CN111524546B - Drug-target interaction prediction method based on heterogeneous information - Google Patents
Drug-target interaction prediction method based on heterogeneous information Download PDFInfo
- Publication number
- CN111524546B CN111524546B CN202010288303.3A CN202010288303A CN111524546B CN 111524546 B CN111524546 B CN 111524546B CN 202010288303 A CN202010288303 A CN 202010288303A CN 111524546 B CN111524546 B CN 111524546B
- Authority
- CN
- China
- Prior art keywords
- matrix
- drug
- target
- loss
- error
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B15/00—ICT specially adapted for analysing two-dimensional or three-dimensional molecular structures, e.g. structural or functional relations or structure alignment
- G16B15/30—Drug targeting using structural data; Docking or binding prediction
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
Landscapes
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Biophysics (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Biotechnology (AREA)
- Chemical & Material Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Public Health (AREA)
- Evolutionary Computation (AREA)
- Epidemiology (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Bioethics (AREA)
- Artificial Intelligence (AREA)
- Medicinal Chemistry (AREA)
- Pharmacology & Pharmacy (AREA)
- Crystallography & Structural Chemistry (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention belongs to the field of computer technology application, and discloses a drug-target interaction prediction method based on heterogeneous information. On the premise that the relation matrixes of the drug-target, the drug-drug and the target-target are known, the relation matrixes are reconstructed by constructing variable characteristic representations of the drug and the target. And taking the difference between each reconstruction matrix and the original matrix as an error matrix, squaring each error matrix element by element, and then adding the error matrixes element by element to obtain a reconstruction error. And (4) reversely deriving the reconstruction error, and synchronously updating variables by a gradient descent method to ensure that the reconstruction error gradually converges to the minimum. And (4) taking the reconstructed drug-target relation matrix when the reconstruction error is minimized, and comparing the reconstructed drug-target relation matrix with the original relation matrix to obtain the newly predicted drug-target interaction relation. The invention can fully integrate the related heterogeneous information of the drug and the target, provides a powerful tool for computer-aided drug discovery and drug relocation, and has good application prospect.
Description
Technical Field
The invention relates to a drug-target interaction prediction method based on heterogeneous information, and belongs to the field of computer technology application.
Background
Most drugs show efficacy through interaction with their targets in vivo. The identification of new targets for existing or obsolete drugs, i.e. drug relocation, is an important component of drug discovery. With the deep understanding of pharmacology, the concept of "multiple targets, multiple drugs" replacing "a single target, a single drug" has been widely accepted. In recent years, computer-assisted drug screening has provided a powerful aid to the field of drug development, benefiting from the constant progress of information technology. Although many drug-target interaction prediction techniques based on matrix decomposition and traditional machine learning have appeared in the past, problems of dependence on tedious manual feature extraction and the like are common, and accuracy needs to be improved. The heterogeneous data is fully integrated, more information and visual angles are provided for the research of the drug-target relationship, and therefore the prediction efficiency and accuracy are improved. Therefore, in order to promote the development of the field of computer-aided drug discovery and drug relocation, the invention inspires artificial intelligence end-to-end feature representation learning and provides a drug target interaction prediction method based on heterogeneous information.
Disclosure of Invention
The invention aims to provide a medicine-target interaction prediction method based on heterogeneous information, which overcomes the defect that the medicine-target interaction prediction of the traditional machine learning excessively depends on complicated manual feature extraction, fully integrates the medicine and target related heterogeneous information, expresses learning through end-to-end features, minimizes the reconstruction loss method, predicts a new medicine-target interaction relation and is used for new medicine discovery and medicine relocation.
In order to achieve the above object, the present invention provides a method for predicting drug-target interaction based on heterogeneous information, comprising the steps of:
the first step, data preparation, is to matrix the known isomeric relationship information of the drug and target:
the known interaction relationship of m drugs and n targets is represented as matrix a ═ (a)ij)m×nThe matrix A is a non-negative matrix, wherein any element aij>When 0, it means that there is a size a between the drug i and the target jijThe interaction relationship of (a)ijWhen the value is 0, the medicine i and the target j have no interaction relation;
the relationship between the known m drugs is represented as matrix B ═ (B)ij)m×mAny element B in the matrix BijRepresents a relationship score between drug i and drug j;
the known relationship between n targets is represented as matrix C ═ (C)ij)n×nAny element C in the matrix CijRepresents a relationship score between target i and target j;
secondly, constructing a characteristic matrix with element values as variables to represent characteristic information of the medicine and the target; constructing a projection matrix with element values as variables for mapping the characteristics of the drug and the target:
constructing a matrix with variable values of elements representing the characteristics of the drug, D ═ (D)ij)m×kAny ith row of the matrix D represents a k-dimensional feature vector of the drug i, and the transpose matrix of D is DTInitializing the elements of the matrix D to take values randomly;
constructing a matrix F ═ (F) with the values of the elements representing the features of the target as variablesij)n×sAny ith row of the matrix F represents an s-dimensional feature vector of the target i, and the transpose matrix of F is FTRandomly initializing the elements of the matrix F to take values;
constructing projection matrix P with element value as variableA＝(pAij)k×sAs a feature mapping space for drugs and targets, and mapping matrix PARandomly initializing and taking values of the elements;
constructing a projection matrix P with element values as variablesB＝(pBij)k×kAs a drug and drug feature mapping space, and mapping the matrix PBRandomly initializing and taking values of the elements;
constructing a projection matrix P with element values as variablesC＝(pCij)s×sMapping space as a feature of the target and the target, and mapping the matrix PCRandomly initializing and taking values of the elements;
thirdly, respectively reconstructing each relation matrix of the medicine and the target by using the variable matrix constructed in the last step:
reconstructing a drug-target relationship matrix: a. the1＝D×PA×FT；
Reconstructing a drug-drug relationship matrix: b is1＝D×PB×DT；
Reconstructing a target-target relationship matrix: c1＝F×PC×FT；
Fourthly, respectively calculating error matrixes of the reconstruction relation matrixes and the original relation matrixes, and subtracting the elements of the corresponding matrixes one by one:
error matrix of the reconstructed relation matrix and the original relation matrix of the drug-target: a. theloss＝A-A1；
Error matrix of the reconstructed relation matrix of the drug-drug and the original relation matrix: b isloss＝B-B1；
An error matrix of the target-target reconstruction relation matrix and the original relation matrix is as follows: closs＝C-C1；
Fifthly, the error matrixes are respectively squared element by element and then added element by element, namely, the matrix A is firstly addedloss、Bloss、ClossRespectively solving the Hadamard products of the two matrixes, and respectively adding the obtained matrixes element by element to obtain a reconstruction error value:
reconstruction error values of matrix a: lossA＝∑∑(Aloss*Aloss)；
Reconstruction error value of matrix B: lossB＝∑∑(Bloss*Bloss)；
Reconstruction error values of matrix C:lossC＝∑∑(Closs*Closs)；
and sixthly, summing the error values of the reconstruction matrixes to obtain a total error:
total error: loss is lossA+lossB+lossC；
And step seven, synchronously updating the element values of the variable matrixes to minimize the total error:
with the goal of minimizing the value of the total error loss, the loss is separately mapped to a matrix D, F, PA、PB、PCThe matrix D, F, P is synchronously updated by adopting a gradient descent method to carry out derivation on each element variable in theA、PB、PCThe value of loss is gradually converged to the minimum, and the matrix A obtained at this time is taken out1And is denoted by Anew；
And eighthly, extracting potential drug-target interaction relations from the drug-target reconstruction relation matrix with the minimized total error:
will matrix AnewComparing with A, taking out AnewAll of A ini,j0 or an element A in the corresponding positionnewi,jThen A isnewi,jThe numerical value of (A) is the size of the newly predicted interaction relationship between the drug i and the target j, and A isnewi,jThe most likely several drug-target interaction relationships can be obtained from the numerical values sorted from large to small.
The invention has the beneficial effects that:
the drug-target interaction prediction method based on the heterogeneous information overcomes the defect that the drug-target interaction prediction of the traditional machine learning excessively depends on complicated manual feature extraction, and can fully integrate the known drug-target, drug-drug and target-target heterogeneous relation information and predict the potential drug-target interaction relation by using a method for minimizing the reconstruction error through end-to-end feature representation learning and variable update iteration, so that valuable reference is provided for drug discovery and drug relocation.
Drawings
FIG. 1 is a schematic flow chart of a drug-target interaction prediction method based on heterogeneous information according to the present invention.
Detailed Description
The invention is further illustrated with reference to the following figures and examples.
As shown in fig. 1, the flow chart of the drug-target interaction prediction method based on heterogeneous information is schematic.
A method for predicting drug-target interactions based on isomerism, comprising the steps of:
the first step, data preparation, is to matrix the known isomeric relationship information of the drug and target:
the known interaction relationship between 1000 drugs and 2000 targets is expressed as matrix a ═ (a)ij)1000×2000The matrix A is a 0-1 matrix, wherein any element aijWhen 1, it means that there is a known interaction relationship between drug i and target j, aijWhen the value is 0, the medicine i and the target j have no interaction relation;
the relationship between 1000 known drugs based on molecular structural similarity is represented as matrix B ═ (B)ij)1000×1000Any element B in the matrix BijRepresenting a similarity relationship score between drug i and drug j based on molecular structure;
the relationship between the 2000 targets known based on protein sequence similarity is represented as matrix C ═ (C)ij)2000×2000Any element C in the matrix CijRepresents a protein sequence similarity relationship score between target i and target j;
secondly, constructing a characteristic matrix with element values as variables to represent characteristic information of the medicine and the target; constructing a projection matrix with element values as variables for mapping the characteristics of the drug and the target:
constructing a matrix with variable values of elements representing the characteristics of the drug, D ═ (D)ij)1000×512Any ith row of the matrix D represents 512-dimensional feature vectors of the medicine i, and the transpose matrix of D is DTAnd the elements of the D matrix are averaged to 0 and standard deviation to 1Carrying out normal distribution random initialization value taking;
constructing a matrix F ═ (F) with the values of the elements representing the features of the target as variablesij)2000×1024Any ith row of the matrix F represents a 1024-dimensional feature vector of the target i, and the transpose matrix of the matrix F is FTAnd initializing the elements of the F matrix randomly according to normal distribution with the average value of 0 and the standard deviation of 1;
constructing a projection matrix P with element values as variablesA＝(pAij)512×1024As a feature mapping space for drugs and targets, and PAElements of the matrix are randomly initialized according to normal distribution with the average value of 0 and the standard deviation of 1;
constructing a projection matrix P with element values as variablesB＝(pBij)512×512As a drug and drug feature mapping space, and PBElements of the matrix are randomly initialized according to normal distribution with the average value of 0 and the standard deviation of 1;
constructing a projection matrix P with element values as variablesC＝(pCij)1024×1024As a feature mapping space for the target and the target, and PCElements of the matrix are randomly initialized according to normal distribution with the average value of 0 and the standard deviation of 1;
thirdly, respectively reconstructing each relation matrix of the medicine and the target by using the variable matrix constructed in the last step:
reconstructing a drug-target relationship matrix: a. the1＝D×PA×FT；
Reconstructing a drug-drug relationship matrix: b is1＝D×PB×DT；
Reconstructing a target-target relationship matrix: c1＝F×PC×FT；
Fourthly, respectively calculating error matrixes of the reconstruction relation matrixes and the original relation matrixes, and subtracting the elements of the corresponding matrixes one by one:
error matrix of the reconstructed relation matrix and the original relation matrix of the drug-target: a. theloss＝A-A1；
Drug-drug reconstitution relationshipsError matrix of matrix and original relation matrix: b isloss＝B-B1；
Error matrix of target-target reconstruction relation matrix and original relation matrix: closs＝C-C1；
Fifthly, the error matrixes are respectively squared element by element and then added element by element, namely, the matrix A is firstly addedloss、Bloss、ClossRespectively solving the Hadamard products of the two matrixes, and respectively adding the obtained matrixes element by element to obtain a reconstruction error value:
reconstruction error values of matrix a: lossA＝∑∑(Aloss*Aloss)；
Reconstruction error value of matrix B: lossB＝∑∑(Bloss*Bloss)；
Reconstruction error values of matrix C: lossC＝∑∑(Closs*Closs)；
And sixthly, summing the error values of the reconstruction matrixes to obtain a total error:
total error: loss is lossA+lossB+lossC；
And step seven, synchronously updating the element values of the variable matrixes to minimize the total error:
with the goal of minimizing the value of the total error loss, the loss is separately mapped to a matrix D, F, PA、PB、PCThe matrix D, F, P is synchronously updated by gradient descent method with learning rate lr being 0.001A、PB、PCIterating 5000 rounds to make the loss value converge to the minimum, and taking out the matrix A obtained at this time1And is denoted by Anew；
And eighthly, extracting potential drug-target interaction relations from the drug-target reconstruction relation matrix with minimized total errors:
will matrix AnewComparing with A, taking out AnewAll of A ini,j0 or an element A in the corresponding positionnewi,jThen A isnewi,jThe values of (a) are the newly predicted drug i and targetThe magnitude of the interaction between the indices j, Anewi,jThe top 100 pairs of drug-target combinations are taken out in descending order of magnitude, from which the most likely 100 drug-target interactions are obtained.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present invention, and do not limit the protection scope of the present invention. With reference to the description of the embodiment, those skilled in the art will understand and make modifications or substitutions related to the technical solution of the present invention without departing from the spirit and scope of the present invention.
Claims (1)
1. A drug-target interaction prediction method based on heterogeneous information is characterized by comprising the following steps:
the first step, data preparation, is to matrix the known isomeric relationship information of the drug and target:
the known interaction relationship of m drugs and n targets is represented as matrix a ═ (a)ij)m×nThe matrix A is a non-negative matrix, wherein any element aij>0 indicates a size a between drug i and target protein jijThe interaction relationship of (a)ijWhen 0, the drug i and the protein j have no interaction relationship;
the known relationship matrix between m drugs is denoted as B ═ (B)ij)m×mAny element B in the matrix BijRepresents a relationship score between drug i and drug j;
the known relationship matrix between n targets is denoted as C ═ (C)ij)n×nAny element C in the matrix CijRepresents a relationship score between target i and target j;
secondly, constructing a feature matrix with element values as variables to represent feature information of the medicine and the target; constructing a projection matrix with element values as variables for mapping the characteristics of the drug and the target:
constructing a matrix with variable values of elements representing the characteristics of the drug, D ═ (D)ij)m×kAny ith generation of matrix DK-dimensional feature vector of table drug i, and transpose matrix of D is DTRandomly initializing the elements of the matrix D to take values;
constructing a matrix F ═ (F) with the values of the elements representing the features of the target as variablesij)n×sAny ith row of the matrix F represents an s-dimensional feature vector of the target i, and the transpose matrix of F is FTRandomly initializing the elements of the matrix F to take values;
constructing a projection matrix P with element values as variablesA＝(pAij)k×sAs a feature mapping space for drugs and targets, and mapping matrix PARandomly initializing and taking values of the elements;
constructing a projection matrix P with element values as variablesB＝(pBij)k×kAs a drug and drug feature mapping space, and mapping the matrix PBRandomly initializing and taking values of the elements;
constructing a projection matrix P with element values as variablesC＝(pCij)s×sMapping space as a feature of the target and the target, and mapping the matrix PCRandomly initializing and taking values of the elements;
thirdly, respectively reconstructing each relation matrix of the medicine and the target by using the variable matrix constructed in the last step:
reconstructing a drug-target relationship matrix: a. the1＝D×PA×FT；
Reconstructing a drug-drug relationship matrix: b is1＝D×PB×DT；
Reconstructing a target-target relationship matrix: c1＝F×PC×FT；
Fourthly, respectively calculating error matrixes of the reconstruction relation matrixes and the original relation matrixes, and subtracting the elements of the corresponding matrixes one by one:
error matrix of the reconstructed relation matrix and the original relation matrix of the drug-target: a. theloss＝A-A1；
Error matrix of the reconstructed relation matrix of the drug-drug and the original relation matrix: b isloss＝B-B1；
Target-target reconstruction relation matrix and sourceError matrix of the relationship matrix: closs＝C-C1；
Fifthly, the error matrixes are respectively squared element by element and then added element by element, namely, the matrix A is firstly addedloss、Bloss、ClossRespectively solving the Hadamard products of the two matrixes, and respectively adding the obtained matrixes element by element to obtain a reconstruction error value:
reconstruction error values of matrix a: lossA＝∑∑(Aloss*Aloss)；
Reconstruction error value of matrix B: lossB＝∑∑(Bloss*Bloss)；
Reconstruction error values of matrix C: lossC＝∑∑(Closs*Closs)；
And sixthly, summing the error values of the reconstruction matrixes to obtain a total error:
total error: loss is lossA+lossB+lossC；
And step seven, synchronously updating the element values of the variable matrixes to minimize the total error:
with the goal of minimizing the value of the total error loss, the loss is separately mapped to a matrix D, F, PA、PB、PCThe matrix D, F, P is synchronously updated by adopting a gradient descent method to carry out derivation on each element variable in theA、PB、PCThe value of loss is gradually converged to the minimum, and the matrix A obtained at this time is taken out1And is denoted by Anew；
And eighthly, extracting potential drug-target interaction relations from the drug-target reconstruction relation matrix with minimized total errors:
will matrix AnewComparing with A, taking out AnewAll of A ini,j0 or an element A in the corresponding positionnewi,jThen A isnewi,jThe numerical value of (A) is the size of the newly predicted interaction relationship between the drug i and the target j, and A isnewi,jThe most likely several drug-target interaction relationships can be obtained from the numerical values sorted from large to small.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010288303.3A CN111524546B (en) | 2020-04-14 | 2020-04-14 | Drug-target interaction prediction method based on heterogeneous information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010288303.3A CN111524546B (en) | 2020-04-14 | 2020-04-14 | Drug-target interaction prediction method based on heterogeneous information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111524546A CN111524546A (en) | 2020-08-11 |
| CN111524546B true CN111524546B (en) | 2022-05-03 |
Family
ID=71903130
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010288303.3A Active CN111524546B (en) | 2020-04-14 | 2020-04-14 | Drug-target interaction prediction method based on heterogeneous information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111524546B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112420126A (en) * | 2020-12-07 | 2021-02-26 | 湖南大学 | Drug target prediction method based on multi-source data fusion and network structure disturbance |
| CN112863693B (en) * | 2021-02-04 | 2021-09-28 | 东北林业大学 | Drug target interaction prediction method based on multi-channel graph convolution network |
| CN113140255B (en) * | 2021-04-19 | 2022-05-10 | 湖南大学 | Method for predicting interaction of lncRNA-miRNA of plant |
| CN113223609B (en) * | 2021-05-17 | 2023-05-02 | 西安电子科技大学 | Drug target interaction prediction method based on heterogeneous information network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2600269A2 (en) * | 2011-12-03 | 2013-06-05 | Medeolinx, LLC | Microarray sampling and network modeling for drug toxicity prediction |
| CN106529205A (en) * | 2016-11-03 | 2017-03-22 | 中南大学 | Drug target relation prediction method based on drug substructure and molecule character description information |
| CN108520166A (en) * | 2018-03-26 | 2018-09-11 | 中山大学 | A kind of drug targets prediction technique based on multiple similitude network wandering |
| CN109887540A (en) * | 2019-01-15 | 2019-06-14 | 中南大学 | A kind of drug targets interaction prediction method based on heterogeneous network insertion |
| CN110415763A (en) * | 2019-08-06 | 2019-11-05 | 腾讯科技（深圳）有限公司 | The interaction prediction method of drug and target, device, equipment and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11037684B2 (en) * | 2014-11-14 | 2021-06-15 | International Business Machines Corporation | Generating drug repositioning hypotheses based on integrating multiple aspects of drug similarity and disease similarity |
-
2020
- 2020-04-14 CN CN202010288303.3A patent/CN111524546B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2600269A2 (en) * | 2011-12-03 | 2013-06-05 | Medeolinx, LLC | Microarray sampling and network modeling for drug toxicity prediction |
| CN106529205A (en) * | 2016-11-03 | 2017-03-22 | 中南大学 | Drug target relation prediction method based on drug substructure and molecule character description information |
| CN108520166A (en) * | 2018-03-26 | 2018-09-11 | 中山大学 | A kind of drug targets prediction technique based on multiple similitude network wandering |
| CN109887540A (en) * | 2019-01-15 | 2019-06-14 | 中南大学 | A kind of drug targets interaction prediction method based on heterogeneous network insertion |
| CN110415763A (en) * | 2019-08-06 | 2019-11-05 | 腾讯科技（深圳）有限公司 | The interaction prediction method of drug and target, device, equipment and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| Target identification among known drugs by deep;Xiangxiang Zeng等;《Chemical Science》;20200113;第1175-1797页 * |
| 基于多标记学习预测药物-靶标相互作用;彭利红 等;《计算机工程与应用》;20160527;第53卷(第15期);第260-265页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111524546A (en) | 2020-08-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111524546B (en) | Drug-target interaction prediction method based on heterogeneous information | |
| Tari et al. | Fuzzy c-means clustering with prior biological knowledge | |
| AU2021212102A1 (en) | Local genetic ethnicity determination system | |
| US11075008B2 (en) | Generating drug repositioning hypotheses based on integrating multiple aspects of drug similarity and disease similarity | |
| Stahura et al. | Virtual screening methods that complement HTS | |
| Luo et al. | A dynamically growing self-organizing tree (DGSOT) for hierarchical clustering gene expression profiles | |
| US6421668B1 (en) | Method and system for partitioning data into subsets of related data | |
| CN112037868B (en) | Training method and device for neural network for determining molecular reverse synthetic route | |
| AU2011226739A1 (en) | Annotation of a biological sequence | |
| CN109637579B (en) | Tensor random walk-based key protein identification method | |
| Chipman et al. | Clustering microarray data | |
| US20140052686A1 (en) | Method, system and software arrangement for reconstructing formal descriptive models of processes from functional/modal data using suitable ontology | |
| Jena et al. | Soft computing methodologies in bioinformatics | |
| CN112418286A (en) | Multi-view clustering method based on constrained non-negative matrix factorization | |
| Mavroudi et al. | Gene expression data analysis with a dynamically extended self-organized map that exploits class information | |
| RoyChoudhury et al. | Consistency of a phylogenetic tree maximum likelihood estimator | |
| Utsunomiya et al. | Unsupervised detection of ancestry tracks with the GHap r package | |
| Dutta et al. | Incomplete multi-view gene clustering with data regeneration using Shape Boltzmann Machine | |
| Sino et al. | Review of deep learning algorithms in computational biochemistry | |
| Roy et al. | A hidden-state Markov model for cell population deconvolution | |
| Sidheekh et al. | Building Expressive and Tractable Probabilistic Generative Models: A Review | |
| Liu et al. | Multi-objective ant colony optimization biclustering of microarray data | |
| Alinsaif | Leveraging Random Forest and Graph-based Centralities to Predict Yeast Essential Genes | |
| Alashwal et al. | A Bayesian kernel for the prediction of protein-protein interactions | |
| CN114898863B (en) | Auxiliary recognition system for intervention occasion of neonatal necrotizing enterocolitis surgery |
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 |