CN117038105A - Drug repositioning method and system based on information enhancement graph neural network - Google Patents
Drug repositioning method and system based on information enhancement graph neural network Download PDFInfo
- Publication number
- CN117038105A CN117038105A CN202311291649.9A CN202311291649A CN117038105A CN 117038105 A CN117038105 A CN 117038105A CN 202311291649 A CN202311291649 A CN 202311291649A CN 117038105 A CN117038105 A CN 117038105A
- Authority
- CN
- China
- Prior art keywords
- information
- disease
- medicine
- drug
- node
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000009511 drug repositioning Methods 0.000 title claims abstract description 38
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 20
- 239000003814 drug Substances 0.000 claims abstract description 92
- 201000010099 disease Diseases 0.000 claims abstract description 87
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 87
- 239000013598 vector Substances 0.000 claims abstract description 53
- 229940079593 drug Drugs 0.000 claims abstract description 51
- 230000002776 aggregation Effects 0.000 claims abstract description 28
- 238000004220 aggregation Methods 0.000 claims abstract description 28
- 230000007246 mechanism Effects 0.000 claims abstract description 22
- 238000011176 pooling Methods 0.000 claims abstract description 22
- 230000008569 process Effects 0.000 claims abstract description 22
- 230000004931 aggregating effect Effects 0.000 claims abstract description 21
- 238000012545 processing Methods 0.000 claims description 23
- 230000006870 function Effects 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 7
- 238000013075 data extraction Methods 0.000 claims description 6
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 4
- 125000004122 cyclic group Chemical group 0.000 claims description 4
- 238000002347 injection Methods 0.000 claims description 3
- 239000007924 injection Substances 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 4
- 230000002708 enhancing effect Effects 0.000 abstract description 2
- 230000003993 interaction Effects 0.000 description 5
- 238000012512 characterization method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 239000000243 solution Substances 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 229940000406 drug candidate Drugs 0.000 description 1
- 230000008406 drug-drug interaction Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 239000002547 new drug Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 229940126585 therapeutic drug Drugs 0.000 description 1
- 230000003313 weakening effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H70/00—ICT specially adapted for the handling or processing of medical references
- G16H70/40—ICT specially adapted for the handling or processing of medical references relating to drugs, e.g. their side effects or intended usage
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Primary Health Care (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Epidemiology (AREA)
- Evolutionary Computation (AREA)
- Pathology (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Medicinal Chemistry (AREA)
- Databases & Information Systems (AREA)
- Pharmacology & Pharmacy (AREA)
- Toxicology (AREA)
- Chemical & Material Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The application discloses a drug repositioning method and a system based on an information enhancement graph neural network, belonging to a drug repositioning technology, comprising the following steps: constructing a priori knowledge model for representing a first relation between medicines and diseases, a second relation between different medicines and a third relation between different diseases, and carrying out information aggregation through K nearest neighbor, average pooling and graph annotation force mechanisms to obtain a first embedded vector of medicine nodes of the priori knowledge model and a second embedded vector of disease nodes; carrying out Hadamard product operation on the first embedded vector and the second embedded vector, obtaining the associated prediction between the medicine and the disease by utilizing a multi-layer perceptron MLP, and repositioning and positioning the medicine; in the information aggregation process, the fixed value is used for replacing the similarity score to define the aggregation coefficient of the neighbor node information, and the average pooling is used for aggregating the homogeneous information, so that the neighbor information is fully utilized, and the effect of enhancing the local information is realized.
Description
Technical Field
The application relates to the technical field of drug repositioning, in particular to a drug repositioning method and system based on an information enhancement graph neural network.
Background
Drug repositioning refers to the discovery of drug candidates for rare or no therapeutic drug diseases, and deep learning techniques have become one of the dominant techniques for drug repositioning. Generally, a deep learning-based drug repositioning model aims at effectively integrating various network structure information, so that high-quality characterization is learned for each disease and drug, and finally, the purpose of prediction is achieved.
Drugs and diseases typically constitute three networks, namely, a drug-drug network, a disease-disease network, and a drug disease-associated network, both of which contain rich structural information and one heterogeneous network. However, some of the information is important, some is not important and can even be regarded as noise information, so there is an urgent need to design a new drug repositioning technique that learns reliable characterization by differentiating between rich information.
Disclosure of Invention
In order to solve the problems, the application provides a drug repositioning method based on an information enhancement graph neural network, which comprises the following two stages:
and a data processing stage: constructing a priori knowledge model, wherein the priori knowledge model is used for representing a first relation between medicines and diseases, a second relation between different medicines and a third relation between different diseases, the first relation represents a correlation relation between medicines and diseases, the second relation represents a similarity relation between medicines, the third relation represents a similarity relation between diseases, and the priori knowledge model represents a model constructed by the prior knowledge of known medicines and diseases;
according to the prior knowledge model, information aggregation is carried out through K neighbor, average pooling and graph injection force mechanisms, a first embedded vector of a medicine node of the prior knowledge model and a second embedded vector of a disease node of the prior knowledge model are obtained, wherein the embedded vectors represent node feature vectors generated after aggregation is carried out through heterogeneous information and isomorphic information of the nodes;
prediction stage: and carrying out Hadamard product operation on the first embedded vector and the second embedded vector, and obtaining the associated prediction between the medicine and the disease by utilizing the multi-layer perceptron MLP to reposition the medicine.
Preferably, in the process of constructing the prior knowledge model in the data processing stage, each node information of the prior knowledge model is generated by acquiring heterogeneous information and neighborhood isomorphic information according to the first relationship, the second relationship and the third relationship.
Preferably, in the process of acquiring the first embedded vector in the data processing stage, based on the first relation and the third relation, according to the disease node corresponding to the medicine node, the first heterogeneous information is aggregated through a graph attention mechanism;
selecting the first K neighbor drug nodes of the drug nodes through average pooling, and aggregating the first homogeneous information;
and generating a first embedded vector of the drug node according to the first heterogeneous information and the first homogeneous information.
Preferably, in the process of acquiring the second embedded vector in the data processing stage, based on the first relation and the second relation, aggregating the second heterogeneous information through a graph attention mechanism according to the drug node corresponding to the disease node;
selecting the first K neighbor disease nodes of the disease nodes through average pooling, and aggregating second homogeneous information;
and generating a second embedded vector of the disease node according to the second heterogeneous information and the second homogeneous information.
Preferably, in the process of aggregating heterogeneous information in the data processing stage, different weights are adaptively allocated to different heterogeneous nodes through a graph attention mechanism, heterogeneous information is aggregated, a fixed value is used for defining an aggregation coefficient of neighbor node information instead of a similarity score, and average pooling is used for aggregating homogeneous information.
Preferably, during model training to reposition the drug, a binary cross entropy loss function is selected as the loss function of the model training process, and optimized through an Adam optimizer and a cyclic learning rate.
The application discloses a drug repositioning system based on an information enhancement graph neural network, which comprises the following components:
the data acquisition module is used for acquiring medicine information and disease information;
the data processing module is used for constructing a priori knowledge model according to the medicine information and the disease information, wherein the priori knowledge model is used for representing a first relationship between medicines and diseases, a second relationship between different medicines and a third relationship between different diseases;
the data extraction module is used for carrying out information aggregation according to the prior knowledge model through K neighbor, average pooling and graph annotation force mechanism to obtain a first embedded vector of a medicine node of the prior knowledge model and a second embedded vector of a disease node of the prior knowledge model;
and the medicine repositioning module is used for carrying out Hadamard product operation on the first embedded vector and the second embedded vector, acquiring the association prediction between the medicine and the disease by utilizing the multi-layer perceptron MLP, and repositioning the medicine.
Preferably, the data processing module is configured to obtain heterogeneous information and neighborhood isomorphic information by acquiring the first relationship, the second relationship and the third relationship, and generate each node information of the prior knowledge model.
Preferably, the data extraction module is further configured to adaptively allocate different weights to different heterogeneous nodes through a graph attention mechanism, aggregate heterogeneous information, replace a similarity score with a fixed value to define an aggregation coefficient of neighbor node information, and aggregate homogeneous information by using average pooling.
The application discloses the following technical effects:
according to the application, a graph annotation mechanism is introduced for the first time to distinguish the correlation between different heterogeneous node information of the medicine and the disease, and high-quality medicine and disease embedding is learned;
in the information aggregation process, the application omits the step of own node information aggregation, uses a fixed value to replace similarity score to define the aggregation coefficient of neighbor node information, uses average pooling to aggregate homogeneous information, fully utilizes neighbor information and realizes the effect of local information enhancement.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of a drug repositioning method according to the present application.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application more apparent, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. The components of the embodiments of the present application generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the application, as presented in the figures, is not intended to limit the scope of the application, as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by a person skilled in the art without making any inventive effort, are intended to be within the scope of the present application.
As shown in fig. 1, the application provides a drug repositioning method based on an information enhancement graph neural network, which comprises the following two stages:
and a data processing stage: constructing a priori knowledge model for representing a first relationship between drugs and diseases, a second relationship between different drugs and a third relationship between different diseases, wherein the first relationship represents a treatment relationship (representing a correlation relationship between drugs and diseases) of the diseases by drug treatment, the second relationship represents a similarity relationship between drugs, and the third relationship represents a similarity relationship between diseases, and the priori knowledge model represents a model constructed by known prior knowledge of drugs and diseases;
according to the prior knowledge model, information aggregation is carried out through K neighbor, average pooling and graph injection force mechanisms, a first embedded vector of a medicine node of the prior knowledge model and a second embedded vector of a disease node of the prior knowledge model are obtained, wherein the embedded vectors represent node feature vectors generated after aggregation is carried out through heterogeneous information and isomorphic information of the nodes;
prediction stage: and carrying out Hadamard product operation on the first embedded vector and the second embedded vector, and obtaining the associated prediction between the medicine and the disease by utilizing the multi-layer perceptron MLP to reposition the medicine.
Further preferably, in the drug repositioning method provided by the application, in the process of constructing the priori knowledge model in the data processing stage, each node information of the priori knowledge model is generated by acquiring heterogeneous information and neighborhood isomorphic information according to the first relationship, the second relationship and the third relationship.
Further preferably, in the drug repositioning method provided by the application, in the process of acquiring the first embedded vector in the data processing stage, based on the first relationship and the third relationship, the application aggregates the first heterogeneous information through a graph attention mechanism according to the disease node corresponding to the drug node;
selecting the first K neighbor drug nodes of the drug nodes through average pooling, and aggregating the first homogeneous information;
and generating a first embedded vector of the drug node according to the first heterogeneous information and the first homogeneous information.
Further preferably, in the drug repositioning method provided by the application, in the process of acquiring the second embedded vector in the data processing stage, based on the first relationship and the second relationship, the drug repositioning method aggregates the second heterogeneous information through a graph attention mechanism according to the drug node corresponding to the disease node;
selecting the first K neighbor disease nodes of the disease nodes through average pooling, and aggregating second homogeneous information;
and generating a second embedded vector of the disease node according to the second heterogeneous information and the second homogeneous information.
Further preferably, in the drug repositioning method provided by the application, in the process of aggregating heterogeneous information in a data processing stage, different weights are distributed for different heterogeneous nodes in a self-adaptive manner through a graph attention mechanism, the heterogeneous information is aggregated, a fixed value is used for replacing similarity scores to define an aggregation coefficient of neighbor node information, and average pooling is used for aggregating homogeneous information.
Further preferably, in the drug repositioning method provided by the application, in the process of model training for repositioning the drug, a binary cross entropy loss function is selected as a loss function in the model training process, and is optimized through an Adam optimizer and a cyclic learning rate.
The application also discloses a drug repositioning system based on the information enhancement graph neural network, which is used for realizing the drug repositioning method and comprises the following steps:
the data acquisition module is used for acquiring medicine information and disease information;
the data processing module is used for constructing a priori knowledge model according to the medicine information and the disease information, wherein the priori knowledge model is used for representing a first relationship between medicines and diseases, a second relationship between different medicines and a third relationship between different diseases;
the data extraction module is used for carrying out information aggregation according to the prior knowledge model through K neighbor, average pooling and graph annotation force mechanism to obtain a first embedded vector of a medicine node of the prior knowledge model and a second embedded vector of a disease node of the prior knowledge model;
and the medicine repositioning module is used for carrying out Hadamard product operation on the first embedded vector and the second embedded vector, acquiring the association prediction between the medicine and the disease by utilizing the multi-layer perceptron MLP, and repositioning the medicine.
Further preferably, the data processing module of the drug repositioning system disclosed by the application is used for obtaining heterogeneous information and neighborhood isomorphic information by obtaining the first relationship, the second relationship and the third relationship, and generating each node information of the priori knowledge model.
Still preferably, the data extraction module of the drug relocation system disclosed by the application is further used for adaptively distributing different weights for different heterogeneous nodes through a graph attention mechanism, aggregating heterogeneous information, defining aggregation coefficients of neighbor node information by replacing similarity scores with fixed values, and aggregating homogeneous information by using average pooling.
Example 1: aiming at the noise information problem existing in medicine repositioning, if each medicine and disease are regarded as nodes, the goal can be changed into distinguishing the correlation degree between different heterogeneous node pairs, different weights are distributed for different heterogeneous nodes in a self-adaptive mode, heterogeneous information aggregation is carried out according to the size of the weights, characterization is obtained, and errors caused by noise information are overcome.
As shown in fig. 1, the application provides a drug repositioning technology based on an information enhancement graph neural network. Specifically, a drug-drug similarity network, a disease-disease similarity network and a known drug-disease association network are first constructed to aggregate heterogeneous information and neighborhood isomorphic information to obtain more complete node information. Similar to some previous studies, attention is directed here to aggregating the information of the first K neighbors to avoid noise information from the neighborhood. Meanwhile, in the information aggregation process, a graph annotation meaning mechanism is introduced, and different correlation coefficients are distributed for different heterogeneous nodes in a self-adaptive mode, so that the target node is ensured to be capable of aggregating more effective information. Unlike the previous work, the present application does not consider the aggregation of own node information in the information aggregation process, and avoids that excessive information is embedded in a limited vector space, thereby weakening useful heterogeneous information and isomorphic information. In the prediction stage, hadamard product operation is firstly carried out on the medicine embedded vector and the disease embedded vector to fully fuse the information of the medicine embedded vector and the disease embedded vector, and then a multi-layer perceptron (MLP) is utilized to model complex medicine-disease association, so that final association prediction is obtained. The method specifically comprises the following steps:
(1) Disease modeling. For each disease, disease modeling learns the corresponding potential vector representations, namely drug-disease interactions and disease-disease interactions, by aggregating two interactions. In particular, heterogeneous information is aggregated by a drug node associated with the disease node; the homogeneity information is aggregated based on the first K neighbor nodes (disease nodes) of the disease node.
(2) Drug modeling. For each drug, drug modeling learns the corresponding potential vector representations, namely drug-disease interactions and drug-drug interactions, by aggregating the two interactions. In particular, heterogeneous information is aggregated by a disease node associated with the drug node; the homogeneity information is aggregated based on the top K neighbor nodes (drug nodes) of the drug node.
(3) In aggregating heterogeneous information, a graph attention mechanism is used to adaptively assign different weights to different heterogeneous nodes, thereby achieving high quality characterization. As for neighbor information aggregation, the application uses a fixed value to replace a similarity score to define the aggregation coefficient of neighbor node information, and uses average pooling to ensure a concise aggregation process and simultaneously retain the effect of enhancing neighbor information.
(4) The Hadamard product was used to fully integrate drug and disease representations.
(5) A multi-layer perceptron (MLP) is applied to model complex drug-disease associations, yielding final prediction results.
In model training, a binary cross entropy loss function is used, and to optimize this loss function, the application uses an Adam optimizer and a cyclic learning rate.
TABLE 1
In table 1, the meaning of english is that AUROC represents the area under the specificity-sensitivity curve, avg represents the average value, AUPR accuracy-recall ratio, MKGAT represents the graph meaning network model based on double laplace regularization least squares, IRNMF represents the drug repositioning model based on index regularization non-negative matrix factorization, SCPMF represents the drug repositioning model based on similarity constraint probability matrix factorization, MKGCN represents the multi-core fusion relationship prediction model on the graph neural network, DRWBNCF represents the weighted bilinear neural collaborative filtering drug repositioning model, DRGANN represents the drug repositioning model of the information enhancement graph neural network, and the application compares with 5 advanced models on the 3 data sets of data set 1 (Fdataset), data set 2 (Cdataset) and data set 3 (LRSSL) in order to demonstrate the superiority of the model of the application. AUROC and AUPR have been widely used in bioinformatics research and therefore are used to evaluate the overall performance of models. Table 1 shows the performance of the model of the application in 10 fold cross-validation versus other models, with 2 indices on 3 data sets being consistently better than all the comparative models, with average AUROC and AUPR of 0.947 and 0.571, respectively, 1.4% and 10.1% higher than the second, good model DRWBNCF.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In the description of the present application, it should be understood that the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more such feature. In the description of the present application, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (9)
1. The drug repositioning method based on the information enhancement graph neural network is characterized by comprising the following two stages:
and a data processing stage: constructing a priori knowledge model for representing a first relationship between a drug and a disease, a second relationship between different drugs and a third relationship between different diseases, wherein the first relationship represents a correlation relationship between the drug and the disease, the second relationship represents a similarity relationship between the drugs, the third relationship represents a similarity relationship between the diseases, and the priori knowledge model represents a model constructed by the prior knowledge of the known drugs and the diseases;
according to the prior knowledge model, information aggregation is carried out through K neighbor, average pooling and graph injection force mechanisms, a first embedded vector of a medicine node of the prior knowledge model and a second embedded vector of a disease node of the prior knowledge model are obtained, wherein the embedded vectors represent node feature vectors generated after aggregation through heterogeneous information and isomorphic information of the nodes;
prediction stage: and carrying out Hadamard product operation on the first embedded vector and the second embedded vector, and obtaining the association prediction between the medicine and the disease by utilizing a multi-layer perceptron MLP, so as to reposition the medicine.
2. The drug repositioning method based on the information enhancement graph neural network according to claim 1, wherein the drug repositioning method is characterized in that:
in the process of constructing a priori knowledge model in a data processing stage, according to the first relation, the second relation and the third relation, each node information of the priori knowledge model is generated by acquiring heterogeneous information and neighborhood isomorphic information.
3. The method for drug repositioning based on the information enhancement map neural network according to claim 2, wherein the method comprises the following steps:
in the process of acquiring a first embedded vector in a data processing stage, based on the first relation and the third relation, according to the disease node corresponding to the medicine node, first heterogeneous information is aggregated through the graph annotation force mechanism;
selecting the first K neighbor drug nodes of the drug nodes through the average pooling, and aggregating first homogeneous information;
and generating the first embedded vector of the medicine node according to the first heterogeneous information and the first homogeneous information.
4. A method for drug repositioning based on an information enhancement graph neural network according to claim 3, wherein:
in the process of acquiring a second embedded vector in a data processing stage, based on the first relation and the second relation, according to the medicine node corresponding to the disease node, second heterogeneous information is aggregated through the graph annotation force mechanism;
selecting the first K neighbor disease nodes of the disease nodes through the average pooling, and aggregating second homogeneous information;
and generating the second embedded vector of the disease node according to the second heterogeneous information and the second homogeneous information.
5. The method for drug repositioning based on the information enhancement map neural network of claim 4, wherein the method comprises the steps of:
in the process of aggregating heterogeneous information in a data processing stage, different weights are distributed for different heterogeneous nodes in a self-adaptive mode through the graph annotation mechanism, heterogeneous information is aggregated, a fixed value is used for replacing similarity scores to define aggregation coefficients of neighbor node information, and average pooling is used for aggregating homogeneous information.
6. The method for drug repositioning based on the information enhancement map neural network of claim 5, wherein the method comprises the steps of:
in the model training process of repositioning the medicine, a binary cross entropy loss function is selected as a loss function in the model training process, and optimization is carried out through an Adam optimizer and a cyclic learning rate.
7. A drug repositioning system based on an information enhancement map neural network, comprising:
the data acquisition module is used for acquiring medicine information and disease information;
the data processing module is used for constructing a priori knowledge model according to the medicine information and the disease information, wherein the priori knowledge model is used for representing a first relationship between medicines and diseases, a second relationship between different medicines and a third relationship between different diseases;
the data extraction module is used for carrying out information aggregation according to the prior knowledge model through K nearest neighbor, average pooling and graph annotation force mechanism, and obtaining a first embedded vector of a medicine node of the prior knowledge model and a second embedded vector of a disease node of the prior knowledge model;
and the medicine repositioning module is used for carrying out Hadamard product operation on the first embedded vector and the second embedded vector, acquiring the association prediction between the medicine and the disease by utilizing the multi-layer perceptron MLP and repositioning the medicine.
8. The drug repositioning system based on an information enhancement graph neural network of claim 7, wherein:
the data processing module is used for obtaining heterogeneous information and neighborhood isomorphic information by obtaining the first relation, the second relation and the third relation, and generating each node information of the priori knowledge model.
9. The drug repositioning system based on an information enhancement graph neural network of claim 8, wherein:
the data extraction module is further configured to adaptively allocate different weights to different heterogeneous nodes through the graph annotation mechanism, aggregate heterogeneous information, replace similarity scores with a fixed value to define aggregation coefficients of neighbor node information, and aggregate homogeneous information by using average pooling.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311291649.9A CN117038105B (en) | 2023-10-08 | 2023-10-08 | Drug repositioning method and system based on information enhancement graph neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311291649.9A CN117038105B (en) | 2023-10-08 | 2023-10-08 | Drug repositioning method and system based on information enhancement graph neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117038105A true CN117038105A (en) | 2023-11-10 |
| CN117038105B CN117038105B (en) | 2023-12-15 |
Family
ID=88635846
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311291649.9A Active CN117038105B (en) | 2023-10-08 | 2023-10-08 | Drug repositioning method and system based on information enhancement graph neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117038105B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117976139A (en) * | 2024-03-29 | 2024-05-03 | 武汉纺织大学 | Drug repositioning method and system based on deviation correcting mechanism and contrast learning |
Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109446338A (en) * | 2018-09-20 | 2019-03-08 | 大连交通大学 | Drug disease relationship classification method neural network based |
| CN111951886A (en) * | 2019-05-17 | 2020-11-17 | 天津科技大学 | Drug relocation prediction method based on Bayesian inductive matrix completion |
| CN112037856A (en) * | 2020-09-30 | 2020-12-04 | 华中农业大学 | Drug interaction and event prediction method and model based on attention neural network |
| CN112331275A (en) * | 2020-10-28 | 2021-02-05 | 东南大学 | Memory network and attention-based drug relocation calculation method |
| WO2021151325A1 (en) * | 2020-09-09 | 2021-08-05 | 平安科技（深圳）有限公司 | Method and apparatus for triage model training based on medical knowledge graphs, and device |
| WO2021179838A1 (en) * | 2020-03-10 | 2021-09-16 | 支付宝(杭州)信息技术有限公司 | Prediction method and system based on heterogeneous graph neural network model |
| CN114242186A (en) * | 2021-12-30 | 2022-03-25 | 湖南大学 | Chinese and western medicine relocation method and system fusing GHP and GCN and storage medium |
| CN114613452A (en) * | 2022-03-08 | 2022-06-10 | 电子科技大学 | Drug relocation method and system based on drug classification map neural network |
| CN114842927A (en) * | 2022-05-20 | 2022-08-02 | 华东师范大学 | Medicine and pathway association prediction method of knowledge graph attention network |
| CN114913916A (en) * | 2022-04-19 | 2022-08-16 | 广东工业大学 | Drug relocation method for predicting new coronavirus adaptive drugs |
| CN115293851A (en) * | 2022-08-12 | 2022-11-04 | 西安交通大学 | Recommendation method for introducing item category information into graph neural network |
| CN115376704A (en) * | 2022-09-15 | 2022-11-22 | 大连大学 | Medicine-disease interaction prediction method fusing multi-neighborhood correlation information |
| CN115527627A (en) * | 2022-10-08 | 2022-12-27 | 湖州师范学院 | Drug relocation method and system based on hypergraph convolutional neural network |
| CN115579053A (en) * | 2022-10-27 | 2023-01-06 | 京东方科技集团股份有限公司 | Drug relocation method, device, electronic device and storage medium |
| CN115858812A (en) * | 2022-05-16 | 2023-03-28 | 吉林大学 | Embedded alignment method constructed by computer |
| CN116189847A (en) * | 2023-05-05 | 2023-05-30 | 武汉纺织大学 | Safety medicine recommendation method based on LSTM-CNN strategy of attention mechanism |
| CN116386899A (en) * | 2023-02-15 | 2023-07-04 | 深圳大学 | Graph learning-based medicine disease association relation prediction method and related equipment |
-
2023
- 2023-10-08 CN CN202311291649.9A patent/CN117038105B/en active Active
Patent Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109446338A (en) * | 2018-09-20 | 2019-03-08 | 大连交通大学 | Drug disease relationship classification method neural network based |
| CN111951886A (en) * | 2019-05-17 | 2020-11-17 | 天津科技大学 | Drug relocation prediction method based on Bayesian inductive matrix completion |
| WO2021179838A1 (en) * | 2020-03-10 | 2021-09-16 | 支付宝(杭州)信息技术有限公司 | Prediction method and system based on heterogeneous graph neural network model |
| WO2021151325A1 (en) * | 2020-09-09 | 2021-08-05 | 平安科技（深圳）有限公司 | Method and apparatus for triage model training based on medical knowledge graphs, and device |
| CN112037856A (en) * | 2020-09-30 | 2020-12-04 | 华中农业大学 | Drug interaction and event prediction method and model based on attention neural network |
| CN112331275A (en) * | 2020-10-28 | 2021-02-05 | 东南大学 | Memory network and attention-based drug relocation calculation method |
| CN114242186A (en) * | 2021-12-30 | 2022-03-25 | 湖南大学 | Chinese and western medicine relocation method and system fusing GHP and GCN and storage medium |
| CN114613452A (en) * | 2022-03-08 | 2022-06-10 | 电子科技大学 | Drug relocation method and system based on drug classification map neural network |
| CN114913916A (en) * | 2022-04-19 | 2022-08-16 | 广东工业大学 | Drug relocation method for predicting new coronavirus adaptive drugs |
| CN115858812A (en) * | 2022-05-16 | 2023-03-28 | 吉林大学 | Embedded alignment method constructed by computer |
| CN114842927A (en) * | 2022-05-20 | 2022-08-02 | 华东师范大学 | Medicine and pathway association prediction method of knowledge graph attention network |
| CN115293851A (en) * | 2022-08-12 | 2022-11-04 | 西安交通大学 | Recommendation method for introducing item category information into graph neural network |
| CN115376704A (en) * | 2022-09-15 | 2022-11-22 | 大连大学 | Medicine-disease interaction prediction method fusing multi-neighborhood correlation information |
| CN115527627A (en) * | 2022-10-08 | 2022-12-27 | 湖州师范学院 | Drug relocation method and system based on hypergraph convolutional neural network |
| CN115579053A (en) * | 2022-10-27 | 2023-01-06 | 京东方科技集团股份有限公司 | Drug relocation method, device, electronic device and storage medium |
| CN116386899A (en) * | 2023-02-15 | 2023-07-04 | 深圳大学 | Graph learning-based medicine disease association relation prediction method and related equipment |
| CN116189847A (en) * | 2023-05-05 | 2023-05-30 | 武汉纺织大学 | Safety medicine recommendation method based on LSTM-CNN strategy of attention mechanism |
Non-Patent Citations (1)
| Title |
|---|
| 张志扬;张凤荔;陈学勤;王瑞锦;: "基于分层注意力的信息级联预测模型", 计算机科学, no. 06 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117976139A (en) * | 2024-03-29 | 2024-05-03 | 武汉纺织大学 | Drug repositioning method and system based on deviation correcting mechanism and contrast learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117038105B (en) | 2023-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gao et al. | MSCFNet: A lightweight network with multi-scale context fusion for real-time semantic segmentation | |
| CN111755078B (en) | Drug molecule attribute determination method, device and storage medium | |
| Hu et al. | 3D separable convolutional neural network for dynamic hand gesture recognition | |
| Min et al. | Deep learning in bioinformatics | |
| Song et al. | Unified binary generative adversarial network for image retrieval and compression | |
| EP3989109A1 (en) | Image identification method and device, identification model training method and device, and storage medium | |
| WO2020048445A1 (en) | End-to-end structure-aware convolutional networks for knowledge base completion | |
| WO2021098362A1 (en) | Video classification model construction method and apparatus, video classification method and apparatus, and device and medium | |
| CN108664999A (en) | A kind of training method and its device, computer server of disaggregated model | |
| CN117038105B (en) | Drug repositioning method and system based on information enhancement graph neural network | |
| CN109255340A (en) | It is a kind of to merge a variety of face identification methods for improving VGG network | |
| CN109242013A (en) | A kind of data mask method, device, electronic equipment and storage medium | |
| CN111564186A (en) | Method and system for predicting interaction of graph-volume drug pairs based on knowledge graph | |
| CN111125469A (en) | User clustering method and device for social network and computer equipment | |
| CN116386899A (en) | Graph learning-based medicine disease association relation prediction method and related equipment | |
| WO2023020214A1 (en) | Retrieval model training method and apparatus, retrieval method and apparatus, device and medium | |
| CN112668482A (en) | Face recognition training method and device, computer equipment and storage medium | |
| CN115600691A (en) | Client selection method, system, device and storage medium in federal learning | |
| CN114743133A (en) | Lightweight small sample video classification and identification method and system | |
| Bayoudh | A survey of multimodal hybrid deep learning for computer vision: Architectures, applications, trends, and challenges | |
| CN106033546B (en) | Behavior classification method based on top-down learning | |
| JP7473389B2 (en) | Learning model generation system and learning model generation method | |
| CN117575044A (en) | Data forgetting learning method, device, data processing system and storage medium | |
| Lu et al. | Soft-orthogonal constrained dual-stream encoder with self-supervised clustering network for brain functional connectivity data | |
| Feng et al. | Power law in deep neural networks: Sparse network generation and continual learning with preferential attachment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |