An information processing method and device for predicting drug targets

By obtaining the microspoke spectrum of compounds and target genes, calculating their correlation degree and combining experimental condition data, predicting the probability of the action of compounds on genes, solving the problem of difficult to explore the correlation between compound microspoke spectrum and gene microspoke spectrum in the prior art. , improves the accuracy of drug target prediction.

CN113539366BActive Publication Date: 2024-11-08SHANGHAI INSTITUTE OF MATERIA MEDICA CHINESE ACADEMY OF SCIENCES
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010309556.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-04-17
Publication Date
2024-11-08
Estimated Expiration
2040-04-17

AI Technical Summary

Technical Problem

The prior art is difficult to explore the correlation between compound smuggling spectrum and gene smuggling spectrum in drug target prediction, resulting in insufficient prediction accuracy.

Method used

By obtaining the perturbation spectrum of the compound and the target gene, calculating its correlation degree, and combining preset experimental condition data, Pearson's correlation coefficient and neuronal network are used for feature extraction and classification, the probability of the compound's action on the target gene is predicted.

Benefits of technology

The accuracy of drug target prediction is improved, and the determination process of correlation degree is simplified by considering the correlation between compound smuggling spectrum and gene smuggling spectrum.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113539366B_ABST
    Figure CN113539366B_ABST
Patent Text Reader

Abstract

The present application discloses an information processing method and device for predicting drug targets, which are used to improve the accuracy of drug target prediction. The method includes obtaining a compound perturbation spectrum corresponding to a compound; obtaining a gene perturbation spectrum corresponding to a target gene acted upon by the compound; determining the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum; and predicting the probability that the compound can act on the target gene based on the degree of correlation and preset experimental condition data. By adopting the scheme provided in the present application, the correlation between the compound perturbation spectrum and the gene perturbation spectrum is taken into account in the process of determining whether a compound can act on a target gene, thereby improving the accuracy of drug target prediction.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and in particular to an information processing method and device for predicting drug targets. Background Art

[0002] Computer prediction models of drug targets help deepen our understanding of the molecular mechanisms of drug action, metabolic pathways, adverse effects and drug resistance. In recent years, the rapid increase of multi-omics data and the rapid development of artificial intelligence technology have laid the foundation for the development of computer technology for drug target reasoning and prediction.

[0003] Currently, the technologies for drug target prediction using gene expression profiles or transcriptome data mainly include: comparative analysis methods, network-based analysis methods and machine learning methods.

[0004] Among them, comparative analysis methods make predictions based on the similarity of characteristic differentially expressed genes, such as CMap developed by Broadinstitute. The network-based method starts from the perspective of systems biology and integrates gene expression profiles with cell networks to predict drug targets. For example, the ProTINA method developed by Noa et al. shows good prediction results by establishing a cell type-specific protein-gene regulatory network and using a dynamic model to infer drug targets from differential gene expression profiles. In addition, different machine learning algorithms have also been used to mine transcriptional profile data for drug target prediction. For example, Pabon et al. used a random forest (RF) model to predict drug targets by analyzing the correlation between transcriptional profiles of drug induction and gene knockdown.

[0005] However, the above methods used in the prior art still have drawbacks. For example, they cannot mine the correlation between compound perturbation spectra and gene perturbation spectra, and there is still a lot of room for improvement in the accuracy of drug target prediction. Therefore, how to propose an information processing method for predicting drug targets to mine the correlation between compound perturbation spectra and gene perturbation spectra and improve the accuracy of drug target prediction is a technical problem that needs to be solved urgently. Summary of the invention

[0006] The purpose of the embodiments of the present application is to provide an information processing method for predicting drug targets, so as to improve the accuracy of drug target prediction.

[0007] In order to solve the above technical problems, the embodiments of the present application adopt the following technical solutions: an information processing method for predicting drug targets, comprising:

[0008] Obtain the compound perturbation spectrum corresponding to the compound;

[0009] Obtaining a gene perturbation spectrum corresponding to the target gene acted upon by the compound;

[0010] determining the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum;

[0011] Based on the correlation degree and preset experimental condition data, the probability that the compound can act on the target gene is predicted.

