CN110648726B

CN110648726B - Network target-based drug network pharmacology intelligent and quantitative analysis method and system

Info

Publication number: CN110648726B
Application number: CN201910902205.1A
Authority: CN
Inventors: 李梢
Original assignee: Tsinghua University
Current assignee: Tsinghua University
Priority date: 2019-09-23
Filing date: 2019-09-23
Publication date: 2021-10-01
Anticipated expiration: 2039-09-23
Also published as: CN110648726A

Abstract

The invention provides a pharmacological intelligent and quantitative analysis method and system of a drug network based on the original theory of a network target, which is used for measuring the overall effect of drugs (including various drug types such as Chinese and western drugs) intervening in a disease biological network. The method can synthesize qualitative and/or quantitative biological information related to diseases and drugs, take a disease biological network as a target, measure the effect of drug intervention in the disease biological network from the aspects of system and integration, and reveal the integral action mechanism of the drugs. The method provides a measurement mode based on qualitative and quantitative selection, adopts biological function multi-scale qualitative analysis and/or time-space multi-dimensional quantitative analysis to measure the drug intervention network effect, and provides a new key technology for breaking through the limitation of the traditional drug research mode based on experience or single target, understanding the drug network regulation mechanism, and rapidly and intelligently finding drug effect substances, action mechanisms, objective therapeutic effect indexes, clinical indications and the like.

Description

Network target-based drug network pharmacology intelligent and quantitative analysis method and system

Technical Field

The invention relates to a method and a system for intelligently and quantitatively analyzing a drug network pharmacology based on a network target, belonging to a network pharmacology technology which takes a disease biological network as a target and utilizes biomolecule-biological function-phenotype multi-scale information and/or space and time multi-dimensional information of diseases and drugs to intelligently and quantitatively analyze the overall effect of drug intervention in the biological network.

Background

Diseases (including traditional Chinese medicine syndromes) occur under the complex interaction of a plurality of factors such as environmental exposure, genetic susceptibility, age and the like, but the whole action mechanism of the medicine to disease intervention is difficult to reveal through the long-term 'single-gene, single-target and single-medicine' reduction theory medicine research mode. With the development of system biology and network pharmacology in recent years, the original theory of "network target-system regulation" was first proposed, which converts the drug research mode from descriptive "one drug one target" to predictive "network target" to resolve the new mode of the overall action mechanism of the drug (fig. 1). The theory of "network target-system regulation" suggests that the drug may be the result of "emerging" curative effect by exerting "multifactorial micro-effect" overall regulation effect on the internal biological network of the disease. Therefore, the method takes the key link of the disease biological network as the target, utilizes the network connection of the target effect on time and space, can better depict the overall intervention effect of the drug on the disease biological network, and realizes the effect of the drug on the disease biological networkUnderstanding of the mechanism of biology and the mechanism of action of drugs^[1]. Based on the holistic view of the use of traditional Chinese medicine, the theory and the method of 'network target-system regulation' provide theoretical basis for describing the intervention effect of the medicine on the disease biological network^[2]. The quantitative analysis of the drug intervention disease effect is beneficial to the construction of a drug effect evaluation system and the precise use of the drug, and has important scientific value and practical significance for understanding of the drug action mechanism, screening of effective drug combinations, prevention and treatment of diseases and the like.

To accelerate the evaluation of disease bio-network intervention and drug combination screening efficiency, many computational methods have been developed^[3]. For example, molecular model-based methods can estimate the effect of a combination drug when the target and mechanism of action are clearly defined^[4]. The statistical learning-based model can be trained according to pharmacological information, genomic features, network features, and the like^[5]. However, a method for quantitatively evaluating the network intervention effect of a drug on a disease biological network from time and space on the basis of biomolecule-biological function-phenotype multi-scale is still lacking at present, and particularly, a means for evaluating the network intervention effect of a complex and targetless traditional Chinese medicine prescription is lacking, so that a new method is urgently needed to be established.

Therefore, the invention provides a Network target-based drug Network pharmacology Intelligent and Quantitative analysis method and system (UNIQ).

Disclosure of Invention

Aiming at the bottleneck that the existing research lacks a quantitative description method for integrally evaluating the network effect of the medicament (including the traditional Chinese medicine), the invention provides a calculation method for quantitatively analyzing the intervention network effect of the medicament from the aspect of network adjustment, which integrates time and space dimension information and quantitatively describes the intervention network effect of the medicament on specific diseases (including traditional Chinese medicine syndromes) (figure 2). In the aspect of inputting data, the method provides a plurality of choices, and the input medicines and qualitative or quantitative information of diseases can be selected according to actual conditions. Disease-related inputs include: disease-Gene or Gene product characterization data (e.g.CIPHER^[6]Computational prediction data, database collection data such as OMIM) and disease-gene or gene product quantification data (e.g., from experimental data such as gene expression profiles, proteomics, clinical data, etc.). The drug side inputs include: drug-target or target profile characterization data (e.g., drug cipher)^[7]Calculation of predictive data, database collection of data such as drug bank) and quantification of drug-target or target profiles (e.g., experimental data, clinical data, etc., from gene expression profiles, proteomics, etc., under drug action). In the aspect of research methods, qualitative and quantitative measurement of drug intervention network effect is carried out through targeting diseases (including traditional Chinese medicine syndromes) related multi-scale networks. The multi-scale network comprises a gene or gene product interaction network, an interaction network such as the relation between cells and biological pathways, a network consisting of clinical phenotypes and the like. The method can be used for performing multi-scale qualitative analysis on biological functions and also can be used for performing time-space multi-dimensional quantitative analysis, and a user can select the analysis according to actual conditions. On the research object, the method can be applied to all medicine types, such as Chinese medicine formulas, Chinese medicines, Chinese medicine components, Chinese medicine component combinations, modern medicines such as chemical medicines, biological medicines and the like, and modern medicine combinations. The method can effectively combine macroscopic whole and microscopic mechanisms, is beneficial to discovering drug effect substances (such as traditional Chinese medicine markers), revealing drug network regulation action mechanisms, identifying objective indexes of curative effects (such as the curative effect markers), and prompting clinical indications of the drugs, thereby providing scientific basis for large-scale screening, accurate clinical positioning, side effect prediction and the like of the drugs. In order to achieve the above purpose, the present invention provides a Network target-based method and system uniq (using Network target for Intelligent and Quantitative analysis on drug actions) for pharmacological Network.

