topology parameters are not significantly changed.
  • the present invention provides a method of discriminating or comparing a drug action module, the method comprising the steps of:
  • the module is the function module of the drug; when the first biological network G i (V i , E i ) is the biological network after the first drug intervention, and when the second biological network G j (V j , E j ) is the biological network after the second drug intervention, if ⁇ t/T ⁇ significant threshold W, Then the module is a function difference module of the first drug and the second drug.
  • the method of the present invention can be used to determine whether the action module of the drug is a target module of the drug, that is, the method further comprises:
  • the drug, the first drug, and the second drug are the same type of traditional Chinese medicine or chemical agent for preventing and treating diseases.
  • the first drug and the second drug are medicinal components of the same type of Chinese medicine or chemical.
  • the V i or V j in the first biological network and the second biological network is a set of nodes in the network, and E i or E j is a set of edges in the network.
  • the first biological network and the second biological network are biomolecular networks, for example selected from the group consisting of a gene co-expression network, a protein interaction network, a gene transcription regulatory network, a non-coding gene regulatory network, and a metabolic network. More preferably, the first biological network and the second biological network are the same type of gene co-expression network or protein interaction network; more preferably, the number of nodes of the biomolecular network is ⁇ 200, ⁇ 500, ⁇ 1000 or ⁇ 5000.
  • the module is divided by a network analysis algorithm selected from the group consisting of clustering, heuristic search, seed expansion, matrix decomposition, topology network, etc.; more preferably, the method is selected from the group consisting of hierarchical clustering algorithm, WGCNA, MCODE, MCL, CFinder, CPM, SPC, GN algorithm, ModuLand, DME, MINE, SVD methods are used for module partitioning.
  • a network analysis algorithm selected from the group consisting of clustering, heuristic search, seed expansion, matrix decomposition, topology network, etc.
  • the method is selected from the group consisting of hierarchical clustering algorithm, WGCNA, MCODE, MCL, CFinder, CPM, SPC, GN algorithm, ModuLand, DME, MINE, SVD methods are used for module partitioning.
  • the topological feature M(t) is selected from the group consisting of modularity, degree, connectivity, density, clustering coefficient (clustering coefficient) ), between-betweenness, network diameter, shortest path, entropy, and characteristic path length.
  • the number of topological features m is 1-10, for example any integer number of 1-10.
  • the number of topological features m is 2, the topological feature M(t) comprises density and connectivity; or preferably, the number of topological features m is 3, and the topological feature M(t) comprises density, connectivity and entropy .
  • calculating the topology parameter change ⁇ t/T comprises:
  • step (3) of the method of the present invention using a permutation test or a bootstrap resampling program to calculate a significant boundary value W of the module topology parameter change;
  • the significance threshold W is obtained by:
  • the significance threshold W is equal to the normalized ⁇ t/T value corresponding to p ⁇ 0.05 in the probability of normal distribution.
  • step (4) of the method of the invention the functional module of the drug is subjected to biofunctional enrichment analysis to determine if it is a target module for the drug.
  • biofunctional enrichment analysis there are many function enrichment analysis tools of the module, and the suitable tools can be selected according to the specific conditions of the module.
  • the function enrichment result of the module it can be judged whether the action module of the drug has the corresponding biological function.
  • biological function enrichment analysis can be performed by gene ontology (GO) function and/or pathway analysis.
  • the p-values of the function enrichment are generally calculated from the hypergeometric distribution.
  • the functional enrichment of GO is the hypergeometric distribution relationship of a certain (several) specific branch of the computational module gene in the GO classification.
  • the GO function of the return module enriches the p value, and the small p value indicates that the module is in the GO. Enrichment has occurred in it.
  • pathway enrichment analysis is the hypergeometric distribution of the genes of the computational module in the same Pathway.
  • the return module has an enriched p-value in the pathway. A small p-value indicates that the module appears in the pathway. Enrichment.
  • FIG. 1 A schematic of the method of the invention is shown in Figure 1.
  • the invention provides a method for discriminating and comparing a target target module of a drug by using a topology parameter, a statistical and a functional analysis method of the network, which can effectively find a function module of the drug and compare the similarities and differences of the action modules of different drugs. It provides a new strategy for drug development models with modules as drug targets.
  • the method of the present invention utilizes the topology change of the module in the network before and after the specific drug intervention, and compares and analyzes whether the module is a function module of the drug, and then through functional enrichment analysis, Select the target module for the drug.
  • the method of the present invention can also utilize the similarities and differences of the modules in the network after different drug interventions, and compare the different functional modules of different drugs through comparative analysis to determine the similarities and differences of the drugs.
  • the method of the present invention reflects the mechanism of action of a drug by observing changes in the topology of the module before and after drug intervention (rather than changes in individual gene and protein expression levels).
  • FIG. 1 is a flow chart showing a method for identifying a drug target module according to the present invention
  • Figure 3 shows the gene co-expression module of the BA group in Example 1;
  • Figure 5 shows the difference module obtained by the BA group in the first embodiment compared with the JA group (W>0.2);
  • Figure 6 shows an example of the 16th module of the BA group in Example 1 and the enriched biological function
  • FIG. 7 shows an example of a target module of the BA group in Embodiment 1;
  • FIG. 8 shows an example of a characteristic module of the BA group in Embodiment 1;
  • Figure 9 shows the interaction relationship of each node in the BA drug target module identified in Example 2 in a known database.
  • Fig. 10 shows the expression of genes in the drug target module discriminated in Example 2.
  • Fig. 11 is a view showing the interaction relationship of genes in the drug target module discriminating in Example 2.
  • Fig. 12 shows the result of module division of the effect group in the third embodiment.
  • Figure 13 shows the response module of anti-PD-1 therapeutic melanoma in the effect group of Example 3 compared to the control group.
  • Fig. 14 shows the result of module division of the effect group in the fourth embodiment.
  • Figure 15 shows the response module of infliximab for treatment of RA in the effect group of Example 4 compared to the control group.
  • the gene expression profile data of the cerebral ischemia model of mice treated with the effective components of refined Qingkailing (BA), geniposide (JA) and cholic acid (UA) were used as examples to construct pre-intervention and intervention of different drugs.
  • the post-gene co-expression network, and the implementation of the method is specifically illustrated by the data of the BA group.
  • mice cerebral ischemia model was treated with the effective components of refined Qingkailing (BA), geniposide (JA) and cholic acid (UA).
  • mice MCAO model was prepared by Zea-Longa suture method, slightly improved, and the model of mouse cerebral ischemia for 1.5 h and reperfusion for 24 h was prepared.
  • the mice were fasted for 12 hours before surgery, and they were given free access to water.
  • the mice were anesthetized with 10% chloral hydrate solution.
  • the mice were anesthetized by intraperitoneal injection at a dose of 0.4 ml/100 g body weight.
  • the left common carotid artery, the internal and external carotid arteries were exposed, and the common carotid artery and external carotid artery were carefully removed with forceps to separate the vagus nerve.
  • the common carotid artery was clamped with an artery clamp to temporarily block the blood flow of the common carotid artery and the internal carotid artery.
  • the sterilized suture was ligated at the distal end of the external carotid artery, and the external carotid artery was cut to ensure that the length of the external carotid artery stump was longer than 0.5 cm.
  • the arterial clip is placed at the distal end, and the common carotid artery bifurcation is incision.
  • the artery clip is loosened, and the plug line is inserted into the internal carotid artery for 18-20 mm.
  • the middle cerebral artery is blocked at the beginning of the middle cerebral artery. Then, the common carotid artery was ligated together with the fishing line, and a 10 mm thread was left to suture the skin. When the cerebral ischemia is 1.5h, the plug is withdrawn, and the blood flow is recanalized. After 24 hours, the neurological function score of the mice was observed to judge whether the modeling was successful.
  • the constructed cerebral ischemia model mice were randomly divided into 5 groups, including sham group, vehicle group, BA group, JA group and UA group. After 2 hours of ischemia, the drug group was treated with 2 ml of drug solution/kg body weight. BA (5 mg/ml), UA (7 mg/ml) and JA (25 mg/ml) were injected. Mice in the Sham group and the vehicle group were injected with 0.9% NaCl (2 ml/kg body weight) via the tail vein. The mouse hippocampal RNA was extracted in a single step and the RNA quality was evaluated using the Bioanalyzer gene chip. The gene expression profile of the mice after intervention was analyzed using a mouse Oligo fragment chip.
  • the chip used consisted of 374 c-DNAs associated with cerebral ischemia.
  • the gene expression profile data used in this example was divided into four groups, namely, a model group (Vehicle), a baicalin treatment group (BA), a geniposide treatment group (JA), and a cholic acid treatment group (UA).
  • the model group gene expression profile data was used as data before drug intervention, and the gene expression profile data of BA group, JA group and UA group were used as data after drug intervention.
  • Each group of gene expression profile data consists of 374 genes from 12 samples (Tbp, Zeb1, Pou2f1, Foxb1, Creb1, Camk2g, Csf1, F5, Hspd1, Matn2, Mt1, Adamts1, Klf6, Dffa, Rgs18, Rhoa, Kcnmb1) , Pdcd11, Pdpk1, Casp8ap2, Mogat1, Rps26, Ak1, Csnk2a2, Dkk2, Ppm1e, Tnfrsf22, Trp53i11, Smpd3, Grin1, Cdk5, Jund, E2f1, Apoe, Ilb, Prkar1b, Il7r, Ngfb, Rela, Ifnar1, Adcy6, Bak1 , Fzd6, Prkch, Rgs4, Actg1,
  • the value in each cell is the expression value of the gene (row) in the sample (column).
  • genes with the same expression pattern can be clustered together in this expression spectrum.
  • Co-expressing genes, these co-expressed genes are clustered into "gene co-expression modules.”
  • the gene co-expression network (the number of nodes is 374) of three drug group data was constructed by weighted co-expression network analysis (WGCNA) tool and divided into modules. After the module identifies the BA group, a total of 23 modules are obtained. The size of the module is between 3-149 nodes (genes); the JA group gets 42 modules, the module size is between 3-46 nodes; the UA group gets 15 modules. The module size is between 3-29 nodes.
  • WGCNA weighted co-expression network analysis
  • Fig. 2 the expression of all genes in the whole network is shown in Fig. 2, and the existence of obvious co-expression modules can be seen.
  • the modules of the BA group are obtained based on the hierarchical clustering algorithm, and each color represents a module, and the result is shown in FIG.
  • the permutation test was used to test the topological changes of the two modules.
  • the 23 modules of the BA group were randomly replaced 1000 times.
  • the normalized ⁇ t/T of each module was calculated and standardized.
  • the statistical p value is ⁇ 0.05
  • the corresponding normalized ⁇ t/T value minimum value that is, the boundary value of the module topological parameter significant change after drug intervention.
  • W 0.2 is the critical value of the module's significant change.
  • the module with ⁇ t/T>0.2 in the BA intervention group can be considered as the response module of the drug.
  • the method of the invention can also be used for comparison between different drug groups, the module with ⁇ t/T>0.2 being the difference module between the two groups. Comparing the BA group with the JA group, nine difference modules can be found, and the results are shown in Fig. 5.
  • Functional enrichment analysis was performed on the response modules of the three drug groups using functional enrichment analysis tools ( tools such as DAVIDE), and the significance threshold was analyzed by p ⁇ 0.05.
  • the BA-16 module is enriched for the most functions, including 41 GO functions and 9 KEGG paths, and the BA-6 module is not enriched to any function.
  • the biological functions enriched in the BA-16 module are as follows:
  • the BA-16 module is involved in biological functions such as cell signaling, placental development, and phosphorus metabolism, and the BA-16 module is associated with colon cancer and MAPK signaling pathways, so it is possible to infer the mechanism by which BA intervenes in the BA-16 response module.
  • the above functions are related.
  • Figure 6 shows the BA-16 module and its enriched functionality.
  • the obtained response modules of the three drug groups of BA, JA, and UA were all enriched in biological functions. It was found that compared with the model group, the BA group obtained 11 response modules, 3 of which failed to enrich any biological functions, and the remaining 8 modules were BA target modules; compared with the model group, JA The group obtained 22 response modules, 3 of which failed to enrich any biological functions, and the remaining 19 modules were JA target modules; compared with the model group, the UA group received 8 response modules, of which 1 The modules failed to enrich any biological functions, and the remaining 7 modules were UA target modules.
  • An example of the BA module's target module (BA-5, BA-15) is shown in Figure 7.
  • BA, JA, and UA are compared between two groups. If a group of modules is different from the other two groups, the module is a characteristic module of the group. After comparison, BA has five characteristic modules.
  • An example of the characteristic module (BA-9, BA-21) is shown in Figure 8; JA has 8 characteristic modules, and the UA has 2 characteristic modules.
  • Module-related biological functions can be obtained through functional enrichment analysis of the module, which may be the mechanism of action of the drug.
  • Example 2 verifies the target module of the identified drug
  • the String (functional protein association networks) database can be used to find various interactions of genes (proteins) within known modules.
  • the BA-16 module inputs the nodes in the module into the database to obtain the interaction relationship of the nodes. See Figure 9.
  • the experimental method was used to confirm the compositional significance of the target module by the amount of gene expression.
  • the target module genes of each treatment group were verified by Western Blot.
  • the expression levels of VEGF and B230120H23Rik were significantly different from those of the Vehicle group, see Figures 10A and 10B, respectively.
  • This example applies the method of the present invention to mRNA expression data of anti-PD-1 therapeutic melanoma for discriminating a target module for anti-PD-1 treatment of melanoma.
  • the cases in which anti-PD-1 treatment was produced were divided into one group (effect group), and the cases in which no reaction occurred were divided into one group (control group), and the gene co-expression network of the two groups was constructed by the WGCNA method.
  • the data used in this example was downloaded from the GEO (Gene Expression Omnibus) database. (https://www.ncbi.nlm.nih.gov/geo/, data ID is GSE78220), the data set is mRNA expression data for anti-PD-1 treatment of melanoma, the sample source is human, a total of 28 samples, of which Fifteen samples were the effect group and 13 samples were the control group. The total amount of mRNA contained in the data was 25,268.
  • the specific implementation process is as follows:
  • the downloaded data is organized into the data table format required for the WGCNA analysis.
  • the table is listed as mRNA, the behavioral therapy sample, and the corresponding cell value is the expression value of the mRNA (row) in the sample (column), based on Co-expression of mRNA, network construction and module partitioning.
  • the WLCNA tool was also used to construct the mRNA co-expression network of the effector group and to partition the module. During the network construction process, 3,080 mRNAs were removed due to excessive missing values, and the remaining 22,188 mRNAs were used to construct the network (the number of nodes in the constructed network was 2,1976) and module partitioning. A total of 30 modules were obtained in the effector group, with modules ranging in size from 38-2900 nodes (mRNA). The module division results obtained by the WGCNA identification group are shown in Fig. 12.
  • the topological structural changes of the effect group and the control group were calculated. Select the density and connectivity as the topological indicators of the module, and calculate the density and connectivity of the effect and control modules. The effect group was compared with the control group, and the changes of the two groups of modules in the density and connectivity index were observed.
  • the topology change of the two modules was tested by the bootstrap resampling.
  • the module components of the effect group were resampled 10,000 times.
  • the normalized ⁇ t/T of each module was calculated.
  • the normal distribution probability of the normalized ⁇ t/T in 10,000 resampling processes is obtained.
  • the statistical p value is ⁇ 0.05
  • the corresponding normalized ⁇ t/T value minimum value that is, the module topological parameter significant change threshold value.
  • W 0.26 is the critical value of the module significant change.
  • the module with ⁇ t/T>0.26 in the anti-PD-1 effect group can be considered as a response module of the drug.
  • the function enrichment analysis tool (DAVIDE 6.7 tool) was used to perform functional enrichment analysis of GO function and KEGG pathway in the response module of the effect group, and the significance threshold was analyzed by p ⁇ 0.05.
  • the corresponding function the target module for anti-PD-1 treatment of melanoma, was enriched in all eight response modules in the anti-PD-1 therapeutic effect group.
  • the present invention was applied to the mRNA expression data of infliximab in the treatment of RA, and three topological parameters (density, connectivity, entropy) were used to discriminate the target module of infliximab in the treatment of RA.
  • the cases were divided into control group and treatment group before and after treatment with infliximab.
  • the gene co-expression network of treatment group and control group was constructed by WGCNA method.
  • the data used in this example was downloaded from the GEO (Gene Expression Omnibus) database (https://www.ncbi.nlm.nih.gov/geo/, data ID GSE8350), which is the treatment of RA with infliximab.
  • the mRNA expression data was obtained from human samples. In this example, 36 samples were selected, of which 18 were control group and 18 were treatment group.
  • the specific implementation process is as follows:
  • the WLCNA tool was also used to construct the mRNA co-expression network of the effector group and to partition the module.
  • a total of 776 mRNAs were used to construct the network (the number of nodes in the constructed network was 776) and module partitioning.
  • a total of 46 modules were obtained in the treatment group, and the size of the module was between 3-73 nodes (mRNA).
  • the module division results obtained by the WGCNA identification group are shown in Fig. 14.
  • the topological structural changes of the effect group and the control group were calculated. Density, connectivity, and entropy were selected as the topological indicators of the module, and the density, connectivity, and entropy of the effect and control modules were calculated. The treatment group was compared with the control group, and the changes of the two groups of modules in density, connectivity, and entropy were observed.
  • the Permutation test to make significant changes to the topology of the two modules
  • the test performed 1000 replacements on the module of the treatment group, and compared with the control group, the normalized ⁇ t/T of each module was calculated, and then the normal distribution probability of the normalized ⁇ t/T during 1000 replacements was obtained, when the statistical p value was obtained.
  • ⁇ 0.05 the corresponding standardized ⁇ t/T value minimum value, that is, the module topology parameter significant change threshold value.
  • W 0.15 is the critical value of the module significant change.
  • the module with ⁇ t/T>0.15 in the infliximab treatment group was considered to be the response module of the drug.
  • the function enrichment analysis tool (DAVIDE 6.7 tool) was used to perform functional enrichment analysis of GO function and KEGG pathway in the response module of the effect group, and the significance threshold was analyzed by p ⁇ 0.05.
  • DAVIDE 6.7 tool The function enrichment analysis tool was used to perform functional enrichment analysis of GO function and KEGG pathway in the response module of the effect group, and the significance threshold was analyzed by p ⁇ 0.05.
  • 2 were not enriched for any biological function, and the remaining 12 modules were the target modules for infliximab treatment of RA.
  • the application of the refined protein component of baicalin (BA) in the mouse model of cerebral ischemia was used as an example.
  • the differentially expressed genes in the cerebral ischemia model of BA intervention mice were obtained, and the network mapping of differentially expressed genes was performed based on the String database to construct a protein interaction network of BA intervention mouse cerebral ischemia model.
  • Methods The target module of BA intervention in mouse cerebral ischemia protein interaction network was identified.
  • the module was divided by MCODE method, and a total of 49 modules were obtained.
  • the permutation test was used to test the topological changes of the two modules.
  • the 49 modules of the BA group were randomly replaced 1000 times.
  • the normalized ⁇ t/T of each module was calculated and standardized.
  • the statistical p value is ⁇ 0.05
  • the corresponding normalized ⁇ t/T value minimum value that is, the boundary value of the module topological parameter significant change after drug intervention.
  • W 0.11 is the critical value of the module significant change.
  • the module with ⁇ t/T>0.11 in the BA intervention group can be considered as the response module of the drug.
  • the present invention can effectively identify the target module of the drug in the biological network, and can also obtain the conservative module and the difference module of different drugs by comparing between different drug groups.

