Drug relocation method based on topological information and biochemical information

Abstract

The invention relates to a drug relocation method based on topological information and biochemical information. The method comprises the following steps: acquiring a data set to construct a heterogeneous network; calculating and obtaining a comprehensive topological information matrix and a biochemical closeness matrix of the medicine, a comprehensive topological information matrix and a biochemical closeness matrix of the disease, a medicine characteristic matrix and a disease characteristic matrix; expanding a meta-path in the heterogeneous network, and screening out a more reasonable negative sample; and performing prediction by using an integrated learning classifier, and outputting the probability that the drug-disease pair has a treatment relationship as a prediction result. According to the method, the limitation that heterogeneous network information is not fully utilized in an existing drug relocation technology is overcome, and biochemical information under multiple perspectives is used and optimized, so that more comprehensive feature information representation is obtained; and meanwhile, aiming at the fact that a medicine-disease deterministic negative sample is lacked, various medicine-disease interaction possibilities are fully considered, a reliable negative sample is screened out, and the prediction accuracy is improved.

Description

Technical Field

[0001] The present application relates to the technical field of bioinformatics, and in particular to a drug repositioning method based on topological information and biochemical information. Background Art

[0002] Discovering new drugs is a complex task that requires a lot of knowledge in biology and chemistry, as well as a lot of time and money, and is accompanied by a high chance of failure. At the same time, the value of drugs already on the market is far from being fully utilized, resulting in a huge waste of existing drug resources. Therefore, drug repositioning methods are of great significance to the application and development of drugs.

[0003] In the paper "Drug repositioning based on weighted local information augmented graph neural network" published in 2024, Yajie Meng et al. disclosed a drug repositioning method DRAGNN using weighted local information augmented graph neural network, which integrates relevant data of drugs and diseases for drug repositioning. A heterogeneous network is constructed by integrating drug-drug similarity network, disease-disease similarity network and drug-disease treatment relationship network; heterogeneous information and neighborhood homogeneous information of drugs and diseases are collected and integrated, and MLP is used for final prediction. However, the method is not sufficient in integrating drug and disease information and utilizing heterogeneous network information, so it has certain limitations. At the same time, supervised learning-based methods require both positive and negative samples when training models. However, drug-disease association data is similar to other biological data, lacking experimentally verified negative samples. At present, most methods randomly select some from unknown drug-disease pairs as negative samples, but such negative samples are unreliable, and there may be temporarily undiscovered positive samples. Summary of the invention

[0004] The purpose of the present invention is to provide a drug repositioning method based on topological information and biochemical information, which overcomes the limitation of existing drug repositioning technology that the heterogeneous network information is not fully utilized, can more fully integrate and utilize heterogeneous network information, use biochemical information from multiple perspectives and optimize it, so as to obtain a more comprehensive feature information representation; at the same time, in view of the fact that there is a lack of deterministic negative samples of drugs-diseases, the various drug-disease interaction possibilities are fully considered, and reliable negative samples can be screened out, thus overcoming the defect of random selection of negative samples in general methods and improving the accuracy of prediction.

[0005] The technical solution adopted by the present invention is: a drug repositioning method based on topological information and biochemical information, comprising the following steps:

[0006] S1: Acquire a data set to construct a heterogeneous network; the data set includes a drug-disease treatment relationship data set, a drug-protein interaction relationship data set, and a disease-protein interaction relationship data set; obtain five similarity matrices of drugs and two similarity matrices of diseases according to the data set, the five similarity matrices of drugs include a chemical structure similarity matrix, an ATC coding similarity matrix, a side effect similarity matrix, a drug-drug interaction similarity matrix, and a target spectrum similarity matrix, and the two similarity matrices of diseases include a disease phenotype similarity matrix and a disease ontology similarity matrix;

[0007] S2: Taking the number of one-step neighbors of a drug or disease as the measure of the topological information of the corresponding node, calculate the number of common one-step heterogeneous neighbors between drugs and other drugs and between diseases and other diseases, construct a drug-drug common one-step neighbor matrix and a disease-disease common one-step neighbor matrix, and use cosine similarity to calculate the similarity between two nodes, construct a common one-step neighbor similarity matrix between drug nodes and a common one-step neighbor similarity matrix between disease nodes, and obtain the comprehensive topological information matrix of drugs and the comprehensive topological information matrix of diseases;

[0008] S3: averaging the five similarity matrices of drugs and the two similarity matrices of diseases from various biochemical perspectives in step S1 to obtain an integrated drug similarity matrix and a disease similarity matrix, and adjusting the integrated drug similarity matrix and the disease similarity matrix according to a pruning algorithm to obtain a biochemical affinity matrix of drugs and a biochemical affinity matrix of diseases;

[0009] S4: Extract and integrate the topological information and biochemical information of drugs and diseases, remove the noise of the data, and obtain low-dimensional feature representations of drugs and diseases to obtain drug feature matrices and disease feature matrices;

[0010] S5: Expand “protein” in the meta-path in the heterogeneous network to “protein→drug→protein” or “protein→disease→protein” to screen out more reasonable negative samples, which are drug-disease pairs without therapeutic relationship;

[0011] S6: After concatenating the drug features and disease features obtained in S4, the feature vector of the drug-disease pair is used as input, and the ensemble learning classifier XGBoost is used for prediction, and the probability that the drug-disease pair has a therapeutic relationship is output as the prediction result.

[0012] Furthermore, the similarity range of the five similarity matrices of the drugs and the two similarity matrices of the diseases is [0, 1]. The higher the similarity, the more similar the two drugs or diseases are.

[0013] Furthermore, the specific steps of step S2 are:

[0014] S201: Write the drug-disease treatment relationship dataset, the drug-protein interaction relationship dataset, and the disease-protein interaction relationship dataset into matrix form, which are respectively denoted as drug-disease treatment relationship matrix A ds , Drug-protein interaction matrix A dp and disease-protein interaction matrix A sp ;

[0015] S202: Based on the drug-disease treatment relationship matrix A ds , Drug-protein interaction matrix A dp and disease-protein interaction matrix A sp Calculate the drug-drug common one-step neighbor matrix N dd and the disease-disease common one-step neighbor matrix N ss , the specific calculation formula is:

[0016] N dd =A dp ×(A dp ) T +A ds ×(A ds ) T ;

[0017] N ss =A sp ×(A sp ) T +(A ds ) T ×A ds ;

[0018] Among them, (A dp ) T Represents the drug-protein interaction matrix A dp The transposed matrix of (A ds ) T Represents the drug-disease treatment relationship matrix A ds The transposed matrix of (A sp ) T Represents the disease-protein interaction matrix A sp The transposed matrix of

[0019] Take the drug-drug common one-step neighbor matrix N dd The data in the ith row is used as the common one-step neighbor information between the ith drug node and other drug nodes, and the disease-disease common one-step neighbor matrix N ss The j-th row of data in is used as the common one-step neighbor information between the j-th disease node and other disease nodes;

[0020] S203: Calculate the common one-step neighbor similarity matrix S between drug nodes using cosine similarity d And the common one-step neighbor similarity matrix S between disease nodes s , the specific calculation formula is:

[0021]

[0022] Where m represents the total number of drug nodes, n represents the total number of disease nodes; (x1, x2,…, x m ) represents the public one-step neighbor information of drug x, (y1,y2,…,y m ) represents the public one-step neighbor information of drug y, x i represents the common one-step neighbor information between drug x and the i-th drug, y i represents the common one-step neighbor information between drug y and the i-th drug; (a1, a2, …, a n ) represents the public one-step neighbor information of disease a, (b1,b2,…,b n ) represents the public one-step neighbor information of disease b, a j represents the common one-step neighbor information between disease a and the jth disease, b j Represents the common one-step neighbor information between disease b and the jth disease;

[0023] S204: The common one-step neighbor similarity matrix S between drug nodes d Drug-disease treatment relationship matrix A ds and drug-protein interaction matrix A dp Splice and obtain the comprehensive topological information matrix T of the drug d ; At the same time, the common one-step neighbor similarity matrix S between disease nodes s Drug-disease treatment relationship matrix A ds The transposed matrix (A ds ) T and disease-protein interaction matrix A sp Splice and obtain the comprehensive topological information matrix T of the disease s , the specific calculation formula is:

[0024] T d =[S d ,A ds ,A dp ];

[0025] T s =[S s ,(A ds ) T ,Asp ].

[0026] Furthermore, the specific steps of step S3 are:

[0027] S301: Calculate the average of drug similarity and disease similarity from multiple biochemical perspectives to obtain the integrated drug similarity matrix Sim d and disease similarity matrix Sim s , the specific calculation formula is:

[0028]

[0029] Among them, ChemSim d is the chemical structure similarity matrix, ATCSim d Encoding similarity matrix for ATC, SESim d is the side effect similarity matrix, DrDrSim d is the drug-drug interaction similarity matrix, TargetSim d is the target spectrum similarity matrix; PhenomeSim s is the disease phenotype similarity matrix, MeshSim s is the disease ontology similarity matrix;

[0030] S302: The integrated drug similarity matrix and disease similarity matrix are adjusted according to the pruning algorithm, that is, the first k similarity values ​​of each row of the drug similarity matrix and the disease similarity matrix in descending order are adjusted, and the remaining values ​​are set to zero, so as to obtain the drug biochemical affinity matrix B d and the biochemical affinity matrix B of the disease s , the specific calculation formula is:

[0031] B d =pruningSim(Sim d ,k,η);

[0032] B s =pruningSim(Sim s ,k,η);

[0033] Where pruningSim(·) represents the pruning operation, k represents the number of adjustments, and η represents the decay factor;

[0034] The steps of the pruning algorithm are:

[0035] The biochemical affinity matrix B d Initialized as a zero matrix with m rows and m columns, the biochemical affinity matrix B of the disease sInitialize it to a zero matrix with n rows and n columns; set the drug similarity matrix Sim d The column subscripts of the first k values ​​of each row of similarity values ​​in descending order are stored in each row of the matrix indexes1, and the disease similarity matrix Sim s The column subscripts of the first k values ​​of the similarity values ​​in each row in descending order are stored in each row of the matrix indexes2; according to the column subscripts recorded in each row and column of the indexes1 matrix, the drug similarity matrix Sim d The element value at the corresponding position in is attenuated, and the attenuated value is filled into the biochemical affinity matrix B of the drug d The corresponding position of the drug is obtained by the biochemical affinity matrix B d ; According to the column subscripts recorded in each row and column of the indexes2 matrix, the disease similarity matrix Sim s The element value at the corresponding position in is attenuated, and the attenuated value is filled into the biochemical affinity matrix B of the disease s The corresponding position of the disease is obtained by the biochemical affinity matrix B s , the specific calculation formula is:

[0036] B d (M,indexes1(M,K))=η K-1 ×Sim d (M,indexes1(M,K));

[0037] B s (N,indexes2(N,K))=η K-1 ×Sim s (N,indexes2(N,K));

[0038] Among them, M=1,2,…,m; N=1,2,…,n; K=1,2,…,k.

[0039] Furthermore, the specific steps of step S4 are: respectively performing the comprehensive topological information matrix T of the drug d , the biochemical affinity matrix of drugs B d , the comprehensive topological information matrix T of the disease s and the biochemical affinity matrix B of the disease d Perform singular value decomposition and integrate the matrix obtained after decomposition to obtain the drug feature matrix F d and the disease feature matrix F s , the specific calculation formula is:

[0040] F d =[U Td ,U Bd ];

[0041] F s =[U Ts ,U Bs ];

[0042] Among them, U Td is the comprehensive topological information matrix T of the drug d After singular value decomposition, the characteristic matrix U Bd is the biochemical affinity matrix B of the drug d After singular value decomposition, the characteristic matrix U Ts is the comprehensive topological information matrix T of the disease s After singular value decomposition, the characteristic matrix U Bs is the biochemical affinity matrix B of the disease s The feature matrix after singular value decomposition.

[0043] Furthermore, the specific steps of step S5 are:

[0044] S501: let the path "drug→disease" be meta-path M1, the path "drug→protein→disease" be meta-path M2, the path "drug→protein→drug→disease" be meta-path M31, the path "drug→disease→drug→disease" be meta-path M32, and the path "drug→disease→protein→disease" be meta-path M33;

[0045] S502: Expand “protein” in meta-path M2 to “protein→drug→protein” to obtain meta-path M2P1; expand “protein” in meta-path M2 to “protein→disease→protein” to obtain meta-path M2P2; expand “protein” in meta-path M31 to “protein→drug→protein” to obtain meta-path M31P1; expand “protein” in meta-path M31 to “protein→disease→protein” to obtain meta-path M31P2; expand “protein” in meta-path M33 to “protein→drug→protein” to obtain meta-path M33P1; expand “protein” in meta-path M33 to “protein→disease→protein” to obtain meta-path M33P2;

[0046] S503: Meta-path M1, meta-path M2, meta-path M31, meta-path M32, meta-path M33, meta-path M2P1, meta-path M2P2, meta-path M31P1, meta-path M31P2, meta-path M33P1 and meta-path M33P2 together constitute a meta-path matrix M of the heterogeneous network. The specific calculation formula is:

[0047] M=M1+M2+M31+M32+M33+M2P1+M2P2+M31P1

[0048] <h2 style=";text-align:left;direction:ltr">+M31P2+M33P1+M33P2;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0049] <h2 style=";text-align:left;direction:ltr"> M1=A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0050] <h2 style=";text-align:left;direction:ltr"> M2=A<h2 style=";text-align:left;direction:ltr"> dp <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> sp <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> ;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0051] <h2 style=";text-align:left;direction:ltr"> M31=A<h2 style=";text-align:left;direction:ltr"> dp <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> dp <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0052] <h2 style=";text-align:left;direction:ltr"> M32=A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0053] <h2 style=";text-align:left;direction:ltr"> M33=A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> sp <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> sp <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> ;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0054] <h2 style=";text-align:left;direction:ltr"> M2P1=A<h2 style=";text-align:left;direction:ltr"> dp <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> pdp <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> sp <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> ;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0055] <h2 style=";text-align:left;direction:ltr"> M2P2=A<h2 style=";text-align:left;direction:ltr"> dp <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> pSp <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> sp <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> ;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0056] <h2 style=";text-align:left;direction:ltr"> M31P1=A<h2 style=";text-align:left;direction:ltr"> dp <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> pdp <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> dp <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0057] <h2 style=";text-align:left;direction:ltr"> M31P2=A<h2 style=";text-align:left;direction:ltr"> dp <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> psp <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> dp <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0058] <h2 style=";text-align:left;direction:ltr"> M33P1=A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> sp <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> pdp <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> sp <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> ;<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr">

[0059] <h2 style=";text-align:left;direction:ltr"> M33P2=A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> sp <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> psp <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> sp) T ;

[0060] A pdp =(A dp ) T ×A dp ;

[0061] A psp =(A sp ) T ×A sp ;

[0062] Among them, A pdp Indicates the expansion mode of "protein→drug→protein", A psp Indicates the “protein→disease→protein” expansion mode;

[0063] S504: Determine whether each element value in the element path matrix M is 0, and record the drug-disease combination mapped by the subscript combination corresponding to the element with the element value 0 as a negative sample.

[0064] Furthermore, the specific steps of step S6 are: treating the drug-disease association prediction as a binary classification problem, wherein the positive sample is a drug-disease pair with a therapeutic relationship, and the negative sample is a drug-disease pair without a therapeutic relationship; after concatenating the drug features and disease features obtained in S4, the feature vector of the drug-disease pair obtained is used as input, an initial decision tree is created through the ensemble learning classifier XGBoost, and then a new regression tree is iteratively constructed to reduce the calculation residual, and the scores corresponding to the leaf nodes corresponding to each feature vector on each tree are added as the prediction result corresponding to the feature vector, that is, the probability that the drug-disease pair has a therapeutic relationship.

[0065] The beneficial effects of the present invention are:

[0066] (1) The present invention comprehensively considers topological information and biochemical information when extracting drug and disease information, makes more reasonable use of heterogeneous networks, extracts heterogeneous network information more fully and effectively, uses biochemical information from multiple perspectives and optimizes it, thereby obtaining a more comprehensive representation of feature information and improving the performance of the relocalization method described in the present invention;

[0067] (2) Compared with the method of randomly selecting negative samples from unknown samples adopted in the prior art, the present invention adopts the method of expanding the "protein" in the meta-path to "protein→drug→protein" or "protein→disease→protein", comprehensively considering the situations of 11 meta-paths with drug as the starting point and disease as the end point, and fully considering various drug-disease interaction possibilities, so as to screen out more reliable negative samples and further improve the accuracy of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0069] Figure 1 The figure is a flow chart of a method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0070] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention can also be implemented in other ways different from those described herein, and therefore, the present invention is not limited to the limitations of the specific embodiments disclosed below.

[0071] like Figure 1 As shown, the embodiment of the present invention proposes a drug repositioning method based on topological information and biochemical information, comprising the following steps:

[0072] S1: Acquire a data set to construct a heterogeneous network; the data set includes a drug-disease treatment relationship data set, a drug-protein interaction relationship data set, and a disease-protein interaction relationship data set; obtain five similarity matrices of drugs and two similarity matrices of diseases according to the data set, the five similarity matrices of drugs include a chemical structure similarity matrix, an ATC coding similarity matrix, a side effect similarity matrix, a drug-drug interaction similarity matrix, and a target spectrum similarity matrix, and the two similarity matrices of diseases include a disease phenotype similarity matrix and a disease ontology similarity matrix.

[0073] In an embodiment of the present invention, the similarity range of the 5 similarity matrices of the drugs and the 2 similarity matrices of the diseases is [0,1]. The higher the similarity, the more similar the two drugs or diseases are. The drug-disease association data set includes an Fdataset data set and a Cdataset data set. Among them, the Fdataset data set can be directly obtained from the paper "PREDICT: a method for inferring novel drug indications with application to personalized medicine" published by Assaf Gottlieb et al. in 2011; the Cdataset data set can be directly obtained from the paper "Drug repositioning based on comprehensive similarity measures and Bi-Random Walk algorithm" published by Huimin Luo et al. in 2016. The Fdataset dataset contains 593 drugs, 313 diseases, 1753 proteins, 1933 verified drug-disease associations, 3184 drug-protein interactions, and 1066 disease-protein interactions, and the Cdataset dataset includes 663 drugs, 409 diseases, 1805 proteins, 2532 verified drug-disease associations, 3251 drug-protein interactions, and 1280 disease-protein interactions.

[0074] The five similarity matrices of the drugs and the two similarity matrices of the diseases can be directly obtained from the paper "Computational drug repositioning based on multi-similarities bilinear matrix factorization" published by Mengyun Yang et al. in 2021, or they can be calculated or obtained separately using the following methods:

[0075] (1) Chemical structure similarity of drugs

[0076] First, the SMILES string information of the drug is downloaded from the DrugBank database, and then an open source tool CDK is used to convert the SMILES string of the drug into molecular fingerprint information. Based on the molecular fingerprint information of the two drugs, the Tanimoto metric method is used, that is, the intersection of the two molecular fingerprints is divided by the union to calculate the similarity between the two molecular fingerprints, and the chemical structure similarity of the drug-drug is obtained.

[0077] (2) ATC code similarity of drugs

[0078] First, the ATC codes of drugs were downloaded from the DrugBank database, and then the ATC code similarity of drugs was calculated based on the semantic similarity algorithm introduced in Philip Resnik's paper "Using Information Content to Evaluate Semantic Similarity in a Taxonomy" published in 1995.

[0079] (3) Similarity of drug side effects

[0080] First, the side effect information of the drugs is downloaded from the SIDER database, a total of 5868 side effects. Then each drug is represented as a 5868-bit binary string. If the drug has a side effect, the value of the corresponding position in the binary string is set to 1, otherwise it is set to 0. Finally, the Jaccard similarity between binary strings is calculated to obtain the drug-drug side effect similarity.

[0081] (4) Drug-drug interaction similarity

[0082] First, the drug-drug interaction data are downloaded from the DrugBank database, and then each drug is represented as a binary interaction expression spectrum of the drug. In the binary interaction expression spectrum of the drug, the corresponding position of the drug that interacts with the drug is recorded as 1, otherwise it is recorded as 0. Finally, the Jaccard similarity between the binary strings is calculated to obtain the drug-drug interaction similarity.

[0083] (5) Drug target spectrum similarity

[0084] First, the drug-target interaction data is downloaded from the DrugBank database, and then each drug is represented as a binary interaction expression spectrum of the target. In the binary interaction expression spectrum of the target, the corresponding position of the target that interacts with the drug is recorded as 1, otherwise it is recorded as 0. Finally, the Jaccard similarity between the binary strings is calculated to obtain the drug-drug target spectrum similarity.

[0085] (6) Phenotypic similarity of diseases

[0086] The phenotypic similarity data of diseases were downloaded from the MimMiner database recorded in the paper "A text-mining analysis of the human phenome" published by Marc A van Driel et al. in 2006. The database provides the phenotypic similarities of 5080 diseases in the online Mendelian genetic database OMIM. The similarities of the diseases required in the embodiments of the present invention were extracted to form a disease phenotypic similarity matrix.

[0087] (7) Ontological similarity of diseases

[0088] First, the MeSH descriptors of the diseases were downloaded from the NIH website. According to the MeSH descriptors, the diseases were represented as a directed acyclic graph. Then, the ontological semantic similarity of the diseases was calculated according to the method introduced in the paper “Inferring the human microRNAfunctional similarity and functional network based on microRNA-associateddiseases” published by Dong Wang et al. in 2010.

[0089] S2: If drugs share certain proteins or certain diseases, then these drugs may be more similar. Similarly, if diseases share certain proteins or certain drugs, then these diseases may be more similar. Therefore, the embodiment of the present invention uses the number of one-step neighbors of a drug or disease as the metric of the topological information of the corresponding node, calculates the number of common one-step heterogeneous neighbors (proteins, diseases) between a drug and other drugs, and the number of common one-step heterogeneous neighbors (proteins, drugs) between a disease and other diseases, constructs a drug-drug common one-step neighbor matrix and a disease-disease common one-step neighbor matrix, and uses cosine similarity to calculate the similarity between two nodes, constructs a common one-step neighbor similarity matrix between drug nodes and a common one-step neighbor similarity matrix between disease nodes, and obtains a comprehensive topological information matrix of drugs and a comprehensive topological information matrix of diseases; the specific steps are:

[0090] S201: Write the drug-disease treatment relationship dataset, the drug-protein interaction relationship dataset, and the disease-protein interaction relationship dataset into matrix form, which are respectively denoted as drug-disease treatment relationship matrix A ds , Drug-protein interaction matrix A dp and disease-protein interaction matrix A sp Among them, the drug-disease treatment relationship matrix A dsIf the data in the i-th row and j-th column is 1, it means that the i-th drug and the j-th disease have a therapeutic relationship, and if the data in the i-th row and j-th column is 0, it means that the i-th drug and the j-th disease have no therapeutic relationship; the drug-protein interaction matrix A dp If the data in the i-th row and j-th column is 1, it means that the i-th drug and the j-th protein have an interaction relationship, and if the data in the i-th row and j-th column is 0, it means that the i-th drug and the j-th protein have no interaction relationship; in the disease-protein interaction matrix, if the data in the i-th row and j-th column is 1, it means that the i-th disease and the j-th protein have an interaction relationship, and if the data in the i-th row and j-th column is 0, it means that the i-th disease and the j-th protein have no interaction relationship.

[0091] S202: Based on the drug-disease treatment relationship matrix A ds , Drug-protein interaction matrix A dp and disease-protein interaction matrix A sp Calculate the drug-drug common one-step neighbor matrix N dd and the disease-disease common one-step neighbor matrix N ss , the specific calculation formula is:

[0092] N dd =A dp ×(A dp ) T +A ds ×(A ds ) T ;

[0093] N ss =A sp ×(A sp ) T +(A ds ) T ×A ds ;

[0094] Among them, (A dp ) T Represents the drug-protein interaction matrix A dp The transposed matrix of (A ds ) T Represents the drug-disease treatment relationship matrix A ds The transposed matrix of (A sp ) T Represents the disease-protein interaction matrix A sp The transposed matrix of .

[0095] Take the drug-drug common one-step neighbor matrix N dd The data in the ith row is used as the common one-step neighbor information between the ith drug node and other drug nodes, and the disease-disease common one-step neighbor matrix N ssThe j-th row of data in is used as the common one-step neighbor information between the j-th disease node and other disease nodes.

[0096] S203: Calculate the common one-step neighbor similarity matrix S between drug nodes using cosine similarity d And the common one-step neighbor similarity matrix S between disease nodes s , the specific calculation formula is:

[0097]

[0098] Where m represents the total number of drug nodes, n represents the total number of disease nodes; (x1, x2,…, x m ) represents the public one-step neighbor information of drug x, (y1,y2,…,y m ) represents the public one-step neighbor information of drug y, x i represents the common one-step neighbor information between drug x and the i-th drug, y i represents the common one-step neighbor information between drug y and the i-th drug; (a1, a2, …, a n ) represents the public one-step neighbor information of disease a, (b1,b2,…,b n ) represents the public one-step neighbor information of disease b, a j represents the common one-step neighbor information between disease a and the jth disease, b j Indicates the common one-step neighbor information between disease b and the jth disease. In the embodiment of the present invention, the Fdataset dataset includes 593 drugs and 313 diseases, and the Cdataset dataset includes 663 drugs and 409 diseases.

[0099] S204: The common one-step neighbor similarity matrix S between drug nodes d Drug-disease treatment relationship matrix A ds and drug-protein interaction matrix A dp Splice and obtain the comprehensive topological information matrix T of the drug d ; At the same time, the common one-step neighbor similarity matrix S between disease nodes s Drug-disease treatment relationship matrix A ds The transposed matrix (A ds ) T and disease-protein interaction matrix A sp Splice and obtain the comprehensive topological information matrix T of the disease s , the specific calculation formula is:

[0100] T d =[S d ,A ds ,A dp];

[0101] T s =[S s ,(A ds ) T ,A sp ].

[0102] S3: averaging the similarity matrices of the five drugs and the similarity matrices of the two diseases under various biochemical perspectives in step S1 to obtain an integrated drug similarity matrix and a disease similarity matrix, and adjusting the integrated drug similarity matrix and the disease similarity matrix according to a pruning algorithm to obtain a biochemical affinity matrix of drugs and a biochemical affinity matrix of diseases; the specific steps are:

[0103] S301: Calculate the average of drug similarity and disease similarity from multiple biochemical perspectives to obtain the integrated drug similarity matrix Sim d and disease similarity matrix Sim s , the specific calculation formula is:

[0104]

[0105] Among them, ChemSim d is the chemical structure similarity matrix, ATCSim d Encoding similarity matrix for ATC, SESim d is the side effect similarity matrix, DrDrSim d is the drug-drug interaction similarity matrix, TargetSim d is the target spectrum similarity matrix; PhenomeSim s is the disease phenotype similarity matrix, MeshSim s is the disease ontology similarity matrix.

[0106] S302: The integrated drug similarity matrix and disease similarity matrix are adjusted according to the pruning algorithm, that is, the first k similarity values ​​of each row of the drug similarity matrix and the disease similarity matrix in descending order are adjusted, and the remaining values ​​are set to zero, so as to obtain the drug biochemical affinity matrix B d and the biochemical affinity matrix B of the disease s , the specific calculation formula is:

[0107] B d =pruningSim(Sim d ,k,η);

[0108] B s =pruningSim(Sim s ,k,η);

[0109] Wherein, pruningSim(·) represents a pruning operation, k represents the number of adjustments, and η represents a decay factor. In the embodiment of the present invention, the value of k is 35, and the value of the decay factor η is 0.65.

[0110] The steps of the pruning algorithm are:

[0111] The biochemical affinity matrix B d Initialized as a zero matrix with m rows and m columns, the biochemical affinity matrix B of the disease s Initialize it to a zero matrix with n rows and n columns; set the drug similarity matrix Sim d The column subscripts of the first k values ​​of each row of similarity values ​​in descending order are stored in each row of the matrix indexes1, and the disease similarity matrix Sim s The column subscripts of the first k values ​​of the similarity values ​​in each row in descending order are stored in each row of the matrix indexes2; according to the column subscripts recorded in each row and column of the indexes1 matrix, the drug similarity matrix Sim d The element value at the corresponding position in is attenuated, and the attenuated value is filled into the biochemical affinity matrix B of the drug d The corresponding position of the drug is obtained by the biochemical affinity matrix B d ; According to the column subscripts recorded in each row and column of the indexes2 matrix, the disease similarity matrix Sim s The element value at the corresponding position in is attenuated, and the attenuated value is filled into the biochemical affinity matrix B of the disease s The corresponding position of the disease is obtained by the biochemical affinity matrix B s , the specific calculation formula is:

[0112] B d (M,indexes1(M,K))=η K-1 ×Sim d (M,indexes1(M,K));

[0113] B s (N,indexes2(N,K))=η K-1 ×Sim s (N,indexes2(N,K));

[0114] Among them, M=1,2,…,m; N=1,2,…,n; K=1,2,…,k.

[0115] S4: Extract and integrate the topological information and biochemical information of drugs and diseases, remove the noise of the data, and obtain the low-dimensional feature representation of drugs and diseases, and obtain the drug feature matrix and disease feature matrix. Since any matrix F can be approximately decomposed into the product of three matrices U, Σ, and V through singular value decomposition, that is:

[0116] F≈UΣV T ;

[0117] Among them, U and V are orthogonal matrices, and Σ is a non-negative diagonal matrix. Therefore, the drug comprehensive topological information matrix T d , drug biochemical affinity matrix B d , disease comprehensive topological information matrix T s , Disease biochemical affinity matrix B d Perform singular value decomposition and obtain the drug feature matrix F d and the disease feature matrix F s , the specific calculation formula is:

[0118] F d =[U Td ,U Bd ];

[0119] F s =[U Ts ,U Bs ];

[0120] Among them, U Td is the comprehensive topological information matrix T of the drug d After singular value decomposition, the characteristic matrix U Bd is the biochemical affinity matrix B of the drug d After singular value decomposition, the characteristic matrix U Ts is the comprehensive topological information matrix T of the disease s After singular value decomposition, the characteristic matrix U Bs is the biochemical affinity matrix B of the disease s The feature matrix after singular value decomposition.

[0121] S5: In the heterogeneous network G = (V, E), there are "targeting / targeted" links between drugs and proteins, "triggering / triggered" links between proteins and diseases, and "treating / treated" links between drugs and diseases. Among them, V is the set of all drug, protein, and disease nodes in the heterogeneous network, and E is the set of all heterogeneous links in the heterogeneous network. Expand the "protein" in the meta-path in the heterogeneous network to "protein→drug→protein" or "protein→disease→protein" to screen out more reasonable negative samples, which are drug-disease pairs without therapeutic relationships; the specific method is:

[0122] S501: Let the path “drug→disease” be meta-path M1, the path “drug→protein→disease” be meta-path M2, the path “drug→protein→drug→disease” be meta-path M31, the path “drug→disease→drug→disease” be meta-path M32, and the path “drug→disease→protein→disease” be meta-path M33.

[0123] S502: Expand the "protein" in the meta-path M2 to "protein→drug→protein" to obtain the meta-path M2P1; expand the "protein" in the meta-path M2 to "protein→disease→protein" to obtain the meta-path M2P2; expand the "protein" in the meta-path M31 to "protein→drug→protein" to obtain the meta-path M31P1; expand the "protein" in the meta-path M31 to "protein→disease→protein" to obtain the meta-path M31P2; expand the "protein" in the meta-path M33 to "protein→drug→protein" to obtain the meta-path M33P1; expand the "protein" in the meta-path M33 to "protein→disease→protein" to obtain the meta-path M33P2.

[0124] S503: Meta-path M1, meta-path M2, meta-path M31, meta-path M32, meta-path M33, meta-path M2P1, meta-path M2P2, meta-path M31P1, meta-path M31P2, meta-path M33P1 and meta-path M33P2 together constitute a meta-path matrix M of the heterogeneous network. The specific calculation formula is:

[0125] M=M1+M2+M31+M32+M33+M2P1+M2P2+M31P1

[0126] +M31P2+M33P1+M33P2;

[0127] M1=A ds ;

[0128] M2=A dp ×(A sp ) T ;

[0129] M31=A dp ×(A dp ) T ×A ds ;

[0130] M32=A ds ×(A ds ) T ×A ds ;

[0131] M33=A ds ×A sp×(A sp ) T ;

[0132] M2P1=A dp ×A pdp ×(A sp ) T ;

[0133] M2P2=A dp ×A pSp ×(A sp ) T ;

[0134] M31P1=A dp ×A pdp ×(A dp ) T ×A ds ;

[0135] M31P2=A dp ×A psp ×(A dp ) T ×A ds ;

[0136] M33P1=A ds ×A sp ×A pdp ×(A sp ) T ;

[0137] M33P2=A ds ×A sp ×A psp ×(A sp ) T ;

[0138] A pdp =(A dp ) T ×A dp ;

[0139] A psp =(A sp ) T ×A sp ;

[0140] Among them, A pdp Indicates the expansion mode of "protein→drug→protein", A psp Indicates the "protein→disease→protein" expansion method.

[0141] S504: Determine whether each element value in the element path matrix M is 0, and record the drug-disease combination mapped by the subscript combination corresponding to the element with the element value 0 as a negative sample.

[0142] Since the meta-path matrix M comprehensively considers the situations of 11 meta-paths with drugs as starting points and diseases as end points, if the value of an element in M ​​is still equal to 0, the possibility that the drugs mapped by its row and column subscripts have a therapeutic relationship with the disease is relatively low, and the drug-disease combination can be regarded as a more reliable negative sample. Compared with the existing similar research methods of randomly selecting an unknown drug-disease combination as a negative sample, the negative samples screened out by the embodiment of the present invention are more reliable.

[0143] S6: Use the ensemble learning classifier XGBoost to predict the drug feature matrix and the disease feature matrix, and output the probability that the drug-disease pair has a therapeutic relationship as the prediction result; the specific steps are: regard the drug-disease association prediction as a binary classification problem, in which the positive sample is a drug-disease pair with a therapeutic relationship, and the negative sample is a drug-disease pair without a therapeutic relationship; after the drug features and disease features obtained in S4 are concatenated, the characteristic vector of the drug-disease pair obtained is used as input, and an initial decision tree is created through the ensemble learning classifier XGBoost, and then a new regression tree is iteratively constructed to reduce the calculation residual, and the scores corresponding to the leaf nodes corresponding to each feature vector on each tree are added as the prediction result corresponding to the feature vector, that is, the probability that the drug-disease pair has a therapeutic relationship. In the embodiment of the present invention, in order to prevent overfitting in the training stage, in the embodiment of the present invention, the number of decision trees is set to 500, and the maximum depth of the tree is set to 6.

[0144] In terms of experiments, the embodiment of the present invention compares the prediction performance of the method described in the embodiment of the present invention (referred to as TBDR in Tables 1 to 3) with other classic recommendation system methods in the data sets Fdataset and Cdataset based on the 5-fold cross-validation method. The negative samples of the embodiment of the present invention are negative samples of the same number as the positive samples randomly selected after screening by the method described in the embodiment of the present invention. In order to ensure fairness, when performing 5-fold cross-validation, the training set and test set used by each method each time are the same. The average value of the area AUC under the ROC curve obtained 5 times is used as the final AUC result score, and the average value of the area AUPR under the PR curve is used as the final AUPR result score. The experimental results shown in Table 1 can be obtained:

[0145] Table 1 AUC and AUPR results of various methods in 5-fold cross validation experiments on different data sets

[0146]

[0147] As can be seen from Table 1, since the method described in the embodiment of the present invention comprehensively considers topological information and biochemical information when extracting drug and disease information, makes more reasonable use of heterogeneous networks, and optimizes biochemical information to obtain a more comprehensive feature information representation, the performance of the method described in the embodiment of the present invention is better than other methods. The AUC result score and AUPR result score of the method described in the embodiment of the present invention in the Fdataset data set are 0.952 and 0.965 respectively, and the AUC result score and AUPR result score in the Cdataset data set are 0.963 and 0.973 respectively, which are better than other methods.

[0148] In addition, the method described in the embodiment of the present invention was also used in the experiment to predict candidate drugs for the treatment of breast cancer and small cell lung cancer. When predicting candidate drugs for a disease, all known drug-disease associated data sets can be used as training sets for the Cdataset data set, and the remaining candidate sets are sorted after the predicted probabilities of the treatment relationships of all drug-disease pairs in the candidate sets are calculated by the method described in the embodiment of the present invention, and the prediction results shown in Tables 2 and 3 can be obtained. Among the top 10 drugs predicted by the method described in the embodiment of the present invention that may treat breast cancer, 8 drugs have been verified by various evidences from authoritative sources and clinical trials; among the top 10 drugs predicted that may treat small cell lung cancer, 7 drugs have been verified by various evidences from authoritative sources and clinical trials.

[0149] Table 2 The top ten drugs predicted to be effective in treating breast cancer

[0150] Serial number DrugBank IDs Predicted drug candidates Literature verification 1 DB00255 Diethylstilbestrol Not verified yet 2 DB00635 prednisone √ 3 DB01204 Mitoxantrone √ 4 DB00860 Prednisolone √ 5 DB00755 Retinoic acid √ 6 DB01196 Estramustine √ 7 DB00305 Mitomycin √ 8 DB00773 Etoposide √ 9 DB00531 Cyclophosphamide √ 10 DB00687 Fludrocortisone Not verified yet

[0151] Table 3 The top ten drugs predicted to be effective in treating small cell lung cancer

[0152]

[0153]

[0154] In summary, the embodiment of the present invention comprehensively considers topological information and biochemical information when extracting drug and disease information, makes more reasonable use of heterogeneous networks, extracts heterogeneous network information more fully and effectively, uses biochemical information from multiple perspectives and optimizes it, thereby obtaining a more comprehensive representation of feature information, and improving the performance of the relocalization method described in the present invention; compared with the method of randomly selecting negative samples from unknown samples adopted in the prior art, the embodiment of the present invention adopts a method of expanding the "protein" in the meta-path to "protein→drug→protein" or "protein→disease→protein", and more comprehensively considers the "targeting / targeted" link relationship between drugs and proteins in heterogeneous networks, and the "triggering / triggered" link relationship between proteins and diseases. Finally, it comprehensively considers the situations of 11 meta-paths with drugs as the starting point and diseases as the end point, and can screen out more reasonable negative samples, thereby further improving the accuracy of the prediction results.

[0155] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A drug repositioning method based on topological information and biochemical information, characterized in that: The steps include: S1: Acquire data sets to construct heterogeneous networks; the data sets include drug-disease treatment relationship data sets, drug-protein interaction relationship data sets, and disease-protein interaction relationship data sets; According to the data set, five similarity matrices of drugs and two similarity matrices of diseases are obtained, wherein the five similarity matrices of drugs include a chemical structure similarity matrix, an ATC code similarity matrix, a side effect similarity matrix, a drug-drug interaction similarity matrix, and a target spectrum similarity matrix, and the two similarity matrices of diseases include a disease phenotype similarity matrix and a disease ontology similarity matrix; S2: Taking the number of one-step neighbors of a drug or disease as the measure of the topological information of the corresponding node, calculate the number of common one-step heterogeneous neighbors between drugs and other drugs and between diseases and other diseases, construct a drug-drug common one-step neighbor matrix and a disease-disease common one-step neighbor matrix, and use cosine similarity to calculate the similarity between two nodes, construct a common one-step neighbor similarity matrix between drug nodes and a common one-step neighbor similarity matrix between disease nodes, and obtain the comprehensive topological information matrix of drugs and the comprehensive topological information matrix of diseases; S3: averaging the five similarity matrices of drugs and the two similarity matrices of diseases from various biochemical perspectives in step S1 to obtain an integrated drug similarity matrix and a disease similarity matrix, and adjusting the integrated drug similarity matrix and the disease similarity matrix according to a pruning algorithm to obtain a biochemical affinity matrix of drugs and a biochemical affinity matrix of diseases; S4: Extract and integrate the topological information and biochemical information of drugs and diseases, remove the noise of the data, and obtain low-dimensional feature representations of drugs and diseases to obtain drug feature matrices and disease feature matrices; S5: Expand "protein" in the meta-path in the heterogeneous network to "protein→drug→protein" or "protein→disease→protein" to screen out more reasonable negative samples, which are drug-disease pairs without therapeutic relationship; S6: After concatenating the drug features and disease features obtained in S4, the feature vector of the drug-disease pair is used as input, and the ensemble learning classifier XGBoost is used for prediction, and the probability that the drug-disease pair has a therapeutic relationship is output as the prediction result.

2. A drug repositioning method based on topological information and biochemical information according to claim 1, characterized in that: The similarity range in the five similarity matrices of the drugs and the two similarity matrices of the diseases is [0, 1]. The higher the similarity, the more similar the two drugs or diseases are.

3. The drug repositioning method based on topological information and biochemical information according to claim 1, characterized in that: The specific steps of step S2 are: S201: Write the drug-disease treatment relationship dataset, the drug-protein interaction relationship dataset, and the disease-protein interaction relationship dataset into matrix form, and record them as drug-disease treatment relationship matrix A ds , Drug-protein interaction matrix A dp and disease-protein interaction matrix A sp ; S202: Based on the drug-disease treatment relationship matrix A ds , Drug-protein interaction matrix A dp and disease-protein interaction matrix A sp Calculate the drug-drug common one-step neighbor matrix N dd and the disease-disease common one-step neighbor matrix N ss , the specific calculation formula is: N dd =A dp ×(A dp ) T +A ds ×(A ds ) T ; N ss =A sp ×(A sp ) T +(A ds ) T ×A ds ; Among them, (A dp ) T Represents the drug-protein interaction matrix A dp The transposed matrix of ds ) T Represents the drug-disease treatment relationship matrix A ds The transposed matrix of sp ) T Represents the disease-protein interaction matrix A sp The transposed matrix of Take the drug-drug common one-step neighbor matrix N dd The data in the ith row is used as the common one-step neighbor information between the ith drug node and other drug nodes, and the disease-disease common one-step neighbor matrix N ss The j-th row of data in is used as the common one-step neighbor information between the j-th disease node and other disease nodes; S203: Calculate the common one-step neighbor similarity matrix S between drug nodes using cosine similarity d And the common one-step neighbor similarity matrix S between disease nodes s , the specific calculation formula is: Where m represents the total number of drug nodes, n represents the total number of disease nodes; (x1, x2,…, x m ) represents the public one-step neighbor information of drug x, (y1,y2,…,y m ) represents the public one-step neighbor information of drug y, x i represents the common one-step neighbor information between drug x and the i-th drug, y i represents the common one-step neighbor information between drug y and the i-th drug; (a1, a2, …, a n ) represents the public one-step neighbor information of disease a, (b1,b2,…,b n ) represents the public one-step neighbor information of disease b, a j represents the common one-step neighbor information between disease a and the jth disease, b j Represents the common one-step neighbor information between disease b and the jth disease; S204: The common one-step neighbor similarity matrix S between drug nodes d Drug-disease treatment relationship matrix A ds and drug-protein interaction matrix A dp Splice and obtain the comprehensive topological information matrix T of the drug d ; At the same time, the common one-step neighbor similarity matrix S between disease nodes s Drug-disease treatment relationship matrix A ds The transposed matrix (A ds ) T and disease-protein interaction matrix A sp Splice and obtain the comprehensive topological information matrix T of the disease s , the specific calculation formula is: T d =[S d ,A ds ,A dp ]; T s =[S s ,(A ds ) T ,A sp ]。 4. A drug repositioning method based on topological information and biochemical information according to claim 3, characterized in that: The specific steps of step S3 are: S301: Calculate the average of drug similarity and disease similarity from multiple biochemical perspectives to obtain the integrated drug similarity matrix Sim d and disease similarity matrix Sim s , the specific calculation formula is: Among them, ChemSim d is the chemical structure similarity matrix, ATCSim d Encoding similarity matrix for ATC, SESim d is the side effect similarity matrix, DrDrSim d is the drug-drug interaction similarity matrix, TargetSim d is the target spectrum similarity matrix; PhenomeSim s is the disease phenotype similarity matrix, MeshSim s is the disease ontology similarity matrix; S302: The integrated drug similarity matrix and disease similarity matrix are adjusted according to the pruning algorithm, that is, the first k similarity values ​​of each row of the drug similarity matrix and the disease similarity matrix in descending order are adjusted, and the remaining values ​​are set to zero, so as to obtain the drug biochemical affinity matrix B d and the biochemical affinity matrix B of the disease s , the specific calculation formula is: B d =pruningSim(Sim d ,k,η); B s =pruningSim(Sim s ,k,η); Where pruningSim(·) represents the pruning operation, k represents the number of adjustments, and η represents the decay factor; The steps of the pruning algorithm are: The biochemical affinity matrix B d Initialized as a zero matrix with m rows and m columns, the biochemical affinity matrix B of the disease s Initialize it to a zero matrix with n rows and n columns; set the drug similarity matrix Sim d The column subscripts of the first k values ​​of each row of similarity values ​​in descending order are stored in each row of the matrix indexes1, and the disease similarity matrix Sim s The column subscripts of the first k values ​​of the similarity values ​​in each row in descending order are stored in each row of the matrix indexes2; according to the column subscripts recorded in each row and column of the indexes1 matrix, the drug similarity matrix Sim d The element value at the corresponding position in is attenuated, and the attenuated value is filled into the biochemical affinity matrix B of the drug d The corresponding position of the drug is obtained by the biochemical affinity matrix B d ; According to the column subscripts recorded in each row and column of the indexes2 matrix, the disease similarity matrix Sim s The element value at the corresponding position in is attenuated, and the attenuated value is filled into the biochemical affinity matrix B of the disease s The corresponding position of the disease is obtained by the biochemical affinity matrix B s , the specific calculation formula is: B d (M,indexes1(M,K))=η K-1 ×Sim d (M,indexes1(M,K)); <h2 style=";text-align:left;direction:ltr">B<h2 style=";text-align:left;direction:ltr"> s <h2 style=";text-align:left;direction:ltr"> (N,indexes2(N,K))=η<h2 style=";text-align:left;direction:ltr"> K-1 <h2 style=";text-align:left;direction:ltr"> ×Sim<h2 style=";text-align:left;direction:ltr"> s <h2 style=";text-align:left;direction:ltr"> (N,indexes2(N,K)); Among them, M=1,2,…,m; N=1,2,…,n; K=1,2,…,k.

5. The drug repositioning method based on topological information and biochemical information according to claim 4, characterized in that: The specific steps of step S4 are: respectively analyzing the comprehensive topological information matrix T of the drug d , the biochemical affinity matrix of drugs B d , the comprehensive topological information matrix T of the disease s and the biochemical affinity matrix B of the disease d Perform singular value decomposition and integrate the matrix obtained after decomposition to obtain the drug feature matrix F d and the disease feature matrix F s , the specific calculation formula is: F d =[U Td ,IN Bd ]; F s =[U Ts ,IN Bs ]; Among them, U Td is the comprehensive topological information matrix T of the drug d After singular value decomposition, the characteristic matrix U Bd is the biochemical affinity matrix B of the drug d After singular value decomposition, the characteristic matrix U Ts is the comprehensive topological information matrix T of the disease s After singular value decomposition, the characteristic matrix U Bs is the biochemical affinity matrix B of the disease s The feature matrix after singular value decomposition.

6. The drug repositioning method based on topological information and biochemical information according to claim 5, characterized in that: The specific steps of step S5 are: S501: let the path "drug→disease" be meta-path M1, the path "drug→protein→disease" be meta-path M2, the path "drug→protein→drug→disease" be meta-path M31, the path "drug→disease→drug→disease" be meta-path M32, and the path "drug→disease→protein→disease" be meta-path M33; S502: expand "protein" in the meta-path M2 to "protein→drug→protein" to obtain meta-path M2P1; expand "protein" in the meta-path M2 to "protein→disease→protein" to obtain meta-path M2P2; expand "protein" in the meta-path M31 to "protein→drug→protein" to obtain meta-path M31P1; expand "protein" in the meta-path M31 to "protein→disease→protein" to obtain meta-path M31P2; expand "protein" in the meta-path M33 to "protein→drug→protein" to obtain meta-path M33P1; expand "protein" in the meta-path M33 to "protein→disease→protein" to obtain meta-path M33P2; S503: Meta-path M1, meta-path M2, meta-path M31, meta-path M32, meta-path M33, meta-path M2P1, meta-path M2P2, meta-path M31P1, meta-path M31P2, meta-path M33P1 and meta-path M33P2 together constitute a meta-path matrix M of the heterogeneous network. The specific calculation formula is: M=M1+M2+M31+M32+M33+M2P1+M2P2+M31P1 +M31P2+M33P1+M33P2; M1=A ds ; M2=A dp ×(A sp ) T ; M31=A dp ×(A dp ) T ×A ds ; <h2 style=";text-align:left;direction:ltr">M32=A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ×(A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> ×A<h2 style=";text-align:left;direction:ltr"> ds <h2 style=";text-align:left;direction:ltr"> ; M33=A ds ×A sp ×(A sp ) T ; M2P1=A dp ×A pdp ×(A sp ) T ; M2P2=A dp ×A pSp ×(A sp ) T ; M31P1=A dp ×A pdp ×(A dp ) T ×A ds ; M31P2=A dp ×A psp ×(A dp ) T ×A ds ; M33P1=A ds ×A sp ×A pdp ×(A sp ) T ; M33P2=A ds ×A sp ×A psp ×(A sp ) T ; A pdp =(A dp ) T ×A dp ; A psp =(A sp ) T ×A sp ; Among them, A pdp Indicates the expansion mode of "protein→drug→protein", A psp Indicates the "protein→disease→protein" expansion mode; S504: Determine whether each element value in the element path matrix M is 0, and record the drug-disease combination mapped by the subscript combination corresponding to the element with the element value 0 as a negative sample.

7. The drug repositioning method based on topological information and biochemical information according to claim 6, characterized in that: The specific steps of step S6 are: treating the drug-disease association prediction as a binary classification problem, wherein the positive sample is a drug-disease pair with a therapeutic relationship, and the negative sample is a drug-disease pair without a therapeutic relationship; after concatenating the drug features and disease features obtained in S4, the feature vector of the drug-disease pair is used as input, an initial decision tree is created through the ensemble learning classifier XGBoost, and then a new regression tree is iteratively constructed to reduce the calculation residual, and the scores corresponding to the leaf nodes corresponding to each feature vector on each tree are added as the prediction result corresponding to the feature vector, that is, the probability that the drug-disease pair has a therapeutic relationship.

Images