According to one aspect of the present invention, a method for quantitative analysis of network intervention effect of drug therapy for diseases based on network topology is provided. The method utilizes a random walk model to depict the action process of treating diseases by drugs from a spatial dimension by measuring the information such as topological attributes of drug targets distributed on a disease biological network. The method can fully explore the interaction among a plurality of components by utilizing a network target-system regulation mode so as to analyze the interaction of a single drug target on a disease biomolecule network and identify the drug effect 'emerging' effect generated by the superposition effect, thereby understanding the network regulation effect of the drug. The method is applicable to single drugs, drug combinations and drug populations.

According to one aspect of the present invention, a method for the quantitative analysis of the effects of network intervention in the treatment of disease with a kinetic-based drug is provided. The method is characterized in that the evolution of the intervention effect of the drug on a disease biological network along with the time is measured from multi-scale dynamic simulation on biomolecule-biological function-phenotype by referring to a relation curve between blood concentration and time in pharmacokinetics on the basis of the assumption that the drug effect is increased and then decreased on the time dimension. The method divides different stages of the drug effect into a latent period, a duration period and a residual period, and respectively quantitatively describes the different stages, and simultaneously quantitatively analyzes the overall effect of the drug on a disease biological network in a time dimension by integrating the association of a drug target and a regulated molecule on the network. The method has the innovation point that the time multidimensional quantitative simulation for realizing the biological network intervention effect of the disease can be closely related by utilizing a calculation model, so that the accuracy of the evaluation of the treatment effect of the medicine is improved. The method is suitable for all types of medicines, such as Chinese medicine formulas, Chinese medicines, Chinese medicine components, Chinese medicine component combinations, modern medicines such as chemical medicines and biological medicines, and modern medicine combinations.

According to one aspect of the present invention, a method for quantitatively analyzing network effects of drug intervention diseases based on macro-micro multi-scale network information is provided. The method is characterized in that the effects of space and time dimensions are integrated, biomolecule-biological function-phenotype multi-scale simulation can be carried out, qualitative and quantitative analysis is carried out on how micro-level biomolecule changes affect macro-level dynamic processes of drug treatment diseases, and a calculation model is provided for qualitative and quantitative analysis of network effects of drug intervention diseases. The method is suitable for all types of medicines, such as Chinese medicine formulas, Chinese medicines, Chinese medicine components, Chinese medicine component combinations, modern medicines such as chemical medicines and biological medicines, and modern medicine combinations.

According to yet another aspect of the present invention, there is provided a method for evaluating the network effect of a drug intervention disease using artificial intelligence calculations and quantitative analysis. The method provides multiple selectable measurement modes based on qualitative and quantitative data and a multi-scale network, adopts biological function multi-scale qualitative analysis or time-space multi-dimension quantitative analysis to measure the effect of the drug intervention network, and provides a new method for discovering drug effect substances, revealing drug action mechanisms, prompting objective indexes of curative effects, identifying clinical indications and the like. The method is suitable for all types of medicines, such as Chinese medicine formulas, Chinese medicines, Chinese medicine components, Chinese medicine component combinations, modern medicines such as chemical medicines and biological medicines, and modern medicine combinations.

According to one aspect of the invention, a network target-based intelligent and quantitative analysis method for pharmacology of a drug network is provided, which is characterized by comprising the following steps:

A) based on the results of the pre-constructed disease biological network and biological function multi-scale analysis, the quantitative analysis of the network intervention effect on the spatial dimension is carried out, and the method comprises the following steps:

measuring the network intervention effect of the drug according to the mode represented by the formula (1) on a node v in a disease biological network and an initial effect score I (v):

wherein E is_n+1(v) For the effect of network intervention at node v in the (n + 1) th state, E_n(u) is expressed as the network intervention effect of the node u in the nth state, I (v) is the initial effect score of the node v, w (v, u) is the edge weight between the nodes v and u, N (v) is the total node set in the network, alpha is the initial state influence factor and has the value range from 0 to 1,

determining a score of a disease bio-network intervention effect in a spatial dimension in a manner characterized by equation (2):

wherein m represents the mth biofunctional module, NSE, in the network_m(vi) a network intervention effect score for the mth biofunctional module in the spatial dimension, w (v, u) representing the weight of the edge between nodes v and u, E (u) and E (v) the network intervention effects of nodes u and v, respectively, TS (u) and TS (v) the topology property scores of nodes u and v, respectively,

B) based on a pre-constructed disease biological network and biological function multi-scale analysis result, according to the following steps:

characterizing the dynamic change of the biomolecule R regulated by the drug target T along with the time T,

wherein:

M_Ris the regulatory effect score of the R node by the T node,

a1, b1, a2, b2, a3 and b3 are undetermined parameters,

t1 is the end time of the effect latency period, and is also the start time of the effect duration,

t2 is the end time of the duration of the effect, and is also the start time of the residual period of the effect,

and

integrating the association of the drug target and the regulated molecule on the network according to the characterization mode of the formula (4), and determining the disease biological network intervention effect score NTE of the mth biological function module at the t moment in the time dimension_tm：

Wherein M is_t(i) The regulated effect score of the ith node at the time t, n represents the total node number in the disease biological network, d_Rt,TFor the shortest path between the R and T nodes,

C) according to the result of the predetermined biological function multi-scale analysis, the interference effect score NE of the disease biological network is determined by assigning weights to all the paths of the disease biological network and integrating the effect scores of the drug interference in the space dimension and the time dimension in a manner represented by the formula (5):

wherein:

w(bp_m) The weight of the mth biological function module is in a value range of 0-1

Beta is a time and space effect scale factor, and the value range is 0-1.