Landscapes

  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Medical Informatics (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Biophysics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Theoretical Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Biotechnology (AREA)
  • Physiology (AREA)
  • Molecular Biology (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioethics (AREA)
  • Software Systems (AREA)
  • Public Health (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Epidemiology (AREA)
  • Databases & Information Systems (AREA)
  • Chemical & Material Sciences (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)

Abstract

提供一种判别或比较药物作用模块的方法，所述方法包括以下步骤：(1)对两个生物网络进行模块的划分，分别得到模块集合；(2)计算两个模块集合中各个模块的一个或多个拓扑结构特征，得到单一或综合的拓扑指标，并由此计算两个模块集合中对应模块的拓扑参数变化；(3)将对应模块的拓扑参数变化与显著性界值进行比较，当两个生物网络分别为药物干预前和干预后的生物网络时，如果拓扑参数变化≥显著性界值，则所述模块为药物的作用模块；当两个生物网络分别为第一药物干预后和第二药物干预后的生物网络时，如果拓扑参数变化≥显著性界值，则所述模块为第一药物和第二药物的作用差异模块。

Description

一种判别或比较药物作用模块的方法 技术领域
本发明属于生物信息技术领域。具体而言，本发明涉及基于生物网络判别和比较药物靶点模块的方法。
背景技术
系统生物学和网络药理学的兴起，使药物机制研究及新药研发从传统的“单组分、单靶点、单疾病”研究模式向“多组分、多靶点、多途径”的方向发展。整合基因组、转录组、蛋白质组和代谢组等高通量数据，结合各种数学模型和算法，在网络的背景下研究疾病、药物、靶点数据及其相互作用关系也成为疾病和药物作用机制研究的热点。
许多复杂疾病的发生、发展都与一系列相互作用的基因或蛋白相关，疾病表型是不同的生物学过程在一个复杂网络中相互作用的表现。而药物则是通过作用于疾病网络中的多个靶点，对各靶点的作用产生协同效应，从而对疾病的发生、发展进行干预，最终达到治疗效果。因此，在疾病研究和药物研究领域，网络分析方法越来越多地被应用于识别复杂疾病的生物标志物及药物的作用靶点等。生物网络的分析也为更好的理解复杂疾病的病理机制、系统揭示药物的药理机制及新药的研发提供了新的思路。
模块性(modularity)是系统生物网络的一个重要特性。在系统生物网络的背景下，通过各种计算模型和算法对网络进行模块化分析，在模块的层次下对网络进行解构和分析成为网络药理学和系统生物学新的研究方向。模块指生物网络中紧密连接的功能集团，相关文献中亦表示为聚类(cluster)、亚网络(subnetwork)、社团结构(community)，子集(subset)、模体(motif)等。模块在生物网络中具有紧密连接的拓扑结构特征，同时也具有稳定性和功能性的特点。研究者基于网络聚类、启发式搜索、种子扩展、网络拓扑结构、矩阵分解等原理发展了大量的模块划分方法，模块化研究成为生物网络研究的热点。
疾病病理和药物作用机制同样具有模块化的特征。“单组分、单靶点”的药物研究模式忽视了靶点基因或蛋白之间的相互作用关系，因此，一些研究尝试从模块的角度来筛选具有特异的靶点和较低副作用的药物。模块药理学(Modular pharmacology，MP)的概念框架也被提出，从模块化角度揭示药物与疾病间关系，提出通过药物的多靶点、模块化设计来治疗复杂疾病，这种以模块而不是单个基因或蛋白为靶点的方法为研究药物作用机制及新药研发提供了新的策略，而如何在复杂的生物网络中识别药物的靶点模块就成为一种挑战。因此，结合生物网络模块的拓扑结构、稳定性和功能性特点，探寻药物靶点模块识别方法具有重要的意义。
发明内容
本发明提供了一种基于生物网络判别和比较药物作用模块的方法，本方法整合了模块的拓扑结构分析、统计分析来综合判别药物干预后生物网络的作用模块(又称“应答模块”)，组合生物功能分析则可以识别其是否为药物的靶点模块；此外，本发明的方法还可以通过比较不同药物干预后的生物网络的差异模块，来对比分析不同药物的应答差异。
定义
本文所用术语“生物网络”是指由节点与边组成的网络，由G(V，E)表示，V表示网络中的节点的集合，节点可以是基因、蛋白、化合物、代谢物等药物可能作用的靶点；E表示网络中的边的集合，边是节点之间的相互作用关系，例如可以是基因的共表达或相互作用关系、蛋白的相互作用关系、转录调节关系等靶点间可能的相互作用关系。常见的生物网络包括基因共表达网络、蛋白质相互作用网络、代谢网络、基因转录调控网络、非编码基因调控网络及各种跨组学的调控及相互作用网络等。以基因共表达网络为例，基因共表达网络是以基因之间表达谱数据的相关性为基础而构建的，常使用图模型来描述基因之间的关系，图中的节点代表基因，边表示两个基因之间的共表达相互作用关系。
本文所用术语“模块”是指生物网络中由节点和边组成的紧密连接的功能集团，其涵义等同于生物网络中的“子集”、“聚类”、“社团”、“子网”、 “模体”等。
本文所用术语“药物的作用模块”是指与药物作用之前的生物网络中对应模块相比较，拓扑结构参数发生显著性变化的模块，其可与术语“药物的应答模块”互换使用。
本文所用术语“药物的靶点模块”是指将药物的作用模块经功能富集而确定确实具有相关功能的模块。
本文所用术语“药物的差异模块”是指将两种药物作用之后获得的生物网络中对应模块相比较，拓扑结构参数发生显著性变化的模块。
本文所用术语“药物的保守模块”是指将两种药物作用之后获得的生物网络中对应模块相比较，拓扑结构参数未发生显著性变化的模块。
本发明的具体技术方案如下：
本发明提供一种判别或比较药物作用模块的方法，所述方法包括以下步骤：
(1)对第一生物网络Gi(Vi，Ei)和第二生物网络Gj(Vj，Ej)进行模块的划分，分别得到模块集合Gi(Mi)和Gj(Mj)；其中第一生物网络Gi(Vi，Ei)为药物干预前的生物网络，而第二生物网络Gj(Vj，Ej)为药物干预后的生物网络；或者第一生物网络Gi(Vi，Ei)为第一药物干预后的生物网络，而第二生物网络Gj(Vj，Ej)为第二药物干预后的生物网络；
(2)计算模块集合Gi(Mi)和Gj(Mj)中各个模块的一个或多个拓扑结构特征M(t)，得到单一或综合的拓扑指标T＝M(t1，t2...tm)，m为拓扑结构特征数目，并由此计算两个模块集合中对应模块的拓扑参数变化Δt/T；
(3)将对应模块的拓扑参数变化Δt/T与显著性界值W进行比较，当第一生物网络Gi(Vi，Ei)为药物干预前的生物网络，而第二生物网络Gj(Vj，Ej)为药物干预后的生物网络时，如果Δt/T≥显著性界值W，则所述模块为所述药物的作用模块；当第一生物网络Gi(Vi，Ei)为第一药物干预后的生物网络，而第二生物网络Gj(Vj，Ej)为第二药物干预后的生物网络时，如果Δt/T≥显著性界值W，则所述模块为第一药物和第二药物的作用差异模块。
优选地，本发明的方法可以用于判别所述药物的作用模块是否是药物的靶标模块，即该方法还包括：
(4)将所述药物的作用模块进行生物功能富集分析，根据功能富集结果判断所述模块是否是所述药物的靶点模块。
在本发明方法的步骤(1)中，所述药物、第一药物、第二药物为同类型的用以预防和治疗疾病的中药或化学药。优选地，所述第一药物、第二药物为同类型的中药或化学药的药效组分。
所述第一生物网络和第二生物网络中的Vi或Vj为网络中节点的集合，Ei或Ej为网络中边的集合。优选地，所述第一生物网络和第二生物网络为生物分子网络，例如选自基因共表达网络、蛋白质相互作用网络、基因转录调控网络、非编码基因调控网络及代谢网络。更优选地，所述第一生物网络和第二生物网络为相同类型的基因共表达网络或蛋白质相互作用网络；更优选地，所述生物分子网络的节点数目≥200、≥500、≥1000或≥5000。
优选地，采用选自聚类、启发式搜索、种子扩展、矩阵分解、拓扑网络等网络分析算法进行模块划分；更优选地，采用选自分层聚类算法、WGCNA、MCODE、MCL、CFinder、CPM、SPC、G-N algorithm、ModuLand、DME、MINE、SVD的方法进行模块划分。
在本发明方法的步骤(2)中，所述拓扑结构特征M(t)选自模块性(modularity)、度(degree)、连接度(connectivity)、密度(density)、聚类系数(clustering coefficient)、介数中心性(betweenness)、网络直径(diameter)、最短路径(shortest path)、熵(entropy)和特征路径长度(characteristic path length)。
优选地，拓扑结构特征数目m为1-10，例如可以是1-10中的任何整数个。
优选地，拓扑结构特征数目m为2，拓扑结构特征M(t)包括密度和连接度；或者优选地，拓扑结构特征数目m为3，拓扑结构特征M(t)包括密度、连接度和熵。
优选地，计算拓扑参数变化Δt/T包括：
计算第一生物网络中每个模块的拓扑参数与第二生物网络中相应模块的拓扑参数相比的变化量，即各个拓扑参数的Δt/T值，采用离差标准化方法将各个拓扑参数的Δt/T进行标准化处理，经过离差标准化后，各个拓扑参数 的Δt/T应在(0，1)之间。
在本发明方法的步骤(3)中，利用置换检验(permutation test)或自助重采样(bootstrap resampling)程序计算模块拓扑参数变化的显著性界值W；
优选地，显著性界值W通过以下方式获得：
利用置换检验或自助重采样方法得到大于1000个与第一生物网络中每个模块对应的随机模块，计算每个随机模块与第二网络相应模块的标准化Δt/T值，进而计算所有随机模块的标准化Δt/T值的正态分布概率，显著性界值W等于所述正态分布概率中p≤0.05时对应的标准化Δt/T值。
在本发明方法的步骤(4)中，对药物的作用模块进行生物功能富集分析，以判断其是否是药物的靶点模块。模块的功能富集分析工具较多，可根据模块具体情况选择适合的工具。根据模块的功能富集结果可以判断药物的作用模块是否具有相应的生物学功能。例如，可以通过基因本体(GO)功能和/或通路(pathway)分析进行生物学功能富集分析。一般根据超几何分布计算功能富集的p值。GO的功能富集是计算模块组成基因在GO分类中某一(几)个特定的分支的超几何分布关系，返回模块的GO功能富集p值，小的p值则表示这个模块在该GO中出现了富集。同样，通路(pathway)富集分析是计算模块组成基因在同一条Pathway中的超几何分布关系，返回模块在该通路的富集p值，小的p值则表示这个模块在该通路中出现了富集。
本发明的方法的示意图见图1。
在生物网络中，对于判别与比较药物的作用模块一直缺少有效的方法，如何判断药物干预前后生物网络中模块的变化及其显著性一直存在挑战。本发明提出了一种利用网络的拓扑参数、统计和功能分析方法来判别与比较药物的作用靶点模块的方法，该方法能够有效地发现药物的作用模块，比较不同药物的作用模块的异同，为以模块为药物靶点的药物研发模式提供了新的策略。
具体地，药物干预后必然会对生物网络产生影响，而怎么判断哪些模块对药物产生应答，也就是怎样选择药物的靶点模块就是其中的关键问题。模块本身也是由节点和边组成的小网络，所以模块本身具有网络的拓扑结构特征，如模块会有特定的密度、连接度、最短路径、特征路径长度、模块性等。 因此，一方面，本发明的方法利用了在特定药物干预前和干预后的网络中模块的拓扑结构变化，通过对比分析来判断该模块是否是药物的作用模块，进而通过功能富集分析，可以从中选择药物的靶点模块。另一方面，本发明的方法还可以利用在不同药物干预后的网络中模块的拓扑结构异同，通过对比分析来判断不同药物的不同作用模块，从而判断药物的作用异同。由此，本发明的方法是通过观察药物干预前后模块的拓扑结构变化(而不是单个基因、蛋白表达水平的变化)来反应药物的作用机制。
附图说明
以下，结合附图来详细说明本发明的实施方案，其中：
图1示出本发明所述药物靶点模块识别方法的流程图；
图2示出实施例1中BA组数据的基因共表达模式；
图3示出实施例1中BA组的基因共表达模块；
图4示出实施例1中BA组与模型组相比得到的应答模块(W＞0.2)；
图5示出实施例1中BA组与JA组相比得到的差异模块(W＞0.2)；
图6示出实施例1中BA组的16号模块示例及富集到的生物学功能；
图7示出实施例1中BA组的靶点模块示例；
图8示出实施例1中BA组的特征性模块示例；
图9示出实施例2中判别出的BA药物靶点模块中各节点在已知数据库中的相互作用关系。
图10示出实施例2中判别出的药物靶点模块内基因的表达情况。
图11示出实施例2中判别出的药物靶点模块内基因的相互作用关系情况。
图12示出实施例3中效应组的模块划分结果。
图13示出实施例3中效应组与对照组相比的抗PD-1治疗黑色素瘤的应答模块。
图14示出实施例4中效应组的模块划分结果。
图15示出实施例4中效应组与对照组相比的英夫利昔单抗治疗RA的应答模块。
实施发明的最佳方式
以下参照具体的实施例来说明本发明。本领域技术人员能够理解，这些实施例仅用于说明本发明，其不以任何方式限制本发明的范围。
下述实施例中的实验方法，如无特殊说明，均为常规方法。下述实施例中所用的药材原料、试剂材料等，如无特殊说明，均为市售购买产品。
实施例1 清开灵有效组分干预小鼠脑缺血模型的靶点模块识别
本实例以精制清开灵有效组分黄芩苷(BA)、栀子苷(JA)、胆酸(UA)干预小鼠脑缺血模型的基因表达谱数据为例，构建不同药物干预前和干预后的基因共表达网络，并且以BA组的数据来具体说明本方法的实施。
整个实施过程的示意图见图1。
1.获得基因共表达数据
应用精制清开灵有效组分黄芩苷(BA)、栀子苷(JA)、胆酸(UA)分别对小鼠脑缺血模型(MCAO)进行药物干预。
小鼠脑缺血模型的构建：小鼠MCAO模型制备采用Zea-Longa线栓法，稍加改进，致备小鼠脑缺血1.5h再灌注24h的模型。小鼠术前禁食12h，自由饮水，经以10％水合氯醛溶液为麻醉剂，采用腹腔注射的方式，以0.4ml/100g体重的剂量，麻醉小鼠，仰卧固定，颈部正中切口，分离并暴露左侧颈总动脉，颈内外动脉，用镊子小心剥离出颈总动脉、颈外动脉，分离迷走神经。用动脉夹夹住颈总动脉，暂时阻断颈总动脉与颈内动脉的血流。用消毒后的缝合线在颈外动脉远心端结扎，并剪断颈外动脉，确保颈外动脉残端长度长于0.5cm。在颈内动脉近端备线，远端放置动脉夹，颈总动脉分叉膨大处切口，松开动脉夹，向颈内动脉插入栓线18～20mm，在大脑中动脉起始端堵塞大脑中动脉，然后将颈总动脉连同鱼线一起结扎，外留10mm线头，缝合皮肤。脑缺血1.5h时退出线栓，使血流再通。24h后观察小鼠的神经功能评分判断造模是否成功。
将构建得到的脑缺血模型小鼠随机分成5组，包括sham组、vehicle组、BA组、JA组和UA组，缺血后2h，用药组按2ml药液/kg体重分别经尾静 脉注射BA(5mg/ml)、UA(7mg/ml)和JA(25mg/ml)。Sham组和vehicle组小鼠接经尾静脉注射0.9％NaCl(2ml/kg体重)。然后按一步法分别抽提小鼠海马RNA，利用Bioanalyzer基因芯片对RNA质量进行评估。利用小鼠Oligo片段的芯片对干预后小鼠的基因表达谱进行分析。所用芯片由与脑缺血相关的374个c-DNA组成。本实例所用基因表达谱数据共有4组，即模型组(Vehicle)、黄芩苷治疗组(BA)、栀子苷治疗组(JA)及胆酸治疗组(UA)。
将模型组基因表达谱数据作为药物干预前的数据，BA组、JA组和UA组基因表达谱数据作为药物干预后的数据。每组基因表达谱数据都由12个样本的374个基因组成(Tbp、Zeb1、Pou2f1、Foxb1、Creb1、Camk2g、Csf1、F5、Hspd1、Matn2、Mt1、Adamts1、Klf6、Dffa、Rgs18、Rhoa、Kcnmb1、Pdcd11、Pdpk1、Casp8ap2、Mogat1、Rps26、Ak1、Csnk2a2、Dkk2、Ppm1e、Tnfrsf22、Trp53i11、Smpd3、Grin1、Cdk5、Jund、E2f1、Apoe、Il1b、Prkar1b、Il7r、Ngfb、Rela、Ifnar1、Adcy6、Bak1、Fzd6、Prkch、Rgs4、Actg1、Gck、Rgs9、Sox9、Rgs1、Dgke、Rgs20、Map2k2、Pin1、Prkcn、Dgkz、Csnk1g1、Dusp4、Il11、Grb2、Shc1、Syk、Sim2、Ywhah、Fgf13、Bid、Gstm2、Rarg、Pou3f1、Camk2b、Mapkapk2、Tcf4、Sos1、Stat5a、Vegfb、Bad、Etv3、Id1、Lcat、Nf1、Gsn、Bbc3、Clu、Capn9、Ercc5、Comt、Ctsl、Amph、Vegfc、Bax、Cyp51、Sox10、Nfyc、Gata2、Id3、Lef1、Pou6f1、6330503C03Rik、Ech1、Ccl4、Itm2a、Hspa1a、Cbx3、Klf10、Idh3g、Gpx2、Map2k5、Daxx、E2f3、Fgf12、Ikbkg、Btrc、Ikbkap、Ifnar2、Cdk5、Psmb1、Sufu、Gab1、Sox30、Pxn、Pygo2、Ctnnb1、Grin2a、Il5ra、Cdk4、Bcl2l1、Actb、Myb、Prkca、Csf2rb2、Gnaq、B-raf、Wnt6、Adcy7、Cacna1b、Fzd7、Prkcm、Rock1、Adcy8、Prkcc、Sub1、Tuba1b、Rgs6、Plcb1、Mknk1、Diablo、Mef2c、Lrp1b、Dgkg、Rgs12、Serpina5、Hspb1、Ppm1b、Dlk1、Cdc42、Fadd、Mdfi、Fgf11、Map3k4、Klk1b3、Il6ra、Tgfb2、Wnt11、Ccna1、Map2k6、Htr1f、Zmat3、Bnip3、Tsg101、Vim、Srf、D14Abb1e、Cdh11、Vdac2、Tfdp1、Gak、Ccna2、Vegfa、Vegfa、Hdac1、Srebf1、Stch、E2f1、Nfatc1、Gna12、Gna13、Cacnb3、Zic1、Pou4f3、Tcf12、Ldb1、Capns1、Fxyd2、Gcgr、LOC100304588、Syt11、Gadd45a、Pbx2、Ier3、Mapk9、Ctnnbip1、Fgf15、Smad3、Nlk、Mecp2、 Sigirr、Rgs18、Ptk2b、Sap30bp、Pcmt1、Tcf3、Braf、Ankrd6、Rgs5、Rap1gap、Adcy1、Grin2b、Gap43、Map2k1、Mapk10、Tgfb1、Lta、Rps6ka1、Wnt3、Rara、Prkcd、Atf4、Adcyap1r1、Cycs、Hint1、Rdx、Src、Adcy9、Prkce、Shcbp1、Elk3、Rgs14、Rgs17、Dusp10、Tubb3、Cyc1、Dusp16、Plcg2、Fzd10、Dgkd、Stat3、Mapk14、Map2k4、Htr1a、Map3k2、Frat1、Casp7、Eef2k、Thbd、Rarb、Camk4、Htr2c、E2f5、Met、Htr7、Camk2b、Stat6、Sod1、Efna4、Vdac3、Adora1、Bmp1、Vdac1、Grb2、Igfbp2、Top2b、Rpl35、Bdnf、Ppp3cb、Raf1、Cpe、Cacnb3、0610007C21Rik、Gna14、Gna11、Tuba1a、Zic3、Mlx、Id4、Ldb2、Sepp1、Prodh、S100a9、Pgam2、Rcan1、Abcc5、Ccr5、Ap1m1、Map3k5、Csnk1e、Axin1、Freq、Sh2b1、Rps6ka4、Wif1、Nkd1、Pam、Crem、Tgm2、Barhl1、Tradd、Plcd4、Ppp2r4、Otud7b、Rgs7、Casp2、Junb、Il2rg、Bad、Il1a、Egr1、Pdgfa、Gapdh、Eif4e、Apc、Prkcz、Parp1、Egfr、Prkcb1、Rgs2、Traf2、Ccr3、Rgs16、Smpd1、Tbp、Dgka、Mos、B230120H23Rik、Eif4e2、Rgs19、Adcy3、Creb5、Taf7、Pik3ca、Stat1、Il15、Atf3、Dvl3、Map3k3、Casp4、Kcnq1、Ptp4a3、fosB、Wnt3a、Calm1、Htr3a、Crkl、Casp3、Lhx1、Camk4、Selenbp2、Tcfe2a、Scg5、Pold3、Mmp2、Farp2、Pold2、Pold1、Gpx4、App、Mlh3、Rbl2、Tpp2、Cdh3、Fmo2、Pold4、Arf1、Sox1、Arhgef1)，数据构成了374×12的基因表达矩阵，部分数据元素的格式如下：
vehicle组：
Array ID Msam-1 Msam-2 Msam-3 Msam-4 Msam-5 Msam-6
Sc02R02C10 0.908 0.972 0.817 0.849 1.077 1.011
Sc03R02C10 1.052 1.029 1.41 1.267 1.152 1.065
Sc04R02C10 0.615 0.862 0.961 0.953 0.614 0.722
Sc01R03C01 1.127 1.096 1.203 1.307 1.181 1.26
Sc02R03C01 1.017 1.134 0.972 1.169 1.255 1.205
Sc03R03C01 1.082 0.93 1.039 1.154 0.741 1.112
Sc04R03C01 1.112 1.408 1.167 1.201 1.291 1.229
Sc01R03C02 1.763 2.578 2.213 2.703 1.518 2.124
BA组：
Array ID Sam 1 Sam 2 Sam 3 Sam 4 Sam 5 Sam 6
Sc02R02C10 1.019 0.966 1.019 1.121 1.146 0.967
Sc03R02C10 0.79 0.948 1.106 0.758 1.016 0.995
Sc04R02C10 0.931 1.003 0.873 0.712 1.02 0.833
Sc01R03C01 1.089 1.137 1.368 1.326 1.282 0.991
Sc02R03C01 1.269 1.482 1.451 1.522 1.044 1.321
Sc03R03C01 1.21 0.625 1.082 0.903 0.83 0.876
Sc04R03C01 1.822 2.376 0.804 1.226 2.714 1.944
Sc01R03C02 1.782 2.199 1.929 2.252 4.277 4.575
每一个单元格内的数值为该基因(行)在该样本(列)中的表达值，根据网络构建的方法，在这个表达谱中，具有同表达模式的基因可以被聚类在一起，形成共表达基因，这些共表达的基因聚类成为“基因共表达模块”。
2.网络构建和模块划分
用加权共表达网络分析(WGCNA)工具分别构建3个药物组数据的基因共表达网络(节点数目均为374个)并划分模块。经过模块识别BA组一共得到23个模块，模块的大小在3-149个节点(基因)之间；JA组得到42个模块，模块大小在3-46个节点之间；UA组得到15个模块，模块大小在3-29个节点之间。
以BA组的模块为例，其整个网络所有基因的表达情况如图2所示，可以看到明显的共表达模块的存在。基于分层聚类算法得到BA组的模块，每一种颜色代表一个模块，结果见图3。
3.模块拓扑参数计算
计算药物干预前(模型组)、BA干预后(BA组)的不同状态下模块的拓扑结构特征变化。
选择密度、连接度作为模块的拓扑指标，分别计算BA干预组和模型组模块的密度、连接度。将BA干预组与模型组进行对照，观察两组模块在密 度、连接度指标中的变化。
密度：density＝mean(vectorizeMatrix(A)).
连接度：
Figure PCTCN2016109895-appb-000001
4.计算拓扑结构变化的显著性。
用置换检验(Permutation test)对两组模块的拓扑结构变化进行显著性检验，对BA组23个模块随机置换1000次，与模型组相比，计算每个模块的标准化Δt/T，进而得到标准化Δt/T在1000次随机置换中的正态分布概率，当统计p值≤0.05时对应的标准化Δt/T值最小值，即药物干预后模块拓扑参数显著变化的界值。当p≤0.05时，W＝0.2为模块显著变化的界值。与模型组相比，BA干预组中Δt/T＞0.2的模块即可认为是药物的应答模块。
将BA组与模型组相比，共发现11个BA的应答模块，结果见图4。
还发现不同药物相对于模型组的应答模块存在差别，这说明不同药物干预后产生的生物网络中存在的模块具有拓扑结构差异。因此，本发明的方法还可用于不同药物组之间的相比，Δt/T＞0.2的模块为两组之间的差异模块。将BA组与JA组相比，可发现9个差异模块，结果见图5。
5.通过功能分析判别与比较药物的靶点模块
利用功能富集分析工具(DAVIDE等工具)对三个药物组的应答模块进行GO功能和KEGG通路的功能富集分析，以p＜0.05为富集分析显著性阈值。以BA组为例，BA组的11个应答模块中，BA-16模块富集到最多功能，包括41个GO功能和9条KEGG通路，BA-6模块没有富集到任何功能。如BA-16模块富集到的生物学功能如下：
GO功能 P-value
GO：0007267～细胞-细胞信号 0.001834725
GO：0001890～胎盘发育 0.002565481
GO：0006793～磷代谢过程 0.005366167
GO：0006796～磷酸盐代谢过程 0.005366167
GO：0001701～宫内胚胎发育 0.02229598
GO：0042325～磷酸化调控 0.026015859
GO：0019220～磷酸盐代谢过程调控 0.027879898
GO：0051174～磷代谢过程调控 0.027879898
GO：0005216～离子通道活性 0.032346557
GO：0022838～底物特异性通道活性 0.034250297
GO：0022803～跨膜转运蛋白活性 0.035130286
GO：0015267～通道活性 0.035130286
GO：0000187～MAPK活性激活 0.043290991
GO：0004714～跨膜受体蛋白酪氨酸激酶活性 0.045409762
KEGG通路 P-value
mmu05210：结肠直肠癌 0.005863221
mmu04010：MAPK信号通路 0.049483325
可以看到BA-16模块与细胞信号、胎盘发育、磷代谢过程等生物学功能相关，而且BA-16模块与结肠癌和MAPK信号通路相关，因此可以推断BA干预BA-16应答模块的机制与如上的功能有关。图6为BA-16模块及其富集到的功能。
对得到的BA、JA、UA三个药物组的应答模块都进行生物学功能的富集。发现，与模型组相比，BA组得到11个应答模块，其中3个模块未能富集到任何的生物学功能，其余的8个模块为BA的靶点模块；与模型组相比，JA组得到22个应答模块，其中3个模块未能富集到任何的生物学功能，其余的19个模块为JA的靶点模块；与模型组相比，UA组得到8个应答模块，其中1个模块未能富集到任何的生物学功能，其余的7个模块为UA的靶点模块。BA组的靶点模块(BA-5、BA-15)示例见图7。
BA、JA、UA三组间两两比较，如果一组的模块与其他两组相比较均不同，则该模块为该组的特征性模块。经比较发现，BA有5个特征性模块， 特征性模块(BA-9、BA-21)示例见图8；JA有8个特征性模块，UA有2个特征性模块。
通过模块的功能富集分析可以得到模块相关的生物学功能，即可能是药物的作用机制。
实施例2 对判别出的药物的靶点模块进行验证
1.基于数据库已知知识的验证
通过String(functional protein association networks)数据库可以查找已知的模块内基因(蛋白)的各种相互作用关系。如BA-16模块，将模块内的节点输入到数据库中，可以得到节点的相互作用关系，参见图9。
由图9可以看到，BA-16模块内的多个节点有多条边相连，每一条边代表一种相互作用关系的证据类型，证明已知这些节点确实存在着相互作用关系。
2.基于实验的验证
采用实验方法，通过基因的表达量来证实靶点模块的组成意义。
用Western Blot方法对各治疗组的靶点模块基因进行验证。BA-21模块中，VEGF和B230120H23Rik的表达水平与Vehicle组相比具有显著性差异，分别参见图10A和图10B。
利用免疫共沉淀(Co-Immunoprecipitation)证实模块内基因编码蛋白的相互作用关系。BA-16模块内Met和Egfr蛋白的免疫共沉淀结果表明二者有直接的相互作用关系(图11B)，BA-9模块内的Fmo2和Jund的免疫共沉淀结果表明二者有相互作用关系(图11A)。
实施例3 在抗PD-1治疗黑色素瘤基因表达网络中的应用
本实例将本发明的方法应用于抗PD-1治疗黑色素瘤的mRNA表达数据，用以判别抗PD-1治疗黑色素瘤的靶点模块。将抗PD-1治疗产生反应的病例分为一组(效应组)，将未产生反应的病例分为一组(对照组)，利用WGCNA方法分别构建两组的基因共表达网络。
本实例中所用数据下载自GEO(Gene Expression Omnibus)数据库 (https://www.ncbi.nlm.nih.gov/geo/，数据ID为GSE78220)，该数据集为抗PD-1治疗黑色素瘤的mRNA表达数据，样本来源为人，共有28例样本，其中15例样本为效应组，13例样本为对照组。数据中共包含mRNA数量为25268。具体实施过程如下：
1.mRNA表达数据的整理
将下载的数据整理成WGCNA分析所需的数据表格格式，表格的列为mRNA，行为治疗样本，相对应的单元格内数值为该mRNA(行)在该样本(列)中的表达值，基于mRNA的共表达情况，进行网络构建和模块划分。
2.网络构建和模块划分
同样用WGCNA工具构建效应组的mRNA共表达网络并划分模块。在网络构建过程中，有3080个mRNA由于过多的缺失值而被移除，剩余的22188个mRNA被用来构建网络(所构建网络的节点数为21976个)和模块划分。在效应组中共得到30个模块，模块的大小在38-2900个节点(mRNA)之间。经WGCNA识别得到效应组的模块划分结果见图12。
3.模块拓扑参数计算
计算效应组和对照组模块的拓扑结构特征变化。选择密度、连接度作为模块的拓扑指标，分别计算效应组和对照组模块的密度、连接度。将效应组与对照组进行比较，观察两组模块在密度、连接度指标中的变化。
4.计算拓扑结构变化的显著性。
用自助重采样程序(bootstrap resampling)对两组模块的拓扑结构变化进行显著性检验，对效应组的模块组成基因进行10000次重采样，与对照组相比，计算每个模块的标准化Δt/T，进而得到标准化Δt/T在10000次重采样过程中的正态分布概率，当统计p值≤0.05时对应的标准化Δt/T值最小值，即模块拓扑参数显著变化界值。当p≤0.05时，W＝0.26为模块显著变化的界值。与对照组相比，抗PD-1效应组中Δt/T＞0.26的模块即可认为是药物的应答模块。
将效应组与对照组相比，共发现8个抗PD-1治疗黑色素瘤的应答模块，结果见图13。
5.通过功能分析判别抗PD-1治疗黑色素瘤的靶点模块
利用功能富集分析工具(DAVIDE 6.7工具)对效应组的应答模块进行GO功能和KEGG通路的功能富集分析，以p＜0.05为富集分析显著性阈值。在抗PD-1治疗效应组中的8个应答模块中都能够富集到相应的功能，抗PD-1治疗黑色素瘤的靶点模块。
实施例4 在英夫利昔单抗(infliximab)治疗类风湿性关节炎(RA)的基因表达网络中的应用
本实例将本发明应用于英夫利昔单抗治疗RA的mRNA表达数据，用3个拓扑参数(密度、连接度、熵)来判别英夫利昔单抗治疗RA的靶点模块。将病例按照英夫利昔单抗治疗前、后分为对照组和治疗组，利用WGCNA方法构建治疗组和对照组的基因共表达网络。
本实例中所用数据下载自GEO(Gene Expression Omnibus)数据库(https://www.ncbi.nlm.nih.gov/geo/，数据ID为GSE8350)，该数据集为英夫利昔单抗治疗RA的mRNA表达数据，样本来源为人，本实例选用其中36例样本，其中18例为对照组，另18例为治疗组。具体实施过程如下：
1.mRNA表达数据的整理
数据整理过程参见实施例3。
2.网络构建和模块划分
同样用WGCNA工具构建效应组的mRNA共表达网络并划分模块。共有776个mRNA被用来构建网络(所构建网络的节点数为776个)和模块划分，在治疗组中共得到46个模块，模块的大小在3-73个节点(mRNA)之间。经WGCNA识别得到效应组的模块划分结果见图14。
4.模块拓扑参数计算
计算效应组和对照组模块的拓扑结构特征变化。选择密度、连接度、熵作为模块的拓扑指标，分别计算效应组和对照组模块的密度、连接度、熵。将治疗组与对照组进行比较，观察两组模块在密度、连接度、熵指标中的变化。
4.计算拓扑结构变化的显著性。
用置换检验(Permutation test)对两组模块的拓扑结构变化进行显著性 检验，对治疗组的模块进行1000次置换，与对照组相比，计算每个模块的标准化Δt/T，进而得到标准化Δt/T在1000次置换过程中的正态分布概率，当统计p值≤0.05时对应的标准化Δt/T值最小值，即模块拓扑参数显著变化界值。当p≤0.05时，W＝0.15为模块显著变化的界值。与对照组相比，英夫利昔单抗治疗组中Δt/T＞0.15的模块即可认为是药物的应答模块。
将治疗组与对照组相比，共发现14个英夫利昔单抗治疗RA的应答模块，结果见图15。
5.通过功能分析判别英夫利昔单抗治疗RA的靶点模块
利用功能富集分析工具(DAVIDE 6.7工具)对效应组的应答模块进行GO功能和KEGG通路的功能富集分析，以p＜0.05为富集分析显著性阈值。在英夫利昔单抗治疗RA的14个应答模块中，有2个模块未富集到任何生物学功能，其余12个模块为英夫利昔单抗治疗RA的靶点模块。
实施例5 在黄芩苷治疗小鼠脑缺血模型蛋白相互作用网络中的应用
本实例以精制清开灵有效组分黄芩苷(BA)干预小鼠脑缺血模型的蛋白相互作用网络为例进行发明的应用。与模型组对照，得到BA干预小鼠脑缺血模型的差异表达基因，基于String数据库进行差异表达基因的网络映射，从而构建BA干预小鼠脑缺血模型的蛋白相互作用网络，用本发明的方法判别BA干预小鼠脑缺血蛋白相互作用网络的靶点模块。
本实施例与实施例1的差别在于所用数据来源数量不同，且网络类型为蛋白相互作用网络，因此与实施例1中的模块划分结果和应答模块判别结果均存在一定差异。下面仅对与实施例1中差异部分进行描述。
1.蛋白相互作用网络数据。
BA干预小鼠脑缺血模型的差异表达基因获取参见实施例1。利用String数据库，通过差异表达基因的映射构建蛋白相互作用网络，所构建的网络共包含2229个蛋白(节点)。
2.模块划分
BA干预小鼠脑缺血模型的蛋白相互作用网络中，用MCODE方法进行模块划分，共得到49个模块。
3.模块拓扑参数计算
计算BA干预后模块的拓扑结构特征变化。
选择密度、连接度作为模块的拓扑指标，分别计算BA干预组和模型组模块的密度、连接度。将BA干预组组成模块与模型组中相应节点进行对照，观察两组模块在密度、连接度指标中的变化。
4.计算拓扑结构变化的显著性。
用置换检验(Permutation test)对两组模块的拓扑结构变化进行显著性检验，对BA组49个模块随机置换1000次，与模型组相比，计算每个模块的标准化Δt/T，进而得到标准化Δt/T在1000次随机置换中的正态分布概率，当统计p值≤0.05时对应的标准化Δt/T值最小值，即药物干预后模块拓扑参数显著变化的界值。当p≤0.05时，W＝0.11为模块显著变化的界值。与模型组相比，BA干预组中Δt/T＞0.11的模块即可认为是药物的应答模块。
将BA组与模型组相比，共发现13个应答模块。
5.通过功能分析有个6应答模块没有富集到相应功能，其余7个模块为BA在蛋白相互作用网络是的靶点模块。
通过以上实例分析，证明了本发明能够有效地在生物网络中识别药物的靶点模块，还可以通过不同药物组间的比较得到不同药物的保守模块和差异模块。
以上对本发明具体实施方式的描述并不限制本发明，本领域技术人员可以根据本发明作出各种改变或变形，只要不脱离本发明的精神，均应属于本发明所附权利要求的范围。

Claims (10)

  1. 一种判别或比较药物作用模块的方法，所述方法包括以下步骤：
    (1)对第一生物网络Gi(Vi，Ei)和第二生物网络Gj(Vj，Ej)进行模块的划分，分别得到模块集合Gi(Mi)和Gj(Mj)；其中第一生物网络Gi(Vi，Ei)为药物干预前的生物网络，而第二生物网络Gj(Vj，Ej)为药物干预后的生物网络；或者第一生物网络Gi(Vi，Ei)为第一药物干预后的生物网络，而第二生物网络Gj(Vj，Ej)为第二药物干预后的生物网络；
    (2)计算模块集合Gi(Mi)和Gj(Mj)中各个模块的一个或多个拓扑结构特征M(t)，得到单一或综合的拓扑指标T＝M(t1，t2...tm)，m为拓扑结构特征数目，并由此计算两个模块集合中对应模块的拓扑参数变化Δt/T；
    (3)将对应模块的拓扑参数变化Δt/T与显著性界值W进行比较，当第一生物网络Gi(Vi，Ei)为药物干预前的生物网络，而第二生物网络Gj(Vj，Ej)为药物干预后的生物网络时，如果Δt/T≥显著性界值W，则所述模块为所述药物的作用模块；当第一生物网络Gi(Vi，Ei)为第一药物干预后的生物网络，而第二生物网络Gj(Vj，Ej)为第二药物干预后的生物网络时，如果Δt/T≥显著性界值W，则所述模块为第一药物和第二药物的作用差异模块。
  2. 根据权利要求1所述的方法，其特征在于，所述方法还包括：
    (4)将所述药物的作用模块进行生物功能富集分析，根据功能富集结果判断所述模块是否是所述药物的靶点模块。
  3. 根据权利要求1或2所述的方法，其特征在于，步骤(1)中，所述药物、第一药物、第二药物为同类型的用以预防和治疗疾病的中药或化学药；优选地，所述第一药物、第二药物为同类型的中药或化学药的药效组分。
  4. 根据权利要求1至3中任一项所述的方法，其特征在于，步骤(1)中，所述第一生物网络和第二生物网络为生物分子网络，例如选自基因共表达网络、蛋白质相互作用网络、基因转录调控网络、非编码基因调控网络及代谢 网络；更优选地，所述第一生物网络和第二生物网络为相同类型的基因共表达网络或蛋白质相互作用网络；
    更优选地，所述生物分子网络的节点数目≥200、≥500、≥1000或≥5000。
  5. 根据权利要求1至4中任一项所述的方法，其特征在于，步骤(1)中，采用选自聚类、启发式搜索、种子扩展、矩阵分解、拓扑网络等网络分析算法进行模块划分；更优选地，采用选自分层聚类算法、WGCNA、MCODE、MCL、CFinder、CPM、SPC、G-N algorithm、ModuLand、DME、MINE、SVD的方法进行模块划分。
  6. 根据权利要求1至5中任一项所述的方法，其特征在于，步骤(2)中，所述拓扑结构特征M(t)选自模块性(modularity)、度(degree)、连接度(connectivity)、密度(density)、聚类系数(clustering coefficient)、介数中心性(betweenness)、网络直径(diameter)、最短路径(shortest path)、熵(entropy)和特征路径长度(characteristic path length)；
    优选地，拓扑结构特征数目m为1-10；
    更优选地，拓扑结构特征数目m为2，拓扑结构特征M(t)包括密度和连接度；或者优选地，拓扑结构特征数目m为3，拓扑结构特征M(t)包括密度、连接度和熵。
  7. 根据权利要求1至6中任一项所述的方法，其特征在于，步骤(2)中，计算拓扑参数变化Δt/T包括：
    计算第一生物网络中每个模块的拓扑参数与第二生物网络中相应模块的拓扑参数相比的变化量，即各个拓扑参数的Δt/T值，采用离差标准化方法将各个拓扑参数的Δt/T进行标准化处理，经过离差标准化后，各个拓扑参数的Δt/T应在(0，1)之间。
  8. 根据权利要求1至7中任一项所述的方法，其特征在于，步骤(3)中，利用置换检验(permutation test)或自助重采样(bootstrap resampling)程序 计算模块拓扑参数变化的显著性界值W。
  9. 根据权利要求1至8中任一项所述的方法，其特征在于，步骤(3)中，显著性界值W通过以下方式获得：
    利用置换检验或自助重采样方法得到大于1000个与第一生物网络中每个模块对应的随机模块，计算每个随机模块与第二网络相应模块的标准化Δt/T值，进而计算所有随机模块的标准化Δt/T值的正态分布概率，显著性界值W等于所述正态分布概率中p≤0.05时对应的标准化Δt/T值。
  10. 根据权利要求1至9中任一项所述的方法，其特征在于，步骤(4)中，通过基因本体(GO)功能和/或通路(pathway)分析进行生物学功能富集分析。
PCT/CN2016/109895 2016-06-07 2016-12-14 一种判别或比较药物作用模块的方法 Ceased WO2017211059A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP16904507.7A EP3467689A4 (en) 2016-06-07 2016-12-14 METHOD FOR DIFFERENTIATING OR COMPARING AN ACTIVE SUBSTANCE ACTIVITY MODULE

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610399510.XA CN107480467B (zh) 2016-06-07 2016-06-07 一种判别或比较药物作用模块的方法
CN201610399510.X 2016-06-07

Publications (1)

Publication Number Publication Date
WO2017211059A1 true WO2017211059A1 (zh) 2017-12-14

Family

ID=60578315

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/109895 Ceased WO2017211059A1 (zh) 2016-06-07 2016-12-14 一种判别或比较药物作用模块的方法

Country Status (3)

Country Link
EP (1) EP3467689A4 (zh)
CN (1) CN107480467B (zh)
WO (1) WO2017211059A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110957002A (zh) * 2019-12-17 2020-04-03 电子科技大学 一种基于协同矩阵分解的药物靶点相互作用关系预测方法
CN113567639A (zh) * 2021-07-13 2021-10-29 中国食品药品检定研究院 中药材质量综合评价方法
CN114067906A (zh) * 2021-11-15 2022-02-18 扬州大学 一种融合多源生物信息的关键蛋白质识别方法

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110211634B (zh) * 2018-02-05 2022-04-05 深圳华大基因科技服务有限公司 一种多组学数据联合分析的方法
CN109192252A (zh) * 2018-08-23 2019-01-11 南开大学 共表达周期昼夜节律的转录组学在药物作用机制发现中的用途
CN110349625B (zh) * 2019-07-23 2022-02-08 中国科学院心理研究所 一种人类大脑基因表达时空常模的建立方法
CN112053742A (zh) * 2020-07-23 2020-12-08 中南大学湘雅医院 分子靶标蛋白的筛选方法、装置、计算机设备和存储介质
CN120340597B (zh) * 2025-06-19 2025-09-16 吉林大学 一种基于网络变化熵的天然药用提取物药效靶点鉴定方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103514381A (zh) * 2013-07-22 2014-01-15 湖南大学 整合拓扑属性和功能的蛋白质生物网络模体识别方法
CN103778349A (zh) * 2014-01-29 2014-05-07 思博奥科生物信息科技（北京）有限公司 一种基于功能模块的生物分子网络分析的方法
CN103782301A (zh) * 2011-09-09 2014-05-07 菲利普莫里斯生产公司 用于基于网络的生物活性评估的系统与方法
WO2015054266A1 (en) * 2013-10-08 2015-04-16 The Regents Of The University Of California Predictive optimization of network system response

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103049678B (zh) * 2012-11-23 2015-09-09 中国科学院自动化研究所 基于蛋白质交互作用网络的异病同治分子机理分析方法
CN103065066B (zh) * 2013-01-22 2015-10-28 四川大学 基于药物组合网络的药物联合作用预测方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103782301A (zh) * 2011-09-09 2014-05-07 菲利普莫里斯生产公司 用于基于网络的生物活性评估的系统与方法
CN103514381A (zh) * 2013-07-22 2014-01-15 湖南大学 整合拓扑属性和功能的蛋白质生物网络模体识别方法
WO2015054266A1 (en) * 2013-10-08 2015-04-16 The Regents Of The University Of California Predictive optimization of network system response
CN103778349A (zh) * 2014-01-29 2014-05-07 思博奥科生物信息科技（北京）有限公司 一种基于功能模块的生物分子网络分析的方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3467689A4 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110957002A (zh) * 2019-12-17 2020-04-03 电子科技大学 一种基于协同矩阵分解的药物靶点相互作用关系预测方法
CN113567639A (zh) * 2021-07-13 2021-10-29 中国食品药品检定研究院 中药材质量综合评价方法
CN114067906A (zh) * 2021-11-15 2022-02-18 扬州大学 一种融合多源生物信息的关键蛋白质识别方法
CN114067906B (zh) * 2021-11-15 2024-04-05 扬州大学 一种融合多源生物信息的关键蛋白质识别方法

Also Published As

Publication number Publication date
CN107480467A (zh) 2017-12-15
EP3467689A4 (en) 2020-02-26
EP3467689A1 (en) 2019-04-10
CN107480467B (zh) 2020-11-03

Similar Documents

Publication Publication Date Title
WO2017211059A1 (zh) 一种判别或比较药物作用模块的方法
US11566279B2 (en) Droplet-based method and apparatus for composite single-cell nucleic acid analysis
Filigheddu et al. Differential expression of microRNAs between eutopic and ectopic endometrium in ovarian endometriosis
CN112020565B (zh) 用于确保基于测序的测定的有效性的质量控制模板
AU2024219712A1 (en) Interpretation of genetic and genomic variants via an integrated computational and experimental deep mutational learning framework
Iorio et al. Identifying network of drug mode of action by gene expression profiling
CN112391470A (zh) 胰腺癌miRNA预后模型的确立及靶向基因的筛选方法
Chandrasekaran et al. A network view on Parkinson's disease
Zeng et al. Network-based approach to identify molecular signatures in the brains of depressed suicides
CN106503482B (zh) 一种用于定量分析药物干预前后生物分子网络中模块变化的方法
Ashok et al. Integrated systems biology approach using gene network analysis to identify the important pathways and new potential drug targets for Neuroblastoma
Zhang et al. Cross talk of chromosome instability, CpG island methylator phenotype and mismatch repair in colorectal cancer
Jha et al. Prioritizing disease biomarkers using functional module based network analysis: a multilayer consensus driven scheme
Wan et al. Machine learning-based characterization of PANoptosis-related biomarkers and immune infiltration in ulcerative colitis: A comprehensive bioinformatics analysis and experimental validation
Mégret et al. Combining feature selection and shape analysis uncovers precise rules for miRNA regulation in Huntington’s disease mice
CN118486370A (zh) 基于集成学习的成人急性b淋巴细胞白血病的分型辅助模型及其构建方法
Gollapalli et al. Elucidating genes and transcription factors of human peripheral blood lymphocytes involved in the cellular response upon exposure to ionizing radiation for biodosimetry and triage: an in silico approach
Lopes-Ramos et al. Regulation of PD1 signaling is associated with prognosis in glioblastoma multiforme
Pyon et al. Identifying gene signatures from cancer progression data using ordinal analysis
Datta et al. Molecular characterization of circadian gene expression and its correlation with survival percentage in colorectal cancer patients
Lee et al. Tumor type and cell type-specific gene expression alterations in diverse pediatric central nervous system tumors identified using single nuclei RNA-seq
Tani et al. MicroRNAs as Biomarkers for Ulcerative Colitis
US20250201404A1 (en) Rna-based method for seizure assessment
Singh et al. Identification of biomarkers and enriched pathways involved in lung cancer
Li et al. Machine Learning‐Based Integrative Analysis Identifies SUMOylation‐Related Genes Underlying the Immune Heterogeneity of Sepsis

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16904507

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2016904507

Country of ref document: EP

Effective date: 20190107