CN110993037A - Protein activity prediction device based on multi-view classification model - Google Patents
Protein activity prediction device based on multi-view classification model Download PDFInfo
- Publication number
- CN110993037A CN110993037A CN201911031041.6A CN201911031041A CN110993037A CN 110993037 A CN110993037 A CN 110993037A CN 201911031041 A CN201911031041 A CN 201911031041A CN 110993037 A CN110993037 A CN 110993037A
- Authority
- CN
- China
- Prior art keywords
- graph
- node
- matrix
- protein
- classification model
- 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
- 238000013145 classification model Methods 0.000 title claims abstract description 26
- 230000004952 protein activity Effects 0.000 title claims abstract description 23
- 239000013598 vector Substances 0.000 claims abstract description 30
- 238000011176 pooling Methods 0.000 claims abstract description 24
- 108090000623 proteins and genes Proteins 0.000 claims abstract description 24
- 102000004169 proteins and genes Human genes 0.000 claims abstract description 24
- 230000004927 fusion Effects 0.000 claims abstract description 15
- 101710192597 Protein map Proteins 0.000 claims abstract description 13
- 238000004590 computer program Methods 0.000 claims abstract description 8
- 239000000126 substance Substances 0.000 claims abstract description 7
- 239000011159 matrix material Substances 0.000 claims description 34
- 238000000034 method Methods 0.000 claims description 8
- 230000002776 aggregation Effects 0.000 claims description 4
- 238000004220 aggregation Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 238000012549 training Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 10
- 150000001875 compounds Chemical group 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 206010007269 Carcinogenicity Diseases 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000004931 aggregating effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000007670 carcinogenicity Effects 0.000 description 1
- 231100000260 carcinogenicity Toxicity 0.000 description 1
- 238000006555 catalytic reaction Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000010076 replication Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000547 structure data Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000012546 transfer Methods 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06N3/08—Learning methods
-
- 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
- G16C20/80—Data visualisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Chemical & Material Sciences (AREA)
- Crystallography & Structural Chemistry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
本发明公开了一种基于多视图分类模型的蛋白质活性预测装置，包括计算机存储器、计算机处理器以及计算机程序，算机存储器中存有训练好多视图分类模型，所述多视图分类模型包括池化单元、图卷积单元和全连接单元；计算机处理器执行计算机程序时实现以下步骤：将蛋白质结构的原子作为节点，原子之间的化学键作为连边，构建表示蛋白质结构的原始蛋白质图；利用池化单元对原始蛋白质图进行池化操作，获得不同维度的多个视图；利用训练好的图卷积单元分别对多个视图和原始蛋白质图进行图卷积操作，获得多个特征向量；对多个特征向量进行级联获得融合特征向量，利用训练好的全连接单元对融合特征向量进行全连接操作，输出蛋白质活性预测结果。
The invention discloses a protein activity prediction device based on a multi-view classification model, comprising a computer memory, a computer processor and a computer program. The computer memory stores a training multi-view classification model, and the multi-view classification model includes a pooling unit , graph convolution unit and fully connected unit; the computer processor implements the following steps when executing the computer program: using the atoms of the protein structure as nodes and the chemical bonds between atoms as edges to construct the original protein graph representing the protein structure; using pooling The unit performs a pooling operation on the original protein map to obtain multiple views of different dimensions; uses the trained graph convolution unit to perform a graph convolution operation on multiple views and the original protein map respectively to obtain multiple feature vectors; The feature vectors are cascaded to obtain the fusion feature vector, and the fully connected unit is used to perform the full connection operation on the fusion feature vector to output the protein activity prediction result.
Description
Technical Field
The invention belongs to the field of bioinformatics, and particularly relates to a protein activity prediction device based on a multi-view classification model.
Background
A graph consists of a series of nodes and con-vertions. Graph analysis is applied to research directions such as node classification, link prediction, clustering, and graph classification as a kind of non-euclidean data. Graph neural networks (Graph neural networks) are a deep learning method based on Graph domain analysis, and rely on information transfer between nodes in a Graph to capture dependency relationships in the Graph, so as to obtain characteristics of each node. The node features it generates can be used as input to any differentiable prediction layer to train the entire model in an end-to-end fashion.
Some of the studies on GNN focus on feature Learning at the node level (see document 1: Hamilton W L, YingR, Leskovec J. inductive replication Learning on Large Graphs [ J ]. 2017.; i.e., generalized Representation Learning of Large Networks), while others focus on feature Learning at the Graph level (see document 2: Henaff M, Bruna J, Lecun Y. deep computational Networks on Graph-Structured Data [ J ]. Computer science, 2015.; i.e., deep convolution Networks of Graph structure Data). The method for learning the characteristics of the concerned nodes brings improvement to tasks such as node classification and link prediction, and the method for learning the characteristics of the graph mainly promotes graph classification and the like. The task of graph classification is to predict the labels of a given graph using the relevant features of the nodes and the topology of the graph.
When applying GNN to graph classification, the standard approach is to generate features for all nodes in the graph and then to globally assemble all these node features together, such as David Duvenaud's implementation to input molecular structures of arbitrary size and shape into a deep neural network to get a molecular fingerprint vector of fixed dimensions (see document 3: Duvenaud D, Maclaurin D, agiuilera-iparaguerre J. relational network on Graphs for learning molecular Fingerprints 2015.) these methods treat all nodes equally when using node features. In other words, the structural information of the entire graph is completely ignored in this process, and by default, the importance of each node in the graph is the same. But in practice different nodes have different degrees of importance in the graph, and they should contribute differently to the graph characteristics.
Disclosure of Invention
The invention aims to provide a protein activity prediction device based on a multi-view classification model, and the compound function prediction device can realize the prediction of the functions of compounds based on the multi-view classification model.
In order to achieve the above object, the present invention provides a compound function prediction device based on a multi-view classification model, comprising a computer memory, a computer processor and a computer program stored in the computer memory and executable on the computer processor, wherein the computer memory stores a trained multi-view classification model, and the multi-view classification model comprises a pooling unit, a graph volume unit and a full connection unit;
the computer processor, when executing the computer program, performs the steps of:
constructing an original protein graph representing a protein structure by taking atoms of the protein structure as nodes and chemical bonds between the atoms as connecting edges;
performing pooling operation on the original protein map by using a pooling unit to obtain a plurality of views with different dimensions;
performing graph convolution operation on the multiple views and the original protein graph by using a trained graph convolution unit to obtain multiple feature vectors;
and cascading the plurality of feature vectors to obtain a fusion feature vector, performing full-connection operation on the fusion feature vector by using the trained full-connection unit, and outputting a protein activity prediction result.
The invention has the technical effects that:
the invention skillfully converts the protein structure into the graph structure, realizes the prediction of the protein activity by extracting and classifying the graph structure characteristics, mainly utilizes the graph convolution graph network to learn to obtain the node characteristics containing the attribute characteristics and the topological structure, classifies and aggregates part of nodes into a cluster according to the node characteristics, performs pooling dimensionality reduction on the protein graph, then performs weighted summation on the node characteristics of each view, converts the node characteristics into the graph characteristics and performs characteristic fusion, and finally outputs the class mark of the predicted graph through full connection, namely the prediction of the protein activity. The multi-view classification model not only considers the spatial structure and node attribute characteristics of the initial graph, but also considers the influence of the deeper graph structure obtained from the initial graph on the prediction result, so that the accuracy of protein activity prediction is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of protein activity prediction based on a multi-view classification model provided by the examples;
fig. 2 is an illustrative diagram of a new node connectivity representation.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the detailed description and specific examples, while indicating the scope of the invention, are intended for purposes of illustration only and are not intended to limit the scope of the invention.
In bioinformatics, compounds are formed by connecting atoms through chemical bonds, and can be converted into graph networks by combining the existing graph networks, and some properties of the compounds are researched by utilizing the idea of processing the graph networks.
The embodiment provides a protein activity prediction device based on a multi-view classification model, which comprises a computer memory, a computer processor and a computer program stored in the computer memory and executable on the computer processor, wherein the computer memory stores a trained multi-view classification model, and the multi-view classification model comprises a pooling unit, a graph volume unit and a full connection unit.
The graph convolution unit mainly comprises a plurality of convolution layers and is mainly used for extracting features of an input graph, learning a local topological structure and self attributes of each node on the graph, obtaining feature vectors with the same dimensionality and storing the feature vectors in a feature matrix form;
the pooling unit is used for classifying and aggregating nodes with similar characteristics and attributes in the initial graph to form new nodes, then acquiring a linkage relation between the new nodes according to the mapping between the nodes of the initial graph and the new nodes, reducing the number of the nodes and the linkage edges of the obtained new graph to a certain extent, realizing low-dimensional representation of the initial graph, and making the constructed new graph capable of serving as the initial graph to complement representation forms and characteristics under different views, so that the partial characteristics are fully utilized to help the graph classification task to a certain extent;
the input of the full-connection module unit is connected with the output of each graph convolution module, dimension normalization and feature fusion are carried out on the initial graph and the graph features of the multiple views obtained after node aggregation, feature representation containing multiple view information is obtained, and accurate classification of the initial graph is achieved.
Before the multi-view classification model is applied, training samples are required to be adopted for training the multi-view classification model, and when the size and weight of the neighborhood in a good-image convolution unit, the number of view layers in a pooling unit and the hyper-parameters in a full-connection unit are determined, the classification result can be more accurate.
The computer program when executed by a computer processor implements the steps of:
s101, an original protein map showing the protein structure is constructed by using atoms of the protein structure as nodes and chemical bonds between the atoms as connecting edges.
The original protein map G is represented by (a, H), where a represents the adjacency matrix and H represents the node characteristics.
S102, performing pooling operation on the original protein map by using a pooling unit to obtain a plurality of views with different dimensions.
Specifically, pooling the original protein map to obtain multiple views of different dimensions includes:
multiplying the node characteristics H by the weight W to obtain an initial distribution matrix S of the nodes, and normalizing each column in the initial distribution matrix S by using a softmax function to obtain a pooling distribution matrix S', wherein the specific process is as follows:
S＝HW
wherein s isijTo assign an element, s, in the matrixijRepresenting the probability that the node i in the previous layer view is aggregated into the node j in the next layer view after being subjected to pooling;
calculating an adjacent matrix A 'and a node characteristic H' of the next layer view according to the adjacent matrix A and the node characteristic H corresponding to the adjacent matrix A and the node characteristic H before aggregation and the pooling distribution matrix S ', wherein the node characteristic H' represents the self attribute of the node in the next layer view:
A′＝(S′)TAS′
H′＝(S′)TH
and the dimension of the adjacency matrix A 'and the node characteristic H' is smaller than that of the adjacency matrix A and the node characteristic H of the previous layer view.
S103, carrying out graph convolution operation on the multiple views and the original protein graph by using the trained graph convolution unit to obtain multiple feature vectors.
Specifically, the obtaining a plurality of feature vectors by performing a graph convolution operation on the plurality of views and the original protein graph by using a trained graph convolution unit includes:
updating the node characteristics of the input graph by using the following formula, wherein the updated node characteristics are the characteristic vectors:
H″＝ReLU(LHW)
wherein L represents a Laplace matrix L ═ I-D-1/2AD-1/2D is the value matrix of the input adjacent matrix A, H is the input node feature, I is an identity matrix, H' represents the node feature obtained after updating, and W is the value matrix in the graph convolution unitAnd (4) weighting the parameters.
In the embodiment, through k times of graph convolution operations, it can also be understood that node features are obtained through k cascaded graph convolution modules, and each time the graph convolution operation is performed, a neighborhood range captured by the node features is expanded by one node.
And S104, cascading the plurality of feature vectors to obtain a fusion feature vector, performing full-connection operation on the fusion feature vector by using the trained full-connection unit, and outputting a protein activity prediction result.
Specifically, a plurality of feature vectors corresponding to a plurality of views are superimposed by using the following formula to obtain a fusion feature vector H*：
H*＝concat(W″iH″i,i＝0,1,2,3，…，n-1，n)
Where concat (. cndot.) represents a cascade function, Wi"represents the weight parameter corresponding to the ith view, and H" represents the node characteristic corresponding to the ith view; the sequence number 0 indicates the original protein map.
After obtaining the fusion feature vector H*On the basis, carrying out full-connection operation on the fusion characteristic vector according to the following formula to obtain a protein activity prediction result:
C＝soft max(H*W*)
where C represents the matrix of prediction outputs, each element in C represents the probability of whether the function of the protein map is active, W*Representing the weight of the parameter in the fully connected unit.
In protein maps, atoms (nodes) are linked by bonds (edges), and single atoms or chemical bonds have little effect on the activity of the entire protein, but some local structures consisting of groups of atoms and their chemical bonds may represent specific functional units that are critical to the function of the protein, and are then important in characterizing the extracted map. In order to highlight the importance of these functional units, the present embodiment aggregates partial nodes that may have some functions into a new node through a hierarchical pooling process, and then the obtained multi-view with fewer nodes and connecting edges is composed of some functional units, and features extracted on the basis better reflect the functions of proteins. Specifically, the classification of the presence or absence of the activity of the protein may be reflected, including the functions of catalysis, carcinogenicity, and the like.
The multi-view classification model not only considers the topological structure and node attributes of the initial graph, but also considers the structural characteristics of different views obtained by aggregation of partial nodes, and supplements the influence of a deeper graph structure on a prediction result, so that the prediction of protein activity can be better realized.
The above-mentioned embodiments are intended to illustrate the technical solutions and advantages of the present invention, and it should be understood that the above-mentioned embodiments are only the most preferred embodiments of the present invention, and are not intended to limit the present invention, and any modifications, additions, equivalents, etc. made within the scope of the principles of the present invention should be included in the scope of the present invention.
Claims (6)
1. A protein activity prediction apparatus based on a multi-view classification model, comprising a computer memory, a computer processor, and a computer program stored in the computer memory and executable on the computer processor, wherein the computer memory has stored therein a trained multi-view classification model comprising a pooling unit, a volume unit, and a full-connected unit;
the computer processor, when executing the computer program, performs the steps of:
constructing an original protein graph representing a protein structure by taking atoms of the protein structure as nodes and chemical bonds between the atoms as connecting edges;
performing pooling operation on the original protein map by using a pooling unit to obtain a plurality of views with different dimensions;
performing graph convolution operation on the multiple views and the original protein graph by using a trained graph convolution unit to obtain multiple feature vectors;
and cascading the plurality of feature vectors to obtain a fusion feature vector, performing full-connection operation on the fusion feature vector by using the trained full-connection unit, and outputting a protein activity prediction result.
2. The multi-view classification model-based protein activity prediction apparatus of claim 1, wherein the original protein graph G is represented by (a, H), where a represents the adjacency matrix and H represents the node characteristics.
3. The multi-view classification model-based protein activity prediction apparatus of claim 2, wherein the pooling of the original protein map to obtain multiple views of different dimensions comprises:
multiplying the node characteristics H by the weight W to obtain an initial distribution matrix S of the nodes, and normalizing each column in the initial distribution matrix S by using a softmax function to obtain a pooling distribution matrix S', wherein the specific process is as follows:
S＝HW
wherein s isijTo assign an element, s, in the matrixijRepresenting the probability that the node i in the previous layer view is aggregated into the node j in the next layer view after being subjected to pooling;
calculating an adjacent matrix A 'and a node characteristic H' of the next layer view according to the adjacent matrix A and the node characteristic H corresponding to the adjacent matrix A and the node characteristic H before aggregation and the pooling distribution matrix S ', wherein the node characteristic H' represents the self attribute of the node in the next layer view:
A′＝(S′)TAS′
H′＝(S′)TH
and the dimension of the adjacency matrix A 'and the node characteristic H' is smaller than that of the adjacency matrix A and the node characteristic H of the previous layer view.
4. The multi-view classification model-based protein activity prediction apparatus of claim 1, wherein the performing a graph convolution operation on the plurality of views and the original protein graph respectively by using the trained graph convolution unit to obtain a plurality of feature vectors comprises:
updating the node characteristics of the input graph by using the following formula, wherein the updated node characteristics are the characteristic vectors:
H″＝ReLU(LHW)
wherein L represents a Laplace matrix L ═ I-D-1/2AD-1/2D is a value matrix of the input adjacent matrix A, H is the input node characteristic, I is an identity matrix, H' represents the node characteristic obtained after updating, and W is the parameter weight in the graph convolution unit.
5. The multi-view classification model-based protein activity prediction device of claim 1, wherein the fusion feature vector H is obtained by superposing a plurality of feature vectors corresponding to a plurality of views according to the following formula*：
H*＝concat(Wi″H″i,i＝0,1,2,3，…，n-1，n)
Where concat (. cndot.) represents a cascade function, Wi"represents the weight parameter corresponding to the ith view, and H" represents the node characteristic corresponding to the ith view; the sequence number 0 indicates the original protein map.
6. The multi-view classification model-based protein activity prediction apparatus of claim 1, wherein the fusion feature vectors are subjected to full-join operation according to the following formula to obtain a protein activity prediction result:
C＝soft max(H*W*)
where C represents the matrix of prediction outputs, each element in C represents the probability of whether the function of the protein map is active, W*Representing the weight of the parameter in the fully connected unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911031041.6A CN110993037A (en) | 2019-10-28 | 2019-10-28 | Protein activity prediction device based on multi-view classification model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911031041.6A CN110993037A (en) | 2019-10-28 | 2019-10-28 | Protein activity prediction device based on multi-view classification model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110993037A true CN110993037A (en) | 2020-04-10 |
Family
ID=70082538
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911031041.6A Pending CN110993037A (en) | 2019-10-28 | 2019-10-28 | Protein activity prediction device based on multi-view classification model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110993037A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111695702A (en) * | 2020-06-16 | 2020-09-22 | 腾讯科技（深圳）有限公司 | Training method, device, equipment and storage medium of molecular generation model |
| CN111798934A (en) * | 2020-06-23 | 2020-10-20 | 苏州浦意智能医疗科技有限公司 | Molecular property prediction method based on graph neural network |
| CN112598021A (en) * | 2020-11-27 | 2021-04-02 | 西北工业大学 | Graph structure searching method based on automatic machine learning |
| CN113889179A (en) * | 2021-10-13 | 2022-01-04 | 山东大学 | Compound-protein interaction prediction method based on multi-view deep learning |
| CN114023375A (en) * | 2021-03-12 | 2022-02-08 | 浙江工业大学 | Width learning enzyme protein detection method and system based on global sampling subgraph |
| CN114155410A (en) * | 2021-12-09 | 2022-03-08 | 京东科技信息技术有限公司 | Graph pooling, classification model training and reconstruction model training method and device |
| CN116416478A (en) * | 2023-06-06 | 2023-07-11 | 福建工程学院 | A Bioinformatics Classification Model Based on Graph Structure Data Features |
| WO2025036438A1 (en) * | 2023-08-15 | 2025-02-20 | 上海金斯康生物科技有限公司 | Prediction for influence of mutation on protease activity |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110097095A (en) * | 2019-04-15 | 2019-08-06 | 天津大学 | A kind of zero sample classification method generating confrontation network based on multiple view |
| CN110263280A (en) * | 2019-06-11 | 2019-09-20 | 浙江工业大学 | A kind of dynamic link predetermined depth model and application based on multiple view |
| WO2019191777A1 (en) * | 2018-03-30 | 2019-10-03 | Board Of Trustees Of Michigan State University | Systems and methods for drug design and discovery comprising applications of machine learning with differential geometric modeling |
-
2019
- 2019-10-28 CN CN201911031041.6A patent/CN110993037A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019191777A1 (en) * | 2018-03-30 | 2019-10-03 | Board Of Trustees Of Michigan State University | Systems and methods for drug design and discovery comprising applications of machine learning with differential geometric modeling |
| CN110097095A (en) * | 2019-04-15 | 2019-08-06 | 天津大学 | A kind of zero sample classification method generating confrontation network based on multiple view |
| CN110263280A (en) * | 2019-06-11 | 2019-09-20 | 浙江工业大学 | A kind of dynamic link predetermined depth model and application based on multiple view |
Non-Patent Citations (2)
| Title |
|---|
| REX YING: ""Hierarchical Graph Representation Learning with Differentiable Pooling"", 《ARXIV》 * |
| 张媛: ""基于多视图融合的蛋白质功能模块检测方法"", 《电子学报》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111695702A (en) * | 2020-06-16 | 2020-09-22 | 腾讯科技（深圳）有限公司 | Training method, device, equipment and storage medium of molecular generation model |
| CN111695702B (en) * | 2020-06-16 | 2023-11-03 | 腾讯科技（深圳）有限公司 | Training method, device, equipment and storage medium of molecular generation model |
| CN111798934A (en) * | 2020-06-23 | 2020-10-20 | 苏州浦意智能医疗科技有限公司 | Molecular property prediction method based on graph neural network |
| CN111798934B (en) * | 2020-06-23 | 2023-11-14 | 苏州浦意智能医疗科技有限公司 | Molecular property prediction method based on graph neural network |
| CN112598021A (en) * | 2020-11-27 | 2021-04-02 | 西北工业大学 | Graph structure searching method based on automatic machine learning |
| CN114023375A (en) * | 2021-03-12 | 2022-02-08 | 浙江工业大学 | Width learning enzyme protein detection method and system based on global sampling subgraph |
| CN113889179A (en) * | 2021-10-13 | 2022-01-04 | 山东大学 | Compound-protein interaction prediction method based on multi-view deep learning |
| CN113889179B (en) * | 2021-10-13 | 2024-06-11 | 山东大学 | Compound-protein interaction prediction method based on multi-view deep learning |
| CN114155410A (en) * | 2021-12-09 | 2022-03-08 | 京东科技信息技术有限公司 | Graph pooling, classification model training and reconstruction model training method and device |
| CN116416478A (en) * | 2023-06-06 | 2023-07-11 | 福建工程学院 | A Bioinformatics Classification Model Based on Graph Structure Data Features |
| CN116416478B (en) * | 2023-06-06 | 2023-09-26 | 福建工程学院 | Bioinformatics classification model based on graph structure data characteristics |
| WO2025036438A1 (en) * | 2023-08-15 | 2025-02-20 | 上海金斯康生物科技有限公司 | Prediction for influence of mutation on protease activity |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110993037A (en) | Protein activity prediction device based on multi-view classification model | |
| Li et al. | Skeleton-parted graph scattering networks for 3d human motion prediction | |
| Liang et al. | Survey of graph neural networks and applications | |
| CN113705772A (en) | Model training method, device and equipment and readable storage medium | |
| CN107506793B (en) | Method and system for clothing recognition based on weakly labeled images | |
| CN111667022A (en) | User data processing method and device, computer equipment and storage medium | |
| CN108108854A (en) | City road network link prediction method, system and storage medium | |
| Huang et al. | Unconstrained multimodal multi-label learning | |
| CN106528874A (en) | Spark memory computing big data platform-based CLR multi-label data classification method | |
| CN111783688B (en) | A classification method of remote sensing image scene based on convolutional neural network | |
| CN113761250A (en) | Model training method, merchant classification method and device | |
| CN111783754A (en) | Part Context-Based Human Attribute Image Classification Method, System and Device | |
| CN114332893B (en) | Table structure recognition method, device, computer equipment and storage medium | |
| CN115393719A (en) | Hyperspectral Image Classification Method Combining Spatial Spectral Domain Adaptation and Ensemble Learning | |
| Liong et al. | Automatic traditional Chinese painting classification: A benchmarking analysis | |
| CN112560712A (en) | Behavior identification method, device and medium based on time-enhanced graph convolutional network | |
| Uzun et al. | Deep learning based classification of time series of chaotic systems over graphic images | |
| CN116977265A (en) | Training method and device for defect detection model, computer equipment and storage medium | |
| Zhang et al. | A spatial–spectral adaptive learning model for textile defect images recognition with few labeled data | |
| Abed | The Applications of Deep Learning Algorithms for Enhancing Big Data Processing Accuracy | |
| Xu et al. | Multi-level self-adaptive prototypical networks for few-shot node classification on attributed networks | |
| Huang et al. | Hierarchical graph convolutional network for skeleton-based action recognition | |
| Liang et al. | GeoAI-enhanced community detection on spatial networks with graph deep learning | |
| Li et al. | Intelligent unsupervised learning method of physical education image resources based on genetic algorithm | |
| CN114936327B (en) | Element recognition model acquisition method and device, computer equipment and storage medium |
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: 20200410 |