According to another aspect of the present invention, there is provided a use of the above network target-based drug network pharmacological intelligent and quantitative analysis method, which is characterized by comprising applying the quantitative analysis method for determining the effect of a drug on disease intervention to one selected from the following drugs:

a prescription of a Chinese medicine,

the Chinese medicinal materials are prepared into a Chinese medicinal preparation,

the components of the traditional Chinese medicines are mixed,

the components of the traditional Chinese medicines are combined,

modern medicines such as chemical medicines, biological medicines and the like,

modern drug combinations.

According to still another aspect of the present invention, there is provided a network target-based intelligent and quantitative analysis system for pharmacology of a drug network, comprising:

A) the part for carrying out network intervention effect quantification on the spatial dimension based on the pre-constructed biological network and biological function multi-scale analysis result of the drug intervention disease comprises the following steps:

a portion for measuring the network intervention effect of a drug in a manner characterized by formula (1) for a node v in a disease biological network and its initial effect score i (v):

determining a fraction of a score for a disease bio-network intervention effect in a spatial dimension, in a manner characterized by equation (2):

B) based on a pre-constructed biological network and biological function multi-scale analysis result of the drug intervention disease, the method comprises the following steps:

characterizing the part of the dynamic variation of the biomolecule R over time T under the regulation of the drug target T,

wherein:

M_Ris the regulatory effect score of the R node by the T node,

a1, b1, a2, b2, a3 and b3 are undetermined parameters,

t1 is the end time of the effect latency period and also the start time of the effect duration,

t2 is the end time of the effect duration and the start time of the effect residual period,

and

integrating the drug target and the regulated molecule in the network according to the mode characterized by the formula (4)Determining the disease biological network intervention effect score NTE of the mth biological function module at the time t on the time dimension_tmThe following components:

C) according to the result of the predetermined biological function multi-scale analysis, by giving weights to all the paths of the disease biological network and integrating the effect scores of the space dimension and the time dimension of the drug intervention, the part of the disease biological network intervention effect score NE is determined according to the mode represented by the formula (5):

wherein:

Beta is a time and space effect scale factor, and the value range is 0-1.

Drawings

Figure 1 is a graph comparing the "single target-local confrontation" mode and the "network target-system regulation" mode.

Fig. 2 is a flow chart of a network target based drug network pharmacological intelligence and quantification method and system (UNIQ algorithm).

Fig. 3 is a diagram of a new mode of intelligent drug development based on cyber pharmacology.

FIG. 4 is a diagram of the network regulation mechanism of the LIUWEIDIHUANG pill for inhibiting the transformation of digestive tract inflammatory cancer in example.

FIG. 5 is a graph showing the effect of LIUWEIDIHUANG pill on inhibiting the transformation of cancer due to digestive tract inflammation.

Fig. 6 is a network regulation mechanism diagram of the compound Huangdai tablet for treating leukemia according to the embodiment.

FIG. 7 is the network intervention effect of FUFANGHUANGDAI tablet for treating leukemia in example.

Detailed Description

The inventor proposes that the intervention characteristics of the medicine on the complex diseases are not in a form of single target-local antagonism, but exert overall curative effect through network target-systemic regulation. For example, a drug may be "on, off" of the overall effect by intervening on a set of targets with specific associations on the network, with the network connection of the target effects in time, space [2 ]. Ideally, the target effect of the drug action is superimposed or coordinated on the disease biological network, and is transmitted through the biological network, and the effect threshold is exceeded, so that the overall effect is turned on and the therapeutic effect is expressed; meanwhile, the target effect is dispersed or antagonized on a biological network related to toxicity and side effect, and is lower than an effect threshold value, so that the overall effect is turned off, and toxicity is not generated or reduced. The inventor further provides a method and a system for pharmacological Intelligent and Quantitative analysis of a drug Network based on a Network target (UNIQ) according to the present invention, which uses a biological Network as a carrier, maps multi-scale information of diseases and drugs into the biological Network, establishes an Intelligent calculation and Quantitative analysis method for evaluating the Network effect of drug intervention diseases on the basis of analyzing the correlation among various factors on the Network, utilizes a Network dynamics and a Network topology method to evaluate the Network effect of drug intervention diseases from time and space dimensions, and provides a new key technology for discovering pharmacodynamic substances, disclosing a drug action mechanism, prompting objective therapeutic effect indexes, identifying clinical indications, and the like. The analysis flow is shown in fig. 2. The method comprises the following steps:

1. drug target and disease biomolecule calculation analysis and collection module/step

Computational analysis and collection of drug targets and disease-related biomolecules is the primary operation in evaluating the network effects of drug intervention on disease. The module includes qualitative and quantitative analysis and collection of disease-gene or gene product by calculationData and qualitative or quantitative data of drug-target or target profile. In one embodiment of the present invention, first, the input specific disease-related phenotype is read, and a disease-related biomolecule prediction algorithm based on a biological network is executed to obtain a disease-related biomolecule set^[6](ii) a Secondly, reading input medicinal chemical information, executing a medicament target prediction algorithm based on a biological network, and outputting a medicament target set^[7]. In addition, qualitative and quantitative data obtained by experimental modes such as disease-gene information in a public database such as OMIM, drug-target activity data in ChEMBL, disease gene expression profiles, selection of differential expression genes from the gene expression profiles under the action of drug intervention and the like are collected to supplement a disease-related biomolecule and a drug target set. For modern drug combinations of traditional Chinese medicines, traditional Chinese medicine prescriptions, chemical medicines, biological medicines and the like, a plurality of compounds may have common targets which have important significance for intervention effect of disease biological networks and can reveal the synergistic effect existing among component combinations^[8]. According to one embodiment of the invention, the screening method provided by the literature is adopted to count the targets with the frequency of occurrence more remarkable than the random condition in the drug combination, and the targets with the common influence of the drug combination are output^[9]。

2. Disease biological network construction and biological function multi-scale analysis module/step for drug intervention

The disease biological network of drug intervention is mainly a biological network formed by complex interaction relationship between drug targets and disease-related biomolecules and is expanded to biological functions based on biomolecules^[5,10]. Based on qualitative and quantitative data of disease-gene or gene product and qualitative and quantitative data of drug-target or target spectrum, and combining multi-scale interaction relations such as protein interaction, cell or biological signal transduction pathway relation and the like, the disease-related biomolecules and the drug targets are generated to form a multi-scale biological network for drug intervention of diseases. The multi-scale network comprises a gene or gene product interaction network, a cell, a channel and other interaction networks, a clinical phenotype formed network and the like. Network containing multiple biological function modulesBlocks, e.g., biological function modules that have a close association with disease occurrence, biological function modules for drug intervention, etc. These key biofunctional modules are often involved in more biomolecules.

In order to quickly understand the related biological functions of the biomolecules, the operation of the biological function multi-scale analysis method is performed. Inputting a group of related genes or gene products in a disease biological network and a target or a target spectrum of a drug intervention biological network as a gene set to be detected, and inputting a group of biological pathways and biological molecules contained in a biological process as a biological function gene set. In order to reflect the correlation between biological functions in a biological network of drug intervention diseases, the invention provides a method for matching biological function terms by using keyword mapping. Determining the relevance significance between the gene set to be tested and the biological function gene set by adopting a statistical test according to the mode represented by the formula (1)^[11,12]And establishing a keyword mapping table according to the disease and drug related biological function terms to perform biological function term matching. And according to the matching scores, clustering the biological function terms with remarkable association, forming a plurality of key biological function modules in a disease biological network for drug intervention, and obtaining the association among the biological functions for multi-scale qualitative analysis of the network intervention effect from the biological functions.

Wherein S (A, B) is the matching score of the biological function terms, A, B are vectors of the biological function terms, and n is the number of the most words in the biological function terms. If the keywords in A are directly contained in B, multiplying the matching score by the amplification weight omega of the keywords after the keywords appear_A,BValue range omega_A,B>1。

Through the construction of the disease biological network and the multi-scale qualitative analysis of biological functions, the inventor describes the relationship between the target set of the medicine and the key links of the disease biological network, and provides a basis for measuring the effect of the medicine intervening the disease biological network.

3. Disease biological network intervention effect topology quantitative analysis module/step

Suppose that: in the spatial dimension, the space is increased gradually, the effect is increased to be constant, or when the effect is increased to a certain degree, the speed increase is obviously slowed down (refer to the process that the random walk algorithm converges to a steady state).

And (3) carrying out operation of quantitative analysis of network intervention effect on the spatial dimension based on the disease biological network constructed in the module 2 and the obtained biological function multi-scale qualitative analysis result. The prediction of drug targets and disease biomolecules based on biological network is a representative prediction method, firstly, an initial effect score I is defined for each node in the disease biological network, and a disease biomolecule prediction method CIPHER in a module 1^[6]Prediction score of (1) and drug target prediction method in Module 1 DrugCIPHER^[7]Are determined together. The determination method is as follows: disease-related nodes are determined after normalization by CIPHER scores; drug target nodes were determined after normalization by drug cipher prediction score; disease and drug target common nodes were determined by normalization after averaging CIPHER and drug CIPHER predictive scores. Meanwhile, the effect of the drug on the disease biological network is gradually diffused from the drug target node where the drug initially intervenes to other nodes in the network. In turn, the network intervention effect of the drug is measured based on a random walk model. As shown in the following formula (2):

in the formula, E_n+1(v) The network intervention effect of the node v in the state of the (n + 1) th step. E_nAnd (u) is expressed as the network intervention effect of the node u in the nth step state. I (v) is the initial effect score of node v, w (v, u) represents the edge weight between nodes v and u. N (v) represents the total set of nodes in the network. Alpha is an initial state influence factor and has a value ranging from 0 to 1.

Drug targets are mapped onto disease biological networks, so that the measure of network effect is the effect of a single point. In practice, however, the extent to which the disease biological network as a whole is affected is not constituted by the superposition of single node effects alone, but depends on the strength of the action of the synergistic biomolecules. If the target itself is less intervened, the synergy will be less. Therefore, in an embodiment according to the present invention, the operation of measuring the network effect by using the network topology attribute is used, and the effect of a single node and the topological relation between nodes in the network are considered when measuring the network effect, so as to finally give a score of the intervention effect of the disease biological network in one spatial dimension, wherein the specific calculation formula is as follows (3):

in the formula, NSE_mIs the network intervention effect score of the mth biological function module in the spatial dimension. w (v, u) represents the edge weight between nodes v and u. E (u) and e (v) are the network intervention effects of node u and node v, respectively. TS (u) and TS (v) are the topological attribute scores of the node u and the node v respectively, and are determined according to the formula in the literature^[13]. m represents the mth biofunctional module in the network.

4. Disease biological network intervention effect dynamics quantitative analysis module/step

Suppose that: in the time dimension, the time increases, and the effect increases first and then decreases (see plasma concentration-time curve in pharmacokinetics).

Besides other biomolecules affecting biological functions, the degree of the affected biomolecules in biological functions changes with time after the biomolecules in biological functions are affected by drugs^[14]. Based on the constructed disease biological network and biological function enrichment result, for the change of the influence of specific biomolecules by drugs along with time, the operation is carried out according to a blood concentration-time curve in pharmacokinetics:

in the formula (4), MR is the regulatory effect fraction of the R node by the T node. a1, b1, a2, b2, a3 and b3 are undetermined parameters. t1 is the end time of the effect latency period and also the start time of the effect duration, and t2 is the end time of the effect duration and also the start time of the effect residual period.

For a particular disease, the biomolecules that produce a synergistic effect are close in network distance, and the distance is related to the effect produced^[13]. Thus, by integrating the association of drug targets and regulated molecules on the network, the disease bio-network intervention effect score is calculated in the time dimension as follows:

wherein NTE_tmThe biological network intervention effect of the disease of the mth biological function module at the t moment in the time dimension. M_Rt(i) The ith node is the regulated effect at time t. n represents the total number of nodes in the disease biological network, d_Rt,TIs the shortest path between the R and T nodes. m represents the mth biofunctional module in the network.

5. Macro-micro multi-scale quantitative analysis module for intervention effect of disease biological network

The medicine can intervene different biological functions through the target, and plays a comprehensive role in the intervention effect of disease biological networks. According to the operation of the above modules 3-4, each drug intervenes in a process with a spatial dimension network intervention effect score NSE_mAnd network intervention effect score NTE at time t of time dimension_tm. And (3) according to the result of the multi-scale qualitative analysis of the biological function in the module 2, giving weights to all the channels of the disease biological network, integrating the effect scores of the space dimension and the time dimension of the drug intervention, and calculating the intervention effect of the disease biological network as follows:

wherein NE is the disease bio-network intervention effect score. w (bp)_m) The weight of the mth biological function module is in a value range of 0-1.Beta is a time and space effect scale factor, the value range is 0-1, and the beta is determined according to a method described in a literature^[14]。

NE describes the calculation of the intervention effect of a drug on a disease from multiple dimensions in time and space based on multi-scale information. Furthermore, comprehensive intervention mechanisms and effects of traditional Chinese medicine formulas, traditional Chinese medicines, traditional Chinese medicine components, traditional Chinese medicine component combinations, western medicines, western medicine component combinations and the like are understood, and a new key means is provided for research and development of medicines and explanation of the integral action mechanism of the medicines.

The theory and method of "network target-system regulation" provides a theoretical basis for describing the intervention effect of drugs on disease biological networks. The quantitative evaluation of the intervention effect of the medicine on the disease biological network from the time dimension and the space dimension is beneficial to the construction of a medicine effect evaluation system and the precise use of the medicine, so that a new intelligent traditional Chinese medicine research and development mode characterized by 'system and intelligence' is established (figure 3). The mode measures the overall intervention effect of the medicine on a disease biological network in a multi-dimension way of space and time through the high-precision intelligent calculation of the overall action mechanism of the traditional Chinese medicine, and discovers modern clinical indications and objective curative effect indexes of the traditional Chinese medicine rapidly and accurately in a large scale, so that the traditional research and development mode based on experience and trial and error is changed, the traditional efficacy mechanism, clinical positioning and pharmacodynamic substance basis of the traditional Chinese medicine are basically clear, and the original breakthrough of the upgrading of the traditional Chinese medicine industry is realized.

Example 1: example of Liuwei Dihuang Wan (pill of six ingredients with rehmannia)

The previous clinical research shows that the famous traditional Chinese medicine pill of six ingredients with rehmannia has better clinical effect on treating various complex diseases, such as the inhibition of the transformation of digestive tract inflammatory cancers such as esophagus^[15]However, the intrinsic mechanism of inhibition of gut inflammatory cancer transformation, as well as the active ingredient population, is still unclear. The inventor uses a network target intervention effect model UNIQ in quantitative evaluation of the efficacy of the famous traditional Chinese medicine Liuwei Dihuang pill for inhibiting the transformation of the alimentary canal inflammatory cancer, mechanism analysis and active ingredient group discovery.

Through UNIQ algorithm analysis, a network regulation mechanism (shown in figure 4) of the Liuwei Dihuang pill for inhibiting transformation of the digestive tract inflammatory cancer is analyzed, the fact that overall regulation of an 'immune-metabolism' module is possibly a key mechanism is found, an active component group for inhibiting transformation of the digestive tract inflammatory cancer is further identified, through multi-scale quantitative analysis, the fact that intervention of the Liuwei Dihuang prescription on a biological network for inhibiting transformation of the digestive tract inflammatory cancer such as esophageal cancer occurs, and the highest effect value exceeds an effect threshold value is found. The specific implementation steps are as follows:

1. drug targets and disease biomolecule prediction

The pill of six ingredients with rehmannia comprises 6 Chinese medicines of prepared rehmannia root, dogwood, yam, alisma orientale, moutan bark and tuckahoe, 322 chemical components are collected from the literature, and the number of the compounds contained in each Chinese medicine is shown in table 1. Drug target prediction representative algorithm drug cipher^[7]And predicting each compound target, and automatically extracting the first 100 candidate biomolecules of each set according to the highest accuracy of the prediction algorithm on the verification data set for subsequent analysis. Representative algorithm for prediction of disease genes, CIPHER^[6]Predicting the pathogenic genes of gastritis and gastric cancer. Meanwhile, counting the target with the occurrence frequency more obvious than the random condition in the drug combination, and outputting the target which is commonly influenced by the drug combination, namely the whole target^[9]。

TABLE 1 information of Chinese medicinal compounds contained in LIUWEIDIHUANG pill

Chinese medicine	Number of compounds	Reference to the literature
			Prepared rehmannia root	9	[16]
Fructus Corni	129	[16-18]
			Cortex moutan	46	[16,19]
Poria cocos (Schw.) wolf	39	[16,20]
			Rhizoma alismatis	25	[16,21,22]
Chinese yam	74	[16,23]

2. Construction of drug target and disease biological network and multi-scale qualitative analysis of biological function

And selecting the first 100 bits of the whole target and the first 100 bits predicted by the CIPEHR algorithm to construct a disease biological network. In addition, the method collects the activity data of traditional Chinese medicines and western medicines from public databases such as ChEMBL to select targets with stronger activity, and selects differential expression genes from the intervention gene expression data of the components of the traditional Chinese medicines and the western medicines detected in high throughput to supplement candidate biomolecules.

Clinical omics data from gastritis to gastric cancer, enteritis to intestinal cancer, and hepatitis to liver cancer are collected from an international public database, key biological process modules related to transformation of various inflammatory cancers such as cell proliferation, cell apoptosis, cell differentiation, DNA damage, metabolism, immunity and the like are analyzed, an incidence relation between biological processes is constructed by mutation information of genes contained in the biological processes by a disease database, and a disease biological function network is established. Each biological function module comprises a plurality of Gene Ontology entries and KEGG pathway entries.

Finally, Fisher's exact test was used to evaluate biological processes in disease biological networks and Gene Ontology databases and significance levels of signal pathways in KEGG databases, keyword mapping and term matching were used to analyze the association between the enriched disease and biological functions, resulting in a six-ingredient rehmannia pill target enriched disease, as shown in table 2.

TABLE 2 diseases with target enrichment of LIUWEIDIHUANG pill

3. Disease biological network intervention effect topology quantitative analysis

The initial effect score I of a node in a disease biological network defines a rule as follows: the scores of all disease-related nodes were normalized first, and likewise, the scores of all drug target nodes were normalized. The initial effect score of the disease-related node is then determined by the predictive score normalization results of the CIPHER algorithm^[6]The initial effect score of the drug target node is determined by the result of normalization of the prediction score of the drug target node by the drug CIPHER algorithm^[7]. And if a certain node is a disease-related node and a drug target node at the same time, taking the mean value of the two nodes.

The network intervention effect of the drug is measured based on a random walk model. As shown in the following formula:

in the formula, E_n+1(v) The network intervention effect of the node v in the state of the (n + 1) th step. E_nAnd (u) is expressed as the network intervention effect of the node u in the nth step state. I (v) is the initial effect score of node v, w (v, u) represents the edge weight between nodes v and u. In this embodiment, w (v, u) is determined by the STRING database by obtaining a score between v and u, and n (v) 198. α is 0.999.

The score of the total intervention effect of the disease biological network in the spatial dimension is:

in the formula, NSE_mIs the network intervention effect score of the mth biological function module in the spatial dimension. w (v, u) represents the edge weight between nodes v and u. E (u), E (v) is the network intervention effect of the nodes u, v. TS (u), TS (v) is the topological attribute score of the node u, v, and is determined according to the formula in the literature^[13]Here, the attribute of the degree of a node in the network is used. m represents the mth biofunctional module in the network, and m is 6 in the embodiment.

4. Disease biological network intervention effect dynamics quantitative analysis

The specific biomolecules are influenced by the drug with time, and the operation is performed according to a blood concentration-time curve in pharmacokinetics:

in the formula, M_RIs the regulatory effect score of the R node by the T node. a 1-64, b 1-16, a 2-66, b 2-1, a 3-66, b 3-2. t1 is the end time of the effector latency period and the start time of the effector duration, t1 is 2, t2 is the end time of the effector duration and the start time of the effector residual period, and t2 is 3.

The total intervention effect score of the disease bio-network in the time dimension is:

wherein NTE_tmThe biological network intervention effect of the disease of the mth biological function module at the t moment in the time dimension. M_Rt(i) The ith node is the regulated effect at time t. n represents the total number of nodes in the disease biological network, d_Rt,TIs the shortest path between the R and T nodes. m represents the mth biofunctional module in the network, and m is 6.

5. Macro-micro multi-scale quantitative analysis of intervention effect of disease biological network

The integrated results of the effect scores in the spatial dimension and the temporal dimension of the drug intervention are:

wherein NE is the disease bio-network intervention effect score. w (bp)_m) The weight of the mth biological function module is in a value range of 0-1. β is the time and space effect scale factor, β is 0.009. The simulation effect is shown in fig. 5. The "threshold" is the minimum effective concentration of the drug. Simulation results show that the intervention effect of the drug on the network target exceeds a threshold value, namely the effect is generated. At the same time, there is a "highest effect" point, which means: in the time dimension, the drug intervention effect reaches a minimum effective concentration and is at a sustained drug effect duration. Meanwhile, in the spatial dimension, the random walk reaches a steady state and continues at the steady state.

Example 2: compound Huangdai tablet embodiment

The compound Huangdai tablet has wide reports and better clinical effects on treating acute promyelocytic leukemia, but the action mechanism and the drug effect substance basis are still not completely clear. By adopting the network target intervention model UNIQ provided by the invention, main active ingredients such as tetra-arsenic sulfide, indirubin, tanshinone IIA and the like can be found to be capable of regulating a plurality of interrelated biological modules of acute promyelocytic leukemia. Furthermore, the biological basis of the compatibility rule of monarch, minister, assistant and guide of the model is explored, the network intervention module of the compound Huangdai tablet four traditional Chinese medicines is successfully found out, the active ingredients contained in the monarch drug realgar can regulate and control the biological network module of the fusion protein PML-RARA of the acute promyelocytic leukemia, the minister drug and the assistant drug intervene the PML-RARA neighbor network module, and the guide drug targets the membrane transport protein Aquaporin-9 and assists in arsenic transportation.

1. Drug targets and disease biomolecule prediction

The compound Huangdai tablet comprises 4 traditional Chinese medicines of salvia miltiorrhiza, indigo naturalis, radix pseudostellariae and realgar from literatureThe total 94 chemical components are collected, and the number of compounds contained in each traditional Chinese medicine is shown in table 3. Drug target prediction representative algorithm drug cipher^[7]And predicting each compound target, and automatically extracting the first 100 candidate biomolecules of each set according to the highest accuracy of the prediction algorithm on the verification data set for subsequent analysis. Representative algorithm for prediction of disease genes, CIPHER^[6]The pathogenic gene of acute promyelocytic leukemia (OMIM:612376) is predicted. Meanwhile, counting the target with the occurrence frequency more obvious than the random condition in the drug combination, and outputting the target which is commonly influenced by the drug combination, namely the whole target^[1]。

TABLE 3 information of Chinese medicinal compounds contained in FUFANGHUANGDAI tablet

2. Construction of drug target and disease biological network

Analysis shows that key biological process modules related to the acute promyelocytic leukemia comprise various types of apoptosis, cell proliferation, cell differentiation, metabolism, immunity, blood coagulation and the like. And then, establishing an incidence relation among biological processes and establishing a disease biological function network through keyword classification. Each biological function module comprises a plurality of Gene Ontology entries and KEGG pathway entries, keyword mapping and biological function term matching are utilized, the biological function terms with significant association are clustered according to matching scores, and a biological network of the compound Huangdai tablet intervening in the promyelocytic leukemia is obtained, as shown in FIG. 6.

In disease biological networksThe initial effect score I of a node defines a rule as follows: all disease-related node scores were normalized first, and likewise, the scores of all drug target nodes were also normalized. The initial effect score of the disease-related node is then determined by the predictive score normalization results of the CIPHER algorithm^[6]The initial effect score of the drug target node is determined by the result of normalization of the prediction score of the drug target node by the drug CIPHER algorithm^[7]. And if a certain node is a disease-related node and a drug target node at the same time, taking the mean value of the two nodes.

And measuring the network intervention effect of the medicine based on a random walk model. As shown in the following formula:

in the formula, E_n+1(v) The network intervention effect of the node v in the state of the (n + 1) th step. E_nAnd (u) is expressed as the network intervention effect of the node u in the nth step state. I (v) is the initial effect score of node v, w (v, u) represents the edge weight between nodes v and u. In this embodiment, w (v, u) is determined by obtaining a score between v and u from the STRIGN database, and n (v) 192. α is 0.999.

wherein NE is the disease bio-network intervention effect score. w (bp)_m) The weight of the mth biological function module is in a value range of 0-1. β is a time and space effect scale factor, β is 0.0001. The simulation effect is shown in fig. 7. The "threshold" is the minimum effective concentration of the drug. Simulation results show that the intervention effect of the drug on the network target exceeds a threshold value, namelyAn effect is produced. At the same time, there is a "highest effect" point, which means: in the time dimension, the drug intervention effect reaches a minimum effective concentration and is at a sustained drug effect duration. Meanwhile, in the spatial dimension, the random walk reaches a steady state and continues at the steady state.

Reference documents:

[1]Li,S,Zhang,ZQ,Wu,LJ,et al.Understanding ZHENG in traditional Chinese medicine in the context of neuro-endocrine-immune network.2007；IET Syst Biol 1,51-60.

[2]Li S.Mapping ancient remedies:applying a network approach to traditional Chinese medicine.Science 2015；350(6262Suppl):S72-S74.

[3]Oulas A,Minadakis G,Zachariou M,et al.Systems Bioinformatics:increasing precision of computational diagnostics and therapeutics through network-based approaches.Brief.Bioinform.2017；bbx151.

[4]Kitchen D B,Decornez,Hélène,Furr J R,et al.Docking and scoring in virtual screening for drug discovery:methods and applications.Nat Rev Drug Discov,2004；3(11):935-949.

[5]Murphy,Robert F.An active role for machine learning in drug development.Nat Chem Biol,2011；7(6):327-330.

[6]Wu X,Jiang R,Zhang MQ,et al.Network-based global inference of human disease genes.Mol Syst Biol,2008；4:189.

[7]Zhao,S.,Li,S.Network-based relating pharmacological and genomic spaces for drug target identification.PLoS One,2010；5:e11764.

[8]Barabasi AL,Gulbahce N,Loscalzo J.Network medicine:a network-based approach to human disease.Nat Rev Genet,2011；12(1):56-68.

[9]Liang,X,Li,H,Li,S.A novel network pharmacology approach to analyse traditional herbal formulae:the Liu-Wei-Di-Huang pill as a case study.Mol Biosyst,2014；10,1014-1022.

[10]Li S.Network systems underlying traditional Chinese medicine syndrome and herb formula.Curr Bioinformatics,2009,4(3):188.

[11]Fisher,Sir Ronald A.Mathematics of a Lady Tasting Tea.In James Roy Newman.The World of Mathematics,1956(3).

[12]Huang DW,Sherman BT,Lempicki RA.Systematic and integrative analysis of large gene lists using DAVID Bioinformatics Resources.Nature Protoc.2009；4(1):44-57.

[13]Li S,Zhang B,Zhang N.Network target for screening synergistic drug combinations with application to traditional Chinese medicine.BMC Syst Biol,2011；5(Suppl 1):S10.

[14]Guo Y,Nie Q,Maclean AL,et al.Multiscale modeling of inflammation-induced tumorigenesis reveals competing oncogenic and onco-protective roles for inflammation.Cancer Research,2017；canres.1662.

[15]Jiang TL,Yan SC,Zhao LF.Preventing effect of'Liuwei dihuang decoction'on esophageal carcinoma.Gan to kagaku ryoho.Cancer&chemotherapy,1989,16(4Pt 2-2):1511-1518.

[16] fujiaju, Xigui Rong and Severe establishment of Chinese herbs original phytochemistry component set, science Press 2009

[17] Wang student bin, Juan silk, poplar and the like, observing the volatile oil components of dogwood in 3 different producing areas by a GC-MS method, analyzing and testing technology and instrument, 2006

[18] GC-MS analysis of chemical components of dogwood volatile oil, fine chemical industry, 2006 and the like

[19] Zhang Yan, Van Jun an, Chinese medicinal material moutan bark research general overview III cortex moutan chemical composition research general overview Chongqing Chinese herbal medicine research, 2008 and the like

[20]Akihisa T,Uchiyama E,Kikuchi T,Tokuda H,Suzuki T,Kimura Y.Anti-tumor-promoting effects of 25-methoxyporicoic acid A and other triterpene acids from Poria cocos.J Nat Prod.2009

[21] Research on chemical components and pharmacological actions of Wangjianpin, Fu Xuchun and Alisma orientale Zhejiang medical science 2012

[22] Study of chemical components of rhizoma alismatis, Chinese herbal medicine 2008

[23]Bai B,Li MJ,Wang Y,Liu XH.Studies on chemical constituents of Dioscorea opposita.Zhongguo Zhong Yao Za Zhi.2008

[24]Huang,T.K.Handbook of Composition and Pharmacological Action of Commonly-used Traditional Chinese Medicine.Beijing,Publishing House of Medicine and Technology of China,1994.

[25]Ji,Y.B.Pharmacological Action and Application of Blood-activating and Stasis-eliminating Available Composition of Traditional Chinese Medicine.Harbin,Publishing House of Science and Technology of Hei Long Jiang,1999.

[26]Zhao,C.Q.,Y.L.Xu,et al.Available Antitumor Composition of Medicinal Plants.Beijing,Publishing House of Tradition Chinese Medicine of China,1997.

Claims

1. A network target-based intelligent and quantitative analysis method for pharmacology of a drug network is characterized by comprising the following steps:

A) based on a pre-constructed biological network and biological function multi-scale qualitative analysis result of the drug intervention disease, the quantitative analysis of the network intervention effect on the spatial dimension is carried out, and the method comprises the following steps:

B) based on the pre-constructed biological network and biological function multi-scale qualitative analysis result of the drug intervention disease, the method comprises the following steps:

wherein:

M_Ris the regulatory effect score of the R node by the T node,

a1, b1, a2, b2, a3 and b3 are undetermined parameters,

and

Wherein M is_t(i) At the ith time of tThe regulated effect score of each node, n represents the total number of nodes in the disease biological network,

for the shortest path between the R and T nodes,

C) according to the result of the predetermined biological function multi-scale qualitative analysis, the interference effect score NE of the disease biological network is determined by assigning weights to all the channels of the disease biological network and integrating the effect scores of the space dimension and the time dimension of the drug interference according to the mode represented by the formula (5):

wherein:

Beta is a time and space effect scale factor, the value range is 0-1,

wherein:

the pre-constructed biological network and biological function multi-scale qualitative analysis result of the drug intervention disease is established by the following steps:

reading the input one or more specific disease related information, and collecting the information obtained by analysis and/or experiment:

qualitative data of disease-genes and/or gene products, and/or

Quantitative data of disease-genes and/or gene products,

a collection of disease-related biomolecules is obtained,

reading the input information related to one or more specific medicines, and collecting the information obtained by analysis or experiment:

drug-target and/or target profile qualitative data, and/or

Drug-target and/or target profile quantification data,

a set of drug targets is obtained,

based on the disease-related biomolecules and the drug targets, combining a multi-scale interaction relation comprising protein interaction and the relation between cells or biological signal transduction pathways to construct a multi-scale biological network of the disease intervened by the drug, and carrying out qualitative and/or quantitative analysis on the network effect of the disease intervened by the drug based on network topology and network dynamics,

outputting network effect including medicine intervention disease, medicine effect substance including medicine effect component and its combination, network regulation mechanism of medicine intervention disease, objective index of curative effect and clinical indication,

the result of the multi-scale qualitative analysis of the biological function comprises the following steps:

a group of related genes or gene products in a disease biological network and a target or a target spectrum of a drug intervention biological network are used as a gene set to be detected, a biological molecule contained in a biological pathway or a biological process in a public database is used as a biological function gene set,

identifying the association relationship among biological functions in a disease biological network intervened by a medicament, determining the association significance between a gene set to be tested and a biological function gene set by adopting statistical test according to the mode represented by the formula (6), establishing a keyword mapping table according to disease and medicament related biological function terms, matching the biological function terms, clustering the biological function terms with significant association according to matching scores, forming a plurality of key biological function modules in the disease biological network intervened by the medicament to obtain the association among the biological functions, and qualitatively analyzing the network intervention effect from the biological function in a multi-scale way,

(6) wherein S (A, B) is the matching score of the biological function terms, A, B are vectors of the biological function terms, n is the number of the most words in the biological function terms,

if the keywords in A are directly contained in B, multiplying the matching score by the amplification weight omega of the keywords after the keywords appear_A,BValue range omega_A,B>1。

2. The network target-based pharmacological intelligent and quantitative analysis method for network targets according to claim 1, wherein the biofunctional multi-scale qualitative analysis and time-space multi-dimensional quantitative analysis comprises:

the qualitative data of the disease-gene and/or gene product comprises at least one selected from the group consisting of computational predictive data, database-collected data, experimental or clinical test data,

the quantitative data of the disease-gene and/or gene product comprises at least one selected from experimental data and clinical data such as gene expression profile, proteome, metabolome and the like,

the drug-target and/or target profile qualitative data comprises at least one selected from the group consisting of computational predictive data, database compiled data, experimental or clinical test data,

the drug-target and/or target spectrum quantitative data comprises at least one selected from gene expression spectrum, proteome, metabolome and other experimental data and clinical data under the action of the drug.

3. Use of the network target-based drug network pharmacology intelligence and quantification method of claim 1 or 2, comprising applying the network target-based drug network pharmacology intelligence and quantification method to one selected from the following drugs:

a prescription of a Chinese medicine,

the components of the traditional Chinese medicines are mixed,

the components of the traditional Chinese medicines are combined,

modern drugs including chemical and biological drugs,

combination of modern drugs.

4. A computer-readable storage medium storing a computer program capable of causing a processor to execute the network target-based drug network pharmacology intelligence and quantification analysis method of claim 1 or 2.

5. A network target-based intelligent and quantitative drug network pharmacology analysis system, comprising:

A) the part for carrying out network intervention effect quantification on a spatial dimension based on a pre-constructed biological network and biological function multi-scale qualitative analysis result of the drug intervention disease comprises the following steps:

wherein:

M_Ris the regulatory effect score of the R node by the T node,

a1, b1, a2, b2, a3 and b3 are undetermined parameters,

and

integrating the association of the drug target and the regulated molecule on the network according to the characterization mode of the formula (4), and determining the disease biological network intervention effect score NTE of the mth biological function module at the t moment in the time dimension_tmThe following components:

wherein M is_t(i) Is the regulated effect score of the ith node at the time t, n represents the total node number in the disease biological network,

for the shortest path between the R and T nodes,

wherein:

w(bp_m) Is composed ofThe weight of the mth biological function module is in a value range of 0-1

Beta is a time and space effect scale factor, the value range is 0-1,

wherein:

qualitative data of disease-genes and/or gene products, and/or

Quantitative data of disease-genes and/or gene products,

a collection of disease-related biomolecules is obtained,

drug-target and/or target profile qualitative data, and/or

Drug-target and/or target profile quantification data,

a set of drug targets is obtained,

6. The network target-based drug network pharmacology intelligence and quantitative analysis system of claim 5, wherein the biofunctional multi-scale qualitative analysis and temporal-spatial multi-dimensional quantitative analysis comprises: