Prediction method for antidepressant drug

Abstract

According to the antidepressant drug prediction method provided by the invention, screening of potential drugs for depression is realized based on combination of a knowledge graph, a transcriptome and a machine learning technology. Specifically, node files, relation files and transcriptome data related to diseases are integrated to construct information in a triple form, and the information is imported into a model to be converted into digital tensor information for scoring. The training data set comprises anti-depression drug tags from ClinicTrials and DygBank and other non-anti-depression drugs, and differential gene expression data is obtained by combining an RNA-seq technology. A plurality of scoring models such as TransE, TransR, DistMult, ComplEx and the like are used for training and evaluation, so as to screen potential antidepressant drugs. The method has the advantages of being low in cost, high in efficiency and wide in range, the accuracy and efficiency of anti-depression drug screening can be remarkably improved, a new drug discovery approach is provided for treatment of depression, and the research and development process of anti-depression drugs can be accelerated.

Description

Technical Field

[0001] The present invention relates to the technical field of drug screening related to depression diseases, and in particular to a method for screening potential drugs for depression based on a combination of knowledge graphs, transcriptomes and machine learning. Background Art

[0002] Depression is a serious mental illness that affects about 16% of the world's population. It is expected that by the end of 2030, it may become the leading cause of disability. Depression, also known as depressive disorder, is a mental disorder with high morbidity, high clinical cure rate but low treatment acceptance rate and high recurrence rate. Its main characteristics are significant and persistent low mood, and patients may show self-harm, suicidal behavior, and even psychotic symptoms such as delusions or hallucinations. Clinically, depressive disorders can be divided into mild, moderate and severe according to the number, type and severity of symptoms. In addition to low mood, loss of interest and lack of energy, there are also some early symptoms such as slow reaction, slow thinking and memory loss, which vary from individual to individual. The cause of depression is related to chronic stress, intestinal microbiota, genetic factors and drug abuse. Although the specific pathogenesis of depression is still unclear, it is generally believed that it is closely related to the reduction of monoamine neurotransmitter levels. Most antidepressant drugs are currently developed based on this theory.

[0003] RNA sequencing (RNA-seq) is a high-throughput gene expression analysis technology that has been widely used in molecular biology research since its inception, helping to understand the function and regulatory mechanism of genes. RNA-seq is often used to analyze differential gene expression (DGE), and its standard workflow is roughly divided into three steps: first, a wet experiment is performed to extract RNA and enrich mRNA or remove rRNA, followed by synthesis of cDNA and construction of sequencing libraries; then sequencing is performed on a high-throughput sequencing platform, usually with a sequencing depth of 10-30 million reads per sample; finally, data analysis is performed, including alignment of reads to a reference genome or transcript, calculation of the number of alignments between reads and transcripts, and standardization and statistical difference analysis between samples. The application of RNA-seq helps to reveal the complexity of mRNA splicing and the regulatory mechanisms of non-coding RNA and enhancer RNA on gene expression.

[0004] The rich biomedical knowledge obtained from biological experiments and clinical practice is a valuable resource for biomedicine. As an effective knowledge management tool, the Knowledge Graph (KG) provides support for knowledge integration in the fields of biomedicine and life sciences. KG is a multi-relational graph or network used to integrate, coordinate and store biomedical knowledge collected from multiple knowledge sources. KG consists of nodes representing biomedical entities (such as diseases, drugs, genes, and biological processes, etc.), and connects these entities through a series of edges to represent the relationships between them (such as drug-treatment-disease, disease-association-gene, and drug-interaction-drug relationship, etc.).

[0005] According to statistics, pharmaceutical companies invest heavily in developing new chemical entities for FDA approval, with an average R&D cost of up to $2.6 billion per drug. Discovering new indications using existing drugs is a cost-effective strategy that can help develop disease prevention and treatment plans. Recent studies have shown that network-based methods can effectively screen potential new indications for FDA-approved drugs with good pharmacokinetic / pharmacodynamic properties, safety, and tolerability by leveraging the relationship between drug targets and diseases. However, the development of cost-effective treatments for MDD is challenged by a lack of understanding of the complex networks between drugs, targets, major depressive disorder (MDD), and diseases. Therefore, extracting relevant information from large-scale structured medical information repositories is a daunting task. Summary of the invention

[0006] The goal of the present invention is to provide a screening method for drugs with potential antidepressant effects based on knowledge graph combined with transcriptomics prediction, aiming to find new antidepressant drugs and improve the efficiency of existing drug screening technology.

[0007] In order to achieve the above object, the technical solution of the present invention is as follows:

[0008] An antidepressant drug prediction method, comprising:

[0009] Obtain existing knowledge graphs in the fields of biomedicine and life sciences, information related to the drugs to be predicted, and information related to depression; information related to depression includes differentially expressed genes related to depression;

[0010] The collected information about the drugs to be predicted and depression-related information are integrated into triples according to nodes and relationships and combined with the triples of the existing knowledge graphs in the fields of biomedicine and life sciences to form a data set;

[0011] Input the triplets of known relationships in the data set into the neural network model, and train it by minimizing the error between the predicted score output by the neural network model and the actual relationship of the input triplet to obtain a trained neural network model;

[0012] The drugs to be predicted are taken as nodes with depression to form triplets of unknown relationships and are input into the trained neural network model. Potential antidepressant drugs are screened according to the therapeutic relationship scores between the predicted drugs and depression output by the trained neural network model.

[0013] Furthermore, the information related to the drug to be predicted includes diseases, anatomy, pathways, drugs, symptoms, dietary supplement ingredients, dietary supplement products, therapeutic categories, genes and side effects related to the drug to be predicted.

[0014] Furthermore, the node pairs of the triples include: disease-disease, drug-gene, drug-disease and gene-gene; among them, the differentially expressed genes related to depression constitute the triple of depression-gene as potential therapeutic targets.

[0015] Furthermore, the neural network model is one or more of TransE, TransR, DistMult and ComplEx.

[0016] Furthermore, the minimization of the error between the predicted score output by the neural network model and the actual relationship of the input triples is specifically:

[0017] Assuming that the closer the predicted score result is to 1, the stronger the association strength of the node pair is, the score of the correct triplet is maximized while the score of the incorrect triplet is minimized.

[0018] Furthermore, potential antidepressant drugs are screened based on the predicted drug treatment relationship scores output by the trained neural network model and depression, specifically:

[0019] The predicted drug-depression treatment relationship scores output by the trained neural network model are sorted from large to small, and the drugs ranked before the known antidepressant drugs are selected as potential antidepressant drugs.

[0020] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the antidepressant drug prediction method when executing the computer program.

[0021] A storage medium containing computer executable instructions, which, when executed by a computer processor, implement the antidepressant drug prediction method.

[0022] A computer program product comprises a computer program / instruction, which implements the steps of the antidepressant drug prediction method when executed by a processor.

[0023] The present invention can effectively screen out new drugs with antidepressant potential, while improving screening efficiency and accuracy. This method is not only low-cost, but also has broad application prospects and can be used to screen potential drugs and related targets and mechanisms for other diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A flow chart of a method for predicting antidepressant drugs provided by the present invention;

[0025] Figure 2 In the embodiment of the present invention, the existing knowledge graph iBKH data is reorganized into triple information, input into the model for training, the best triple type is selected and input into the model for training, a trained model is obtained, and then the best disease and drug relationship type is selected, and the clinical drugs are scored by the AUC scores in different models TransE, TransR, DistMult and ComplEx, as well as the Ensembl scoring result diagram after the four models are integrated; wherein, the Ensembl scoring after the four models are integrated specifically adds the ranking of each drug in the four training model results, and then normalizes, scales the data to 0 to 1, re-ranks the drugs, and selects the drugs ranked before the known antidepressant drugs as potential antidepressant drugs.

[0026] Figure 3 In the embodiment of the present invention, the existing knowledge graphs iBKH and PrimeKG are processed and integrated, reorganized into triple information, the best triple type is selected to be input into the model for training, and a trained model is obtained. Then, the best disease and drug relationship type is selected, and the AUC scores of clinical drugs in different models TransE, TransR, DistMult and ComplEx are scored, as well as the Ensembl scoring result diagram after the integration of the four models;

[0027] Figure 4 The differential genes obtained by differential analysis of different transcriptome data based on iBKH and primeKG in the embodiment of the present invention are then processed and integrated to be reorganized into triple information, and the best triple type is selected to be input into the model for training to obtain a trained model, and then the best disease and drug relationship type is selected, and the AUC scores of clinical drugs in different models TransE, TransR, DistMult and ComplEx scoring are obtained, as well as the Ensembl scoring result diagram after the integration of the four models. DETAILED DESCRIPTION

[0028] In order to better illustrate the purpose, technical solutions and advantages of the present invention, the present invention will be further described below in conjunction with specific embodiments.

[0029] The present invention provides an antidepressant drug prediction method, which integrates the information of MDD-related drugs, targets and pathways in the existing KG, and combines the transcriptome data related to depression to construct a new knowledge graph and train a neural network model, and then uses the trained neural network model as a scoring model to screen and obtain potential antidepressant drugs according to the output results of the scoring model; Figure 1 As shown, the following steps are included:

[0030] (1) Constructing a training data set, wherein the training data set includes various triplets constructed into a new knowledge graph, which is obtained through disease node files, disease-related other factor relationship type files, disease-related transcriptome data, and antidepressant clinical drugs and antidepressant labels. Specifically, the training data set includes the following sub-steps:

[0031] (1.1) Obtain existing knowledge graphs in the fields of biomedicine and life sciences, information related to the drugs to be predicted, and information related to depression;

[0032] Among them, the existing knowledge graph in the field of biomedicine and life sciences refers to the existing knowledge graph related to biomedicine and life sciences, which generally includes disease node files and relationship type files of other disease-related factors, which can be downloaded from some databases. The embodiment of the present invention downloads 10 node files and 18 relationship files from iBKH (https: / / github.com / wcm-wanglab / iBKH / ), and downloads 10 node files and 26 relationship files from PrimeKG (https: / / github.com / mims-harvard / PrimeKG / ). Based on the node files and relationship files of iBKH, the two types of data are integrated and processed, and finally 366,770 nodes are obtained, including diseases (Disease), anatomy (Anatomy), pathway (Pathway), drug (Drug), symptoms (Symptom), dietary supplement ingredients (DietarySupplement Ingredient, DSI), dietary supplement products (DietarySupplement Product, DSP), therapeutic categories (Therapeutic The results are shown in Table 1.

[0033] The depression-related information in the present invention mainly includes differentially expressed genes related to depression. In the present invention, six depression-related data sets, GSE45468, GSE45642, GSE66277, GSE87610, GSE102556 and GSE185855, were downloaded from the GEO database. The transcriptome data were differentially analyzed using the R language limma package and the DESeq2 package to obtain differentially expressed genes (DEGs, with a screening threshold of p<0.05; |log2FC|>=1.2). The results of the differential analysis of each transcriptome are shown in Table 2.

[0034] The information related to the drug to be predicted in the present invention mainly refers to whether it is an antidepressant drug. In the embodiment of the present invention, known approved drugs and clinical trial drugs are downloaded from the DrugBank (https: / / go.drugbank.com / drugs / ) and ClinicalTrials (https: / / clinicaltrials.gov / ) databases, and the drugs related to depression are set as antidepressants and marked with the corresponding label 1, and other drugs are set as non-antidepressants and marked with the corresponding label 0. The drug results corresponding to each stage are shown in Table 3.

[0035] Table 1: Node and relationship information of the training dataset

[0036] Entity Type quantity disease 37,997 Anatomy 23,186 path 31,658 drug 37,628 symptom 1,361 Dietary Supplement Ingredients 4,101 Dietary Supplement Products 137,568 Treatment category 605 Gene 88,415 side effect 4,251 total 366,770

[0037] Relationship Type quantity DSI-Symptom 2,093 DSI-Disease 5,134 DSI-Drug 3,057 DSI-Anatomy 4,334 DSP-DSI 689,297 DSI-TC 5,430 Disease-Pathway 1,941 Drug-Pathway 3,231 Drug-Side Effect 163,206 Disease-Gene 27,538,774 Disease-Symptom 3,357 Anatomy-Gene 12,171,021 Gene-Pathway 152,243 Disease-Disease 18,904 Drug-Drug 2,684,682 Drug-Gene 1,275,949 Drug-Disease 2,727,267 Gene-Gene 802,707 Total 50,312,276

[0038] Table 2: Transcriptional differential analysis results

[0039]

[0040]

[0041] Table 3: Drug clinical information

[0042] Clinical stage Depression-related medications Phase I clinical trial 29 Phase II 72 Phase III clinical trial 27 Phase IV clinical trial 40 FDA approved for marketing 65

[0043] Among them, MDD related drugs represents the number of antidepressant drugs in different stages of drugs, which is also the number of antidepressant drug labels.

[0044] (1.2) Integrate the collected information related to the drug to be predicted and the information related to depression into triples according to nodes and relationships and construct a data set with the triples of the existing knowledge graph in the field of biomedicine and life sciences;

[0045] Specifically, this embodiment selects different relationship types based on the data collected by iBKH and reorganizes them into new triple information. There are three main selection methods: the first method selects 11 relationship types, which have direct and indirect relationships with diseases and drugs compared to the remaining eight types: Disease-Pathway, Drug-Pathway, Drug-SideEffect, Disease-Gene, Disease-Symptom, Gene-Pathway, Disease-Disease, Drug-Drug, Drug-Gene, Drug-Disease, and Gene-Gene; the second method selects five relationships: Disease-Disease, Drug-Drug, Drug-Gene, Drug-Disease, and Gene-Gene, which further reduces the influence of other factors and only considers the influence of drugs and diseases; the third method: Disease-Disease, Drug-Gene, Drug-Disease, and Gene-Gene exclude the influence of Drug-Drug in the second method. Based on iBKH, PrimeKG data content was added, and then reorganized into new triple information according to the selection method in the first step. Based on iBKH and PrimeKG data, the differential genes obtained from the differential analysis of 6 data sets were added to the Disease-Gene relationship type, and the differential genes were added as potential treatment targets to the Disease-Gene relationship, and then different relationship types were selected and reorganized into new triple information according to the selection method in the first step.

[0046] (2) Training the scoring model, taking the integrated triple information as input, inputting it into a neural network model for feature learning, converting the nodes and relationships into tensors, training the nodes and relationships by minimizing the error between the predicted score and the actual relationship, and obtaining the trained neural network model as the scoring model. The scoring model of the embodiment of the present invention takes the four models of TransE, TransR, DistMult and ComplEx in the Python package DGL-KE as examples, and is specifically described as follows:

[0047] Scoring model based on TransE: TransE is a machine learning model for knowledge graph embedding (KGE). Its core idea is to embed entities and relations in the knowledge graph into a low-dimensional vector space so that the semantic relationship between entities can be captured by operations between vectors. The goal of the model is to learn a function f that maps the head entity h, the relation r, and the tail entity t to a score that represents the possibility that the triple (h, r, t) is established in the knowledge graph. TransE models the relation r as a vector so that for a correct triple (h, r, t), the embedding vector of the head entity h plus the relation vector r should be as close as possible to the embedding vector of the tail entity t. It is assumed that the closer the scoring result is to 1, the stronger the association between h and t.

[0048] TransR is an improvement on the TransE model, aiming to address some limitations of TransE in dealing with complex relationships, especially many-to-many and many-to-one relationships. It retains some basic concepts in TransE, including entities, relations, and triples. However, TransR introduces a relationship matrix to improve the expressiveness of the model. The core idea is to combine the relationship matrix with the entity vector to capture more complex relationship patterns. In TransR, each relationship r has a corresponding relationship matrix Mr, which is used to adjust the interactions between entity vectors. It is assumed that the closer the scoring result is to 1, the stronger the association between h and t.

[0049] DistMult is a machine learning model for predicting entity relationships in knowledge graphs. It is a model based on tensor decomposition and aims to capture complex relationships between entities. The core idea is to map entities and relationships into a low-dimensional continuous vector space. Specifically, the DistMult model uses triples (head entity, relationship, tail entity) to train the model, where each entity and relationship is represented as a vector. The training goal of the model is to maximize the score of the correct triples and minimize the score of the wrong triples. Using a polynomial function as a distance metric function can effectively capture the semantic association between entities and improve the accuracy of entity relationship prediction. Specifically, for a triple, the DistMult model performs a dot product operation on the vectors of the head entity and the relationship, and then performs a dot product operation with the vector of the tail entity to obtain a score value used to represent the confidence of the triple. It is assumed that the closer the score result is to 1, the stronger the association between h and t.

[0050] ComplEX extends DistMult by introducing complex-valued embeddings. Its core feature is that entity and relationship embeddings h, r, t no longer exist in real space, but in complex space. It uses complex vectors to represent entities and relationships in the knowledge graph, so that it can more flexibly capture complex interactions between entities, especially symmetric and asymmetric relationships. It is assumed that the closer the score result is to 1, the stronger the association between h and t.

[0051] The four scoring model formulas of TransE, TransR, DistMult and ComplEx are:

[0052] TransE=-||h+rt||

[0053]

[0054] DisMult=h T diag(r)t

[0055] ComplEx=h T Re(diag(r)t)

[0056] Where M r It is the mapping matrix of relation r, which is used to project entities from entity space to relation space. diag(*) means converting the relation vector r into a diagonal matrix, and Re(*) means taking the real part.

[0057] (3) Screening potential antidepressant drugs based on the predicted drug-depression therapeutic relationship scores output by the trained neural network model.

[0058] In a specific embodiment, the predicted drug-depression treatment relationship scores output by the trained neural network model are sorted from large to small, and the drugs ranked before the known depression drugs are selected as potential antidepressants. Furthermore, the ranking of each drug in the four training model results can be added, and then normalized, the data is scaled to 0 to 1, and the drugs are re-ranked, and the drugs ranked before the known depression drugs are selected as potential antidepressants.

[0059] In this embodiment, clinical drug information was downloaded from the DrugBank and ClinicaTrials databases, and there were 4110 FDA-approved drugs, of which 65 had antidepressant labels; there were 4113 drugs from clinical phase IV to marketed, of which 105 had antidepressant labels; there were 4120 drugs from clinical phase III to marketed, of which 132 had antidepressant labels; there were 4157 drugs from clinical phase II to marketed, of which 204 had antidepressant labels; there were 4176 drugs from clinical phase I to marketed, of which 233 had antidepressant labels; the triplet information of clinical drugs and depression was input into the trained scoring model, and potential antidepressant drugs were screened according to the drug ranking results of the scoring model. Among them, the triple information in the best scoring model built based on the existing iBKH knowledge graph is the model trained with the corresponding relationship types between the three nodes Disease-Gene, Drug-Gene, Drug-Disease and Gene-Gene. Then, the relationship types between diseases and drugs selected according to the scoring model are treatment, alleviation, effect, association, association relationship and semantic relationship. Finally, the AUC score is around 0.7. Figure 2 As shown;

[0060] The best scoring model constructed by adding PrimeKG knowledge graph information to the existing iBKH knowledge graph is trained based on the triple information of the corresponding relationship types between the three nodes of Disease-Gene, Drug-Gene, Drug-Disease and Gene-Gene. Then, the relationship types between diseases and drugs selected by the scoring model are treatment, alleviation, effect, association, association relationship and semantic relationship. Finally, the AUC score can reach up to 0.75, and the model performance is slightly improved. Figure 3 As shown;

[0061] In the existing knowledge graph, there are several types of relationships between diseases and genes: association, downregulation, upregulation, inferred relationship, improper regulation, disease-related, pathogenic mutations, polymorphic changes, risk effects in pathogenesis, possible therapeutic effects, biomarkers (diagnosis), promotion of progression, drug targets, overexpression in disease, and mutations that affect the course of the disease. Depression (MDD) patients and normal samples were differentially analyzed, and the obtained differential genes are closely related to the pathogenesis of depression (MDD), which is helpful for the discovery of drugs related to depression (MDD). For the potential related genes (differential genes) found by the transcriptome, the present invention adds 1 to the association relationship to represent that the gene is associated with the disease, and other non-differential genes are not added. At the same time, other relationships are not changed. Subsequently, after adding the information of different transcriptome differential genes to the iBKH and integrated knowledge graphs, the model training results showed that the model performance was improved to varying degrees, such as Figure 4As shown. After adding the corresponding transcriptome differentially expressed genes to the integrated knowledge graph, the triple information in the corresponding optimal scoring model and the relationship types between diseases and drugs are not the same. For example, when GSE53987 is added, the corresponding triple information is Disease-Gene, Drug-Gene, Drug-Disease, and Gene-Gene. The relationship types between diseases and drugs selected by the scoring model are Treatment and Palliates. The final AUC score can reach up to 0.76; when GSE185885 is added, the corresponding triple information is Disease-Gene, Drug-Gene, Drug-Disease, Drug-Drug, and Gene-Gene. The relationship types between diseases and drugs selected by the scoring model are Treats and Palliates. The final AUC score can reach up to 0.76; when GSE66277 is added, the corresponding triple information is Disease-Pathway, Drug-Pathway, Drug-Side There are 11 types of triples, including Effect, Disease-Gene, Disease-Symptom, Gene-Pathway, Disease-Disease, Drug-Drug, Drug-Gene, Drug-Disease and Gene-Gene. The types of relationships between diseases and drugs selected according to the scoring model are Treats and Palliates, and the final AUC score can reach up to 0.83; GSE87610 is added, and its corresponding triple information is 5 types, including Disease-Gene, Drug-Gene, Drug-Disease, Drug-Drug and Gene-Gene. The types of relationships between diseases and drugs selected according to the scoring model are Treats and Palliates, and the final AUC score can reach up to 0.78; GSE45642 is added, and its corresponding triple information is Disease-Pathway, Drug-Pathway, Drug-Side There are 11 types of relationships, including Effect, Disease-Gene, Disease-Symptom, Gene-Pathway, Disease-Disease, Drug-Drug, Drug-Gene, Drug-Disease and Gene-Gene. The relationship types between diseases and drugs selected by the scoring model are treatment and mitigation. The highest AUC score can reach 0.85; GSE45468 was added, and its corresponding triple information was Disease-Gene, Drug-Gene, Drug-Disease, Drug-Drug, and Gene-Gene. The relationship types between diseases and drugs selected by the scoring model were treatment and mitigation. The final AUC score was as high as 0.85, and the final integrated model also reached 0.81. The drugs finally selected according to the scoring results are shown in Table 4.

[0062] Table 4: The top 20 drugs predicted based on Ensembl’s AUC score of 0.81

[0063]

[0064]

[0065] Candidate represents potential antidepressants, and MDD represents known marketed antidepressants

[0066] An electronic device provided by an embodiment of the present invention includes one or more processors, and is used to implement an antidepressant drug prediction method in the above embodiment.

[0067] The electronic device of the present invention can be applied to any device with data processing capability, and the device with data processing capability can be a device or apparatus such as a computer.

[0068] The device embodiment can be implemented by software, hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by the processor of any device with data processing capability in which it is located to read the corresponding computer program instructions in the non-volatile memory into the memory for execution. From the hardware level, in addition to the processor, memory, network interface, and non-volatile memory, any device with data processing capability in which the device in the embodiment is located can also include other hardware according to the actual function of the device with data processing capability, which will not be described in detail.

[0069] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0070] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of the present invention. Ordinary technicians in this field can understand and implement it without paying creative work.

[0071] An embodiment of the present invention further provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, an antidepressant drug prediction method in the above embodiment is implemented.

[0072] The computer-readable storage medium may be an internal storage unit of any device with data processing capability described in any of the aforementioned embodiments, such as a hard disk or a memory. The computer-readable storage medium may also be any device with data processing capability, such as a plug-in hard disk, a smart media card (SMC), an SD card, a flash card, etc. equipped on the device. Furthermore, the computer-readable storage medium may also include both an internal storage unit of any device with data processing capability and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capability, and may also be used to temporarily store data that has been output or is to be output.

[0073] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the present invention.

Claims

1. A method for predicting antidepressant drugs, characterized in that: include: Obtain existing knowledge graphs in the fields of biomedicine and life sciences, information related to the drugs to be predicted, and information related to depression; The depression-related information includes differentially expressed genes related to depression; The collected information related to the drug to be predicted and the information related to depression are integrated into triples according to nodes and relationships and combined with the triples of the existing knowledge graphs in the fields of biomedicine and life sciences to form a data set; Input the triplets of known relationships in the data set into the neural network model, and train it by minimizing the error between the predicted score output by the neural network model and the actual relationship of the input triplet to obtain a trained neural network model; The drugs to be predicted are taken as nodes with depression to form triplets of unknown relationships and are input into the trained neural network model. Potential antidepressant drugs are screened according to the therapeutic relationship scores between the predicted drugs and depression output by the trained neural network model.

2. The method according to claim 1, characterized in that The information related to the drug to be predicted includes diseases, anatomy, pathways, drugs, symptoms, dietary supplement ingredients, dietary supplement products, therapeutic categories, genes and side effects related to the drug to be predicted.

3. The method according to claim 1, characterized in that The node pairs of the triples include: disease-disease, drug-gene, drug-disease and gene-gene; among them, the differentially expressed genes related to depression constitute the depression-gene triplet as potential therapeutic targets.

4. The method according to claim 1, characterized in that: The neural network model is one or more of TransE, TransR, DistMult and ComplEx.

5. The method according to claim 1, characterized in that The minimization of the error between the predicted score output by the neural network model and the actual relationship of the input triples is specifically: Assuming that the closer the predicted score result is to 1, the stronger the association strength of the node pair is, the score of the correct triplet is maximized while the score of the incorrect triplet is minimized.

6. The method according to claim 1, characterized in that Potential antidepressant drugs are screened based on the predicted drug-depression treatment relationship scores output by the trained neural network model, specifically: The predicted drug-depression treatment relationship scores output by the trained neural network model are sorted from large to small, and the drugs ranked before the known antidepressant drugs are selected as potential antidepressant drugs.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the antidepressant drug prediction method according to any one of claims 1 to 6 is implemented.

8. A storage medium comprising computer executable instructions, wherein the computer executable instructions, when executed by a computer processor, implement an antidepressant drug prediction method according to any one of claims 1 to 6.

9. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the antidepressant drug prediction method according to any one of claims 1 to 6 are implemented.

Images