[0012] The beneficial effect of the present application is that it is possible to determine the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum, and then predict the probability that the compound can have an effect on the target gene based on the degree of correlation and the experimental condition data, so that in the process of determining whether the compound can have an effect on the target gene, the correlation between the compound perturbation spectrum and the gene perturbation spectrum is taken into account, thereby improving the accuracy of drug target prediction.

[0013] In one embodiment, determining the correlation between the compound perturbation spectrum and the gene perturbation spectrum comprises:

[0014] The degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum is calculated based on a first preset algorithm.

[0015] In one embodiment, when the correlation degree is the Pearson correlation coefficient between the compound perturbation spectrum and the gene perturbation spectrum, the probability that the compound can act on the target gene is predicted according to the correlation degree and the preset experimental condition data, including:

[0016] Obtain preset experimental condition data;

[0017] The Pearson correlation coefficient and the experimental condition data are substituted into a second preset algorithm to obtain a score for the probability of interaction between the compound and the target gene.

[0018] In one embodiment, determining the correlation between the compound perturbation spectrum and the gene perturbation spectrum comprises:

[0019] Inputting the compound perturbation spectrum and the gene perturbation spectrum into a feature extraction network to perform feature extraction on the compound perturbation spectrum and the gene perturbation spectrum;

[0020] Obtaining a first vector corresponding to the compound perturbation spectrum and a second vector corresponding to the gene perturbation spectrum output by the feature extraction network;

[0021] Inputting the first vector and the second vector into a calculation module;

[0022] Obtain the Pearson correlation coefficient of the first vector and the second vector output by the calculation module.

[0023] The beneficial effect of this embodiment is that the characteristic vectors corresponding to the compound perturbation spectrum and the gene perturbation spectrum, i.e., the first vector and the second vector, are calculated through a neural network, and then the Pearson correlation coefficient of the first vector and the second vector can be obtained through a calculation module. The Pearson correlation coefficient of the first vector and the second vector, i.e., the Pearson correlation coefficient of the compound perturbation spectrum and the gene perturbation spectrum, is used to characterize the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum. Therefore, this embodiment can obtain the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum through a neural network, thereby simplifying the process of determining the degree of correlation between the two.

[0024] In one embodiment, the prediction of the probability that the compound can act on the target gene according to the correlation degree and the preset experimental condition data includes:

[0025] Obtain preset experimental condition data;

[0026] Inputting the Pearson correlation coefficient and the experimental condition data into a classification module;

[0027] The score of the probability of interaction between the compound and the target gene output by the classification module is obtained.

[0028] In one embodiment, the preset experimental condition data includes at least one of the following data:

[0029] Duration of compound perturbation, compound dose, duration of gene knockdown, and cell type.

[0030] In one embodiment, when there are multiple types of target genes, the method further comprises:

[0031] Obtain scores for the probability of interaction between each type of target gene and the compound;

[0032] Sort the scores corresponding to each type of target gene;

[0033] It is determined that the target gene corresponding to the highest score value interacts with the compound.

[0034] The present application also provides an information processing device for predicting drug targets, comprising:

[0035] A first acquisition module is used to acquire a compound perturbation spectrum corresponding to the compound;

[0036] The second acquisition module is used to obtain the gene perturbation spectrum corresponding to the target gene acted by the compound;

[0037] A determination module, used to determine the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum;

[0038] The prediction module is used to predict the probability that the compound can act on the target gene according to the correlation degree and preset experimental condition data.

[0039] In one embodiment, the determining module includes:

[0040] A first input submodule, used for inputting the compound perturbation spectrum and the gene perturbation spectrum into a feature extraction network to perform feature extraction on the compound perturbation spectrum and the gene perturbation spectrum;

[0041] A first acquisition submodule, used to acquire a first vector corresponding to the compound perturbation spectrum and a second vector corresponding to the gene perturbation spectrum output by the feature extraction network;

[0042] A second input submodule, used for inputting the first vector and the second vector into a calculation module;

[0043] The second acquisition submodule is used to acquire the Pearson correlation coefficient between the first vector and the second vector output by the calculation module.

[0044] In one embodiment, the prediction module includes:

[0045] The third acquisition submodule is used to obtain preset experimental condition data;

[0046] A third input submodule, used for inputting the Pearson correlation coefficient and the experimental condition data into a classification module;

[0047] The fourth acquisition submodule is used to obtain the score of the probability of interaction between the compound and the target gene output by the classification module. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a flow chart of an information processing method for predicting drug targets according to an embodiment of the present application;

[0049] Figure 2 This is a flowchart of an information processing method for predicting drug targets according to another embodiment of the present application;

[0050] Figure 3 This is a flowchart of an information processing method for predicting drug targets according to another embodiment of the present application;

[0051] Figure 4 A block diagram of an information processing device for predicting drug targets according to an embodiment of the present application;

[0052] Figure 5This is a block diagram of an information processing device for predicting drug targets according to another embodiment of the present application, illustrating the main architecture of the determination module of this embodiment;

[0053] Figure 6 This is a block diagram of an information processing device for predicting drug targets according to another embodiment of the present application, illustrating the main architecture of the prediction module of this embodiment. DETAILED DESCRIPTION

[0054] Various aspects and features of the present application are described herein with reference to the accompanying drawings.

[0055] It should be understood that various modifications may be made to the embodiments of the present application. Therefore, the above description should not be considered as limiting, but only as an example of an embodiment. Other modifications within the scope and spirit of the present application will occur to those skilled in the art.

[0056] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the present application and, together with the general description of the present application given above and the detailed description of the embodiments given below, serve to explain the principles of the present application.

[0057] These and other characteristics of the present application will become apparent from the following description of a preferred form of embodiment given as a non-limiting example with reference to the accompanying drawings.

[0058] It should also be understood that, although the present application has been described with reference to some specific examples, those skilled in the art will be able to realize many other equivalent forms of the present application that have the features described in the claims and are therefore within the scope of protection defined thereby.

[0059] The above and other aspects, features and advantages of the present application will become more apparent in view of the following detailed description when taken in conjunction with the accompanying drawings.

[0060] Specific embodiments of the present application are described hereinafter with reference to the accompanying drawings; however, it should be understood that the embodiments applied for are merely examples of the present application, which may be implemented in a variety of ways. Well-known and / or repeated functions and structures are not described in detail to avoid unnecessary or redundant details that obscure the present application. Therefore, the specific structural and functional details applied for herein are not intended to be limiting, but merely serve as a basis and representative basis for the claims to teach those skilled in the art to use the present application in a variety of ways with substantially any suitable detailed structure.

[0061] This specification may use the phrases "in one embodiment," "in another embodiment," "in yet another embodiment," or "in other embodiments," all of which may refer to one or more of the same or different embodiments according to the present application.

[0062] Figure 1 This is a flowchart of an information processing method for predicting drug targets according to an embodiment of the present application. The method includes the following steps S11-S14:

[0063] In step S11, a compound perturbation spectrum corresponding to the compound is obtained;

[0064] In step S12, a gene perturbation spectrum corresponding to the target gene acted by the compound is obtained;

[0065] In step S13, the correlation between the compound perturbation spectrum and the gene perturbation spectrum is determined;

[0066] In step S14, the probability that the compound can act on the target gene is predicted based on the correlation degree and the preset experimental condition data.

[0067] In this embodiment, the compound perturbation spectrum corresponding to the compound is obtained, wherein the compound perturbation spectrum is used to express the difference between the gene expression spectrum of the cell after drug addition and the gene expression spectrum of the cell in a normal state. In this embodiment, the compound refers to the compound in the drug to predict the target.

[0068] Further, the perturbation spectrum of the compound is determined by:

[0069] After the selected small molecule compounds were co-incubated with specific cells, positive and negative control groups were set up, and differentially expressed genes were analyzed using sequencing technology to obtain compound perturbation spectra. In addition, compound perturbation spectra can also be obtained by searching existing databases. 978 iconic characteristic differential genes were extracted from the compound perturbation differential gene expression spectrum and formed into a 978-dimensional feature vector, which represents the compound perturbation spectrum.

[0070] After obtaining the compound perturbation spectrum, obtain the gene perturbation spectrum corresponding to the target gene acted by the compound, wherein the gene perturbation spectrum is used to characterize the difference between the expression spectrum after the cell gene is knocked down and the expression spectrum of the cell in the normal state. Determine the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum, and then predict the probability that the compound can act on the target gene based on the degree of correlation and the preset experimental condition data. It should be noted that the compound, in most cases, physically interacts with the protein in the gene, therefore, the effect of the compound on the target gene includes the effect on the protein encoded by the target gene.

[0071] The beneficial effect of the present application is that it can determine the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum, and then predict the probability that the compound can have an effect on the target gene based on the degree of correlation and the experimental condition data, so that in the process of determining whether the compound can have an effect on the target gene, the correlation between the compound perturbation spectrum and the gene perturbation spectrum is taken into account, thereby improving the accuracy of drug target prediction.

[0072] In one embodiment, the above step S13 may be implemented as follows:

[0073] The degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum is calculated based on a first preset algorithm.

[0074] In this embodiment, the correlation degree of the perturbation spectra of the above compounds can be realized based on an algorithm. Specifically, the algorithm can be input into an application program for realization. The specific algorithm is as follows:

[0075] First, the compound perturbation spectrum and gene perturbation spectrum are obtained. In specific practice, the compound perturbation spectrum characteristics are represented by a 978-dimensional vector, denoted as C, C = (c1, c2, c3...c978), for any i (i = 1-978), ci represents the differential expression value of gene i after compound perturbation, that is, the difference between the gene expression spectrum of cells after drug addition and the gene expression spectrum of cells in normal state.

[0076] Gene perturbation spectrum features (978-dimensional vector), denoted as G, G = (g1, g2, g3...g978), for any i (i = 1-978), gi represents the differential expression value of gene i after gene knockdown, that is, the difference between the expression spectrum of the cell after gene knockdown and the expression spectrum of the cell in the normal state.

[0077] Experimental condition data (4-dimensional vector), E = (t1, d, t2, l), t1 represents the duration of compound perturbation, d represents the compound dose, t2 represents the duration of gene knockdown, and l represents the type of cell line.

[0078] The protein-protein interaction network (PPI network) is represented by a connection matrix, denoted by symbol A.

[0079] For the convenience of explanation, without loss of generality, only the differential expression of two genes is studied, then C = (c1, c2), G = (g1, g2).

[0080] To make the whole process easier to understand, we can set C = (0.1, 0.3), G = (0.1, 0.3), and the connection matrix E = (24, 10, 96, 1).

[0081] From the connection matrix Availability Matrix

[0082] Easy to get

[0083] From the Laplace matrix L = DA, we can get:

[0084] By the regularized Laplace matrix L sys =D -1 / 2 LD -1 / 2 We can get:

[0085]

[0086] Perform a spectral decomposition on this matrix:

[0087] L sys =UλU T

[0088] From this we can get:

[0089] Without loss of generality, we can let the parameter matrix

[0090] Because (f*h) graph =UωU T f

[0091] When f = c,

[0092]

[0093] Define a relu function:

[0094]

[0095] Obviously, l1 relu =relu(l1)=(0.03,0.00)

[0096] For simplicity, a 200-dimensional graph embedding is not generated. Instead, a 2-dimensional compound perturbation graph embedding is generated, which is set as E1.

[0097] Without loss of generality, we can let the parameter matrix

[0098] Compound perturbation graph embedding:

[0099]

[0100] Similarly, we can get: Gene knockdown map embedding E2 = [0.03 0.03 0.03]

[0101] Calculate the Pearson of E1 and E2

[0102] Apparently, Pearson R 2 =r*r=1.

[0103] In one embodiment, when the correlation degree is the Pearson correlation coefficient between the compound perturbation spectrum and the gene perturbation spectrum, the above step S14 can be implemented as the following steps A1-A2:

[0104] In step A1, obtaining preset experimental condition data;

[0105] In step A2, the Pearson correlation coefficient and the experimental condition data are substituted into a second preset algorithm to obtain a score for the probability of interaction between the compound and the target gene.

[0106] In this embodiment, the preset experimental condition data E=(t1, d, t2, l) is obtained. According to the specific experimental conditions, the specific experimental condition data t1=24, d=10, t2=96, l=1 are obtained. 2 Combined with the four-dimensional vector experimental condition data E, we get a five-dimensional vector, denoted as v5.

[0107] Obviously v5 = (24, 10, 96, 1, 1).

[0108] The parameter matrix can be

[0109]

[0110] o exp =e O =(e 132 , e 132 )

[0111] sum=e 132 +e 132

[0112]

[0113] The output is a two-dimensional vector, taking the first dimension as the CPI score.

[0114] That is: CPI score=output[1]=0.5.

[0115] That is, the Pearson correlation coefficient and experimental condition data are substituted into the second preset algorithm, and the score of the probability of interaction between the compound and the target gene is obtained as 0.5.

[0116] In one embodiment, the above step S13 can be implemented as the following steps B1-B4:

[0117] In step B1, the compound perturbation spectrum and the gene perturbation spectrum are input into a feature extraction network to extract features from the compound perturbation spectrum and the gene perturbation spectrum;

[0118] In step B2, a first vector corresponding to the compound perturbation spectrum and a second vector corresponding to the gene perturbation spectrum output by the feature extraction network are obtained;

[0119] In step B3, the first vector and the second vector are input into a calculation module;

[0120] In step B4, the Pearson correlation coefficient of the first vector and the second vector output by the calculation module is obtained.

[0121] In this embodiment, the compound perturbation spectrum (also referred to as the compound perturbation transcription spectrum feature, which is composed of 978-dimensional vectors in the specific practice process) and the gene perturbation spectrum are first obtained (also referred to as the gene knockdown transcription spectrum feature, which is composed of 978-dimensional vectors in the specific practice process), and then the compound perturbation spectrum and the gene perturbation spectrum are passed through the feature extraction network. In this embodiment, the feature extraction network is a spectrum-based graph neural network (GCN). By constructing two parallel GCNs, feature extraction is performed from the compound perturbation spectrum and the gene perturbation spectrum respectively, that is, the key features are extracted to achieve dimensionality reduction. After feature extraction, the feature extraction network outputs the first vector corresponding to the compound perturbation spectrum and the second vector corresponding to the gene perturbation spectrum; the first vector and the second vector are obtained by reducing the dimensionality of the corresponding 978-dimensional feature vectors, therefore, the dimension of the first vector and the second vector is less than 978 dimensions, and the first vector and the second vector are input into the calculation module; the Pearson correlation coefficient of the first vector and the second vector output by the calculation module is obtained.

[0122] The beneficial effect of this embodiment is that the characteristic vectors corresponding to the compound perturbation spectrum and the gene perturbation spectrum, i.e., the first vector and the second vector, are calculated through a neural network, and then the Pearson correlation coefficient of the first vector and the second vector can be obtained through a calculation module. The Pearson correlation coefficient of the first vector and the second vector, i.e., the Pearson correlation coefficient of the compound perturbation spectrum and the gene perturbation spectrum, is used to characterize the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum. Therefore, this embodiment can obtain the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum through a neural network, thereby simplifying the process of determining the degree of correlation between the two.

[0123] In one embodiment, Figure 2 As shown, the above step S14 can be implemented as the following steps S21-S23:

[0124] In step S21, obtaining preset experimental condition data;

[0125] In step S22, the Pearson correlation coefficient and experimental condition data are input into the classification module;

[0126] In step S23, the score of the probability of interaction between the compound and the target gene output by the classification module is obtained.

[0127] Acquire preset experimental condition data. Specifically, the preