CN111564186A - Method and system for predicting interaction of graph-volume drug pairs based on knowledge graph - Google Patents
Method and system for predicting interaction of graph-volume drug pairs based on knowledge graph Download PDFInfo
- Publication number
- CN111564186A CN111564186A CN202010216234.5A CN202010216234A CN111564186A CN 111564186 A CN111564186 A CN 111564186A CN 202010216234 A CN202010216234 A CN 202010216234A CN 111564186 A CN111564186 A CN 111564186A
- Authority
- CN
- China
- Prior art keywords
- drug
- gcn
- prediction model
- pair
- training
- 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.)
- Pending
Links
- 239000003814 drug Substances 0.000 title claims abstract description 165
- 229940079593 drug Drugs 0.000 title claims abstract description 133
- 230000003993 interaction Effects 0.000 title claims abstract description 94
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000012549 training Methods 0.000 claims abstract description 50
- 238000012360 testing method Methods 0.000 claims abstract description 23
- 230000006870 function Effects 0.000 claims abstract description 10
- 238000012795 verification Methods 0.000 claims abstract description 6
- 238000000605 extraction Methods 0.000 claims description 15
- 238000010276 construction Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 238000010200 validation analysis Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 abstract description 7
- 238000010801 machine learning Methods 0.000 description 4
- 230000008406 drug-drug interaction Effects 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 238000002790 cross-validation Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011056 performance test Methods 0.000 description 2
- 206010013710 Drug interaction Diseases 0.000 description 1
- 208000030453 Drug-Related Side Effects and Adverse reaction Diseases 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000012733 comparative method Methods 0.000 description 1
- 238000009509 drug development Methods 0.000 description 1
- 238000007876 drug discovery Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 229940126586 small molecule drug Drugs 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/30—Prediction of properties of chemical compounds, compositions or mixtures
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/50—Molecular design, e.g. of drugs
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/70—Machine learning, data mining or chemometrics
Landscapes
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Crystallography & Structural Chemistry (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computing Systems (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- Medicinal Chemistry (AREA)
- Pharmacology & Pharmacy (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention provides a method for predicting interaction of graph-rolled drug pairs based on a knowledge graph, which comprises the following steps: extracting a drug pair data sample to generate a data set containing a training set, a verification set and a test set; constructing a knowledge graph corresponding to the data set; establishing a GCN drug pair interaction prediction model, and learning the feature information of drugs contained in a drug pair and the neighborhood thereof; inputting the drug pair data samples of the training set into a GCN drug pair interaction prediction model, and training the GCN drug pair interaction prediction model; optimizing a loss function according to a training result, and then sending the training result into a GCN drug pair interaction prediction model for training; training the GCN medicament on an interaction prediction model through iterative calculation; inputting the drug pair data samples in the test set into a GCN drug pair interaction prediction model to obtain a test result; and analyzing the test result to obtain a prediction result. The method and the system for predicting the interaction of the atlas-based atlas medicament pair have high accuracy and short training time.
Description
[ technical field ] A method for producing a semiconductor device
The invention relates to the technical field of medicine pair interaction prediction, in particular to a knowledge graph-based method and a knowledge graph-based system for predicting interaction of a convolved medicine pair.
[ background of the invention ]
Drug development is a high-investment and high-risk field, the average time from development to marketing of a drug is 10-15 years, the average investment is about 26 hundred million dollars, and only one of 10000 molecules entering a development pipeline can be successfully developed on average per 5000-. The cost can be reduced to a great extent by adopting a calculation method, and the method is used as a hotspot for auxiliary calculation in the field of drug discovery by depending on a data-driven machine learning method and training a model through a large amount of open-source experimental data and predicting downstream tasks.
In the related art, machine learning algorithms are widely used for prediction of drug-drug interaction. Most existing artificial intelligence calculation models generally need to integrate drug information of a multi-source data set, such as various information (structure, property and the like) of drugs, part of methods focus on representation learning of drug characteristics, structural characteristics of the drugs are acquired through SMILES molecules, and interaction among the drugs is predicted through structural similarity; or integrating drug side effects and other related relations, some methods predict drug-drug interactions by constructing various relation networks and combining effective graph embedding methods. These methods have difficulty obtaining reasonably distributed data samples for batch training due to the rarity of the types of label data available from the data set. Meanwhile, the current mainstream methods only consider a single relationship between drugs, namely drug-drug interaction, and ignore potential association existing between drugs and other entities (such as targets and genes).
The knowledge map contains rich entities and associated information, so that the possibility of exploring potential association of the medicine and other entities is provided. Meanwhile, due to the obvious effect of the graph neural network in many fields, the effective aggregation method can be designed to effectively represent the node characteristics and the neighborhood information in the graph by iteratively extracting the neighborhood information in the graph, and the potential inter-entity relation in the knowledge graph can be captured better.
Therefore, it is necessary to provide a method and a system for predicting drug pair interaction based on atlas of knowledge to solve the above problems.
[ summary of the invention ]
Aiming at the technical problems to be solved, the invention provides a method and a system for predicting the interaction of the atlas-based drug pair based on the knowledge graph, which have high accuracy and short training time consumption.
The invention provides a method for predicting interaction of graph-rolled drug pairs based on a knowledge graph, which comprises the following steps:
s1, extracting a drug pair data sample to generate a data set, wherein the data set comprises a training set, a verification set and a test set;
s2: constructing a knowledge graph corresponding to the data set;
s3, establishing a GCN drug pair interaction prediction model, and learning the feature information of the drugs contained in the drug pairs and the neighborhood thereof by using the knowledge graph;
s4, inputting the drug pair data samples in the training set into the GCN drug pair interaction prediction model, and training the GCN drug pair interaction prediction model;
s5, the training result of the GCN drug pair interaction prediction model is sent to the GCN drug pair interaction prediction model again to train after optimizing a loss function;
s6, completing the training of the GCN medicament on an interaction prediction model through multiple iterative computations;
s7: inputting the drug pair data samples in the test set into the GCN drug pair interaction prediction model to obtain a test result;
s8: and analyzing the test result to obtain a prediction result of the interaction of the medicament.
Preferably, the proportion of the training set in the data set is 80%, the proportion of the validation set is 10%, and the proportion of the test set is 10%.
Preferably, the step S2 includes the steps of:
s21: downloading a compressed file of each original data in the data set from a website;
s22: converting the compressed file into RDF graph data using a Bio2RDF tool;
s23: and uploading the RDF graph data to an RDF triple memory, and extracting the selected triple by executing reference operation to obtain a knowledge graph of the corresponding data in the data set.
Preferably, the GCN drug pair interaction prediction model comprises an embedded layer and two GCN layers arranged in sequence.
Preferably, the knowledge-graph is input to the GCN layer for neighborhood information clustering operations to train the GCN drug pair interaction prediction model.
Preferably, the step S4 specifically includes: and sequentially inputting the drug pair data samples in the training set into the embedding layer and the GCN layer for processing.
Preferably, the step S4 includes the following steps:
s41: respectively inputting drug entities in a drug pair data sample of the training set to the embedding layer to obtain a random initialization vector;
s42: sending the initialization vector output by the embedding layer into the GCN layer;
s43: and obtaining a scoring result of the medicine pair through neighborhood sampling and clustering operation of the GCN layer.
Preferably, in step S5, specifically, the method includes: and sending the training result of the GCN layer to a classifier, and continuously sending the training result to the embedding layer after the classifier optimizes a loss function so as to continuously train the GCN drug pair interaction prediction model.
Preferably, in step S6, the GCN drug pair interaction prediction model is trained when 50 iterations are performed or when the results of 5 consecutive iterations are not changed.
The invention also provides a system for predicting the interaction of the map-volume drug pair based on the knowledge graph, which comprises the following steps:
the drug pair extraction module is used for extracting drug pair data samples and generating a data set, wherein the data set comprises a training set, a verification set and a test set;
the knowledge graph building module is used for building a knowledge graph corresponding to the data set, and the knowledge graph comprises entities, relations and a triple set;
the GCN drug pair interaction prediction model is used for learning the feature information of drugs and neighborhoods thereof and is arranged at the output ends of the knowledge graph construction module and the drug pair extraction module, the GCN drug pair interaction prediction model comprises an embedded layer and two GCN layers which are sequentially arranged, and the embedded layer is arranged at the output ends of the knowledge graph construction module and the drug pair extraction module; and
and the classifier is used for generating output labels for task classification, and is arranged at the output end of the GCN layer.
Compared with the related technology, the invention provides a method and a system for predicting interaction of a medicine pair based on a knowledge graph volume, wherein the medicine pair in a training set of a medicine pair extraction module is input into a GCN medicine pair interaction prediction model, and the knowledge graph generated by a knowledge graph construction module is input into the GCN medicine pair interaction prediction model to learn medicines and neighborhood information thereof; in addition, by comparing several famous machine learning models for predicting the interaction of the drug pair, the performance of the graph-volume drug pair interaction prediction method and system based on the knowledge graph is superior to that of other comparison models in the performance of a drug bank data set, and the requirements of high stability and high accuracy can be met; the method and the system for predicting the interaction of the atlas-based atlas medicament pair have the advantages of high accuracy and short training time.
[ description of the drawings ]
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without inventive efforts, wherein:
FIG. 1 is a flow chart of a method for predicting the interaction of a mapped product drug pair based on a knowledge-graph according to the present invention;
FIG. 2 is a block diagram of a system for predicting drug pair interactions based on knowledge-graph convolution as provided by the present invention;
fig. 3 is a schematic diagram of the operation of the GCN layer shown in fig. 2.
[ detailed description ] embodiments
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, the present invention provides a method for predicting interaction between a volume drug pair based on a knowledge graph, comprising the following steps:
and S1, extracting the drug pair data samples to generate a data set, wherein the data set comprises a training set, a verification set and a test set.
The proportion of the training set in the data set is 80%, the proportion of the validation set is 10% and the proportion of the test set is 10%. Drug pairs in the dataset that have an interaction are labeled as 1 and set as positive samples, and drug pairs that do not have an interaction are labeled as 0 and set as negative samples. Performing 5-fold cross validation processing on the data set to obtain the training set, the test set and the validation set;
s2: and constructing a knowledge graph corresponding to the data set.
The invention adopts a knowledge map construction module to generate a knowledge map corresponding to each data set. The step S2 specifically includes:
s21: downloading a compressed file of each original data in the data set from a website;
s22: converting the compressed file into RDF graph data using a Bio2RDF tool;
s23: and uploading the RDF graph data to an RDF triple memory, and extracting selected triples by executing benchmark operation to obtain a knowledge graph of the data corresponding to the data set.
S3, establishing a GCN (Graph connected Network) medicine pair interaction prediction model, and learning the medicine contained in the medicine pair and the characteristic information of the neighborhood thereof by using the knowledge Graph.
In the invention, a drug pair extraction module is used for generating a data sample required by a model; the GCN drug pair interaction prediction model is used for learning the feature information of drugs and neighborhoods thereof and is arranged at the output ends of the knowledge map construction module and the drug pair extraction module, and comprises an embedded layer and two GCN layers. The embedding layer is arranged at the output end of the medicine pair extraction module, and the GCN layer is arranged at the output end of the embedding layer.
In the invention, a classifier is adopted to generate an output label of task classification, which is arranged at the output end of a GCN drug pair interaction prediction model.
And S4, inputting the drug pair data samples in the training set into the GCN drug pair interaction prediction model, and training the GCN drug pair interaction prediction model.
The drug pair data samples are sequentially input to the embedding layer and the GCN layer for processing so as to train an interaction prediction model of the GCN drug pair.
Wherein the training process is as follows:
s41: respectively inputting drug entities in a drug pair data sample of the training set to the embedding layer to obtain a random initialization vector;
s42: sending the initialization vector output by the embedding layer into the GCN layer;
s43: and obtaining a scoring result of the medicine pair through neighborhood sampling and clustering operation of the GCN layer.
Specifically, the GCN layer includes neighborhood sampling and clustering operations, and specifically, each GCN layer samples a neighborhood of a drug node in a neighborhood range of a fixed size, and then performs clustering operations to obtain a final drug and a neighborhood representation vector thereof, specifically including:
sum clustering:
concat clustering:
wherein e is expressed as a feature vector of the drug entity,
the feature vector of the neighbor entity of the ith drug in the neighborhood range of S (e), sigma represents an activation function, W represents a learnable weight matrix, and b is a bias.
And S5, the training result of the GCN drug pair interaction prediction model is optimized by a loss function and then is fed into the GCN drug pair interaction prediction model again for training.
Specifically, the training result of the GCN layer is sent to a classifier, and the classifier optimizes the loss function and then continues to send the result to the embedding layer to continue training the GCN drug pair interaction prediction model.
Preferably, binary cross entropy is used as a loss function to calculate the probability value of the classification result
Comparing with the previous original label yi, j, the objective function LOSS can be obtained as:
and S6, completing the training of the GCN medicament on an interaction prediction model through multiple iterative calculations.
And (5) performing 50 iterative calculations or completing training of the GCN drug pair interaction prediction model when the result of 5 continuous iterative calculations is not changed any more.
S7: and inputting the drug pair data samples in the test set into the GCN drug pair interaction prediction model to obtain a test result.
S8: and analyzing the test result to obtain a prediction result of the interaction of the medicament.
Referring to fig. 2 and fig. 3, the present invention further provides a system 100 for predicting drug pair interaction based on atlas of knowledge, including a drug pair extraction module 10, an atlas construction module 20, a GCN drug pair interaction prediction model 30, and a classifier 40, wherein:
the drug pair extraction module 10 is used for generating data samples required by the model;
the knowledge graph building module 20 is configured to generate a knowledge graph corresponding to each data set;
the GCN drug pair interaction prediction model 30 is used for learning the feature information of the drugs and their neighborhoods, and is arranged at the output end of the knowledge map construction module 20 and the drug pair extraction module 10, and the GCN drug pair interaction prediction model 30 includes an embedded layer 310 and two GCN layers 320;
the classifier 40 is used for generating output labels of task classification, and is arranged at the output end of the GCN drug pair interaction prediction model 30.
Specifically, the embedding layer 310 is disposed at the output end of the pair of drugs extraction module 10 and the knowledge graph construction module 20, and the classifier 40 is disposed at the output end of the GCN layer 320.
In the following, the interaction prediction method and system based on the atlas will be evaluated for performance.
It should be noted that the data set adopted in this embodiment is a drug bank data set (V5.1.4, published in 2019, 7, month 2), and the interaction prediction method and system based on the atlas of knowledge base map volume drug pair are subjected to performance evaluation. The data set included 2578 experimentally-proven small molecule drugs, and a unique 612388 drug pairs screened from 13339 drugs that were experimentally validated. The data set and constructed knowledge-graph statistics are shown in table 1 below.
TABLE 1 DrugBank data set and constructed knowledge graph statistics table
Aiming at the performance evaluation of the interaction prediction system based on the atlas chart and the convolution medicine, in the embodiment, the prediction experiment of the interaction between medicines is carried out on potential medicine pairs in the drug bank data set. In this set of experiments, the scores of ACC, AUPR, AUC-ROC and F1 were used as evaluation indexes of model performance, and the 5-fold cross-validation comparison experiment was performed with 9 models of 5 mainstream methods.
Laplacian, GreRep, Deepwalk, Struc2vec, LINE, SDNE, GAE in the comparative methods, all performed based on BioNEV (https:// github. com/xiangyue9607/BioNEV), experimental parameter settings were kept default.
Regarding the deep ddi method, we rebuild the DNN model and modify the multi-class execution into two classes, with the source code address:
(https://bitbucket.org/kaistsystemsbiology/deepddi/src/master/)。
with regard to the KG-ddi method, we re-execute RDF2Vec to generate a 300-dimensional embedded vector, with the source code address:
(https://github.com/rezacsedu/Drug-drug-interaction-prediction)。
the specific experimental parameter settings provided by the present invention are shown in table 2.
TABLE 2 Superparameter settings for experimental execution
| Parameter(s) | Is provided with |
| Batch size | 4096 |
| Learning rate | 1e-2 |
| L2 weight value | 1e-7 |
| Size of vector dimension | 32 |
| Number of |
2 |
| Neighborhood size | 16 |
The results of the performance tests of the method of the present invention on the drug bank data set are shown in table 3.
TABLE 3 Performance test results on DrugBank data set by the method of the present invention
From the experimental results of table 3, it can be seen that: the test performance of the knowledge-graph-based drug pair interaction prediction system provided by the invention on a drug bank data set is superior to that of other comparison methods, and particularly, compared with all comparison methods, the method provided by the invention has at least 11.18%, 6.99%, 6.51% and 11% performance improvement on scores of ACC, AUPR, AUC-ROC and F1 respectively, so that a high-standard classification effect is realized.
Compared with the related technology, the invention provides a method and a system for predicting interaction of a medicine pair based on a knowledge graph volume, wherein the medicine pair in a training set of a medicine pair extraction module is input into a GCN medicine pair interaction prediction model, and the knowledge graph generated by a knowledge graph construction module is input into the GCN medicine pair interaction prediction model to learn medicines and neighborhood information thereof; in addition, by comparing several famous machine learning models for predicting the interaction of the drug pair, the performance of the graph-volume drug pair interaction prediction method and system based on the knowledge graph is superior to that of other comparison models in the performance of a drug bank data set, and the requirements of high stability and high accuracy can be met; the method and the system for predicting the interaction of the atlas-based atlas medicament pair have the advantages of high accuracy and short training time.
While the foregoing is directed to embodiments of the present invention, it will be understood by those skilled in the art that various changes may be made without departing from the spirit and scope of the invention.
Claims (10)
1. A method for predicting interaction of graph-rolled drug pairs based on a knowledge graph is characterized by comprising the following steps:
s1, extracting a drug pair data sample to generate a data set, wherein the data set comprises a training set, a verification set and a test set;
s2: constructing a knowledge graph corresponding to the data set;
s3, establishing a GCN drug pair interaction prediction model, and learning the feature information of the drugs contained in the drug pairs and the neighborhood thereof by using the knowledge graph;
s4, inputting the drug pair data samples in the training set into the GCN drug pair interaction prediction model, and training the GCN drug pair interaction prediction model;
s5, the training result of the GCN drug pair interaction prediction model is sent to the GCN drug pair interaction prediction model again to train after optimizing a loss function;
s6, completing the training of the GCN medicament on an interaction prediction model through multiple iterative computations;
s7: inputting the drug pair data samples in the test set into the GCN drug pair interaction prediction model to obtain a test result;
s8: and analyzing the test result to obtain a prediction result of the interaction of the medicament.
2. The method of predicting drug-pair interaction based on knowledge-graph volume of claim 1, wherein the training set is 80% in the data set, the validation set is 10% in the data set, and the test set is 10% in the data set.
3. The method of predicting a drug-pair interaction based on a knowledge-graph convolution of claim 1, wherein the step S2 includes the steps of:
s21: downloading a compressed file of each original data in the data set from a website;
s22: converting the compressed file into RDF graph data using a Bio2RDF tool;
s23: and uploading the RDF graph data to an RDF triple memory, and extracting the selected triple by executing reference operation to obtain a knowledge graph of the corresponding data in the data set.
4. The method of claim 1, wherein the GCN drug pair interaction prediction model comprises an embedded layer and two GCN layers sequentially arranged.
5. The method of claim 4, wherein the knowledge-graph based atlas drug pair interaction prediction method is input to the GCN layer for neighborhood information clustering operations to train the GCN drug pair interaction prediction model.
6. The method for predicting the drug pair interaction based on the atlas of knowledge as claimed in claim 4, wherein the step S4 is specifically: and sequentially inputting the drug pair data samples in the training set into the embedding layer and the GCN layer for processing.
7. The method of predicting a drug-pair interaction based on a knowledge-graph convolution of claim 6, wherein the step S4 includes the steps of:
s41: respectively inputting drug entities in a drug pair data sample of the training set to the embedding layer to obtain a random initialization vector;
s42: sending the initialization vector output by the embedding layer into the GCN layer;
s43: and obtaining a scoring result of the medicine pair through neighborhood sampling and clustering operation of the GCN layer.
8. The method for predicting the drug pair interaction based on the atlas of knowledge as claimed in claim 1, wherein in the step S5 specifically: and sending the training result of the GCN layer to a classifier, and continuously sending the training result to the embedding layer after the classifier optimizes a loss function so as to continuously train the GCN drug pair interaction prediction model.
9. The method of predicting drug-pair interaction based on knowledge-graph convolution of claim 1, wherein in step S6, the GCN drug-pair interaction prediction model is trained when 50 iterations are performed or when the results of 5 consecutive iterations are not changed.
10. A system for predicting a drug-pair interaction based on a map of knowledge comprising:
the drug pair extraction module is used for extracting drug pair data samples and generating a data set, wherein the data set comprises a training set, a verification set and a test set;
the knowledge graph building module is used for building a knowledge graph corresponding to the data set, and the knowledge graph comprises entities, relations and a triple set;
the GCN drug pair interaction prediction model is used for learning the feature information of drugs and neighborhoods thereof and is arranged at the output ends of the knowledge graph construction module and the drug pair extraction module, the GCN drug pair interaction prediction model comprises an embedded layer and two GCN layers which are sequentially arranged, and the embedded layer is arranged at the output ends of the knowledge graph construction module and the drug pair extraction module; and
and the classifier is used for generating output labels for task classification, and is arranged at the output end of the GCN layer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010216234.5A CN111564186A (en) | 2020-03-25 | 2020-03-25 | Method and system for predicting interaction of graph-volume drug pairs based on knowledge graph |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010216234.5A CN111564186A (en) | 2020-03-25 | 2020-03-25 | Method and system for predicting interaction of graph-volume drug pairs based on knowledge graph |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111564186A true CN111564186A (en) | 2020-08-21 |
Family
ID=72070333
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010216234.5A Pending CN111564186A (en) | 2020-03-25 | 2020-03-25 | Method and system for predicting interaction of graph-volume drug pairs based on knowledge graph |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111564186A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112599186A (en) * | 2020-12-30 | 2021-04-02 | 兰州大学 | Compound target protein binding prediction method based on multi-depth learning model consensus |
| CN113012770A (en) * | 2021-03-17 | 2021-06-22 | 中南大学 | Medicine-medicine interaction event prediction method, system, terminal and readable storage medium based on multi-modal deep neural network |
| CN113140254A (en) * | 2021-04-28 | 2021-07-20 | 厦门大学 | Meta-learning drug-target interaction prediction system and prediction method |
| CN114242162A (en) * | 2021-07-29 | 2022-03-25 | 北京百度网讯科技有限公司 | Method for establishing drug synergy prediction model, prediction method and corresponding device |
| CN114420221A (en) * | 2022-01-04 | 2022-04-29 | 湖南大学 | Knowledge graph-assisted multitask drug screening method and system |
| CN114708929A (en) * | 2022-04-08 | 2022-07-05 | 南京邮电大学 | Model for predicting interaction between medicine and training method thereof |
| WO2022203734A1 (en) * | 2021-03-25 | 2022-09-29 | Google Llc | Machine learning for predicting the properties of chemical formulations |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019107804A1 (en) * | 2017-12-01 | 2019-06-06 | 한국과학기술원 | Method for predicting drug-drug or drug-food interaction by using structural information of drug |
| CN110223751A (en) * | 2019-05-16 | 2019-09-10 | 平安科技（深圳）有限公司 | Prescription evaluation method, system and computer equipment based on medical knowledge map |
| CN110277143A (en) * | 2019-06-21 | 2019-09-24 | 沈阳药科大学 | A kind of important target spot evaluation method of drug based on weighted network |
| CN110289050A (en) * | 2019-05-30 | 2019-09-27 | 湖南大学 | A kind of drug based on figure convolution sum term vector-target interaction prediction method |
-
2020
- 2020-03-25 CN CN202010216234.5A patent/CN111564186A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019107804A1 (en) * | 2017-12-01 | 2019-06-06 | 한국과학기술원 | Method for predicting drug-drug or drug-food interaction by using structural information of drug |
| CN110223751A (en) * | 2019-05-16 | 2019-09-10 | 平安科技（深圳）有限公司 | Prescription evaluation method, system and computer equipment based on medical knowledge map |
| CN110289050A (en) * | 2019-05-30 | 2019-09-27 | 湖南大学 | A kind of drug based on figure convolution sum term vector-target interaction prediction method |
| CN110277143A (en) * | 2019-06-21 | 2019-09-24 | 沈阳药科大学 | A kind of important target spot evaluation method of drug based on weighted network |
Non-Patent Citations (1)
| Title |
|---|
| SOURAV DAS: "《Predicting Unknown Adverse Drug Reactions Using an Unsupervised Node Embedding Algorithm》", 《DEGREE OF MASTER AT MASSACHUSETTS INSTITUTE OR TECHNOLOGY》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112599186A (en) * | 2020-12-30 | 2021-04-02 | 兰州大学 | Compound target protein binding prediction method based on multi-depth learning model consensus |
| CN113012770A (en) * | 2021-03-17 | 2021-06-22 | 中南大学 | Medicine-medicine interaction event prediction method, system, terminal and readable storage medium based on multi-modal deep neural network |
| CN113012770B (en) * | 2021-03-17 | 2022-05-10 | 中南大学 | Multi-modal deep neural network based prediction of drug-drug interaction events |
| WO2022203734A1 (en) * | 2021-03-25 | 2022-09-29 | Google Llc | Machine learning for predicting the properties of chemical formulations |
| CN113140254A (en) * | 2021-04-28 | 2021-07-20 | 厦门大学 | Meta-learning drug-target interaction prediction system and prediction method |
| CN113140254B (en) * | 2021-04-28 | 2023-08-25 | 厦门大学 | Meta-learning drug-target interaction prediction system and prediction method |
| CN114242162A (en) * | 2021-07-29 | 2022-03-25 | 北京百度网讯科技有限公司 | Method for establishing drug synergy prediction model, prediction method and corresponding device |
| CN114242162B (en) * | 2021-07-29 | 2022-11-01 | 北京百度网讯科技有限公司 | Method for establishing drug synergy prediction model, prediction method and corresponding device |
| CN114420221A (en) * | 2022-01-04 | 2022-04-29 | 湖南大学 | Knowledge graph-assisted multitask drug screening method and system |
| CN114420221B (en) * | 2022-01-04 | 2024-06-25 | 湖南大学 | Knowledge graph-assisted multi-task drug screening method and system |
| CN114708929A (en) * | 2022-04-08 | 2022-07-05 | 南京邮电大学 | Model for predicting interaction between medicine and training method thereof |
| CN114708929B (en) * | 2022-04-08 | 2024-03-08 | 南京邮电大学 | Medicine-medicine interaction prediction model and training method thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111564186A (en) | Method and system for predicting interaction of graph-volume drug pairs based on knowledge graph | |
| Ma et al. | Autonomous deep learning: A genetic DCNN designer for image classification | |
| US11075008B2 (en) | Generating drug repositioning hypotheses based on integrating multiple aspects of drug similarity and disease similarity | |
| Elyan et al. | A genetic algorithm approach to optimising random forests applied to class engineered data | |
| US20190050368A1 (en) | Machine learning predictive labeling system | |
| CN112085565B (en) | Deep learning-based information recommendation method, device, equipment and storage medium | |
| Kwon et al. | Deepcci: End-to-end deep learning for chemical-chemical interaction prediction | |
| CN110929029A (en) | Text classification method and system based on graph convolution neural network | |
| Sathya et al. | [Retracted] Cancer Categorization Using Genetic Algorithm to Identify Biomarker Genes | |
| Pham et al. | Graph classification via deep learning with virtual nodes | |
| US20200142957A1 (en) | Learning property graph representations edge-by-edge | |
| US20140280257A1 (en) | Data Analysis Computer System and Method For Parallelized and Modularized Analysis of Big Data | |
| US20140278339A1 (en) | Computer System and Method That Determines Sample Size and Power Required For Complex Predictive and Causal Data Analysis | |
| Teng et al. | Customer credit scoring based on HMM/GMDH hybrid model | |
| CN113241115A (en) | Depth matrix decomposition-based circular RNA disease correlation prediction method | |
| Sarwar et al. | A survey of big data analytics in healthcare | |
| Potla | Scalable Machine Learning Algorithms for Big Data Analytics: Challenges and Opportunities | |
| Hadipour et al. | Deep clustering of small molecules at large-scale via variational autoencoder embedding and K-means | |
| CN115812210A (en) | Method and apparatus for enhancing performance of machine learning classification tasks | |
| Ma et al. | NinimHMDA: neural integration of neighborhood information on a multiplex heterogeneous network for multiple types of human microbe–disease association | |
| Teji et al. | Predicting missing links in gene regulatory networks using network embeddings: A qualitative assessment of selective embedding techniques | |
| Lee et al. | Deep single-cell RNA-seq data clustering with graph prototypical contrastive learning | |
| Pu et al. | An integrated network representation of multiple cancer-specific data for graph-based machine learning | |
| Bhadoria et al. | Bunch graph based dimensionality reduction using auto-encoder for character recognition | |
| Gao et al. | AiProAnnotator: Low-rank Approximation with network side information for high-performance, large-scale human Protein abnormality Annotator |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200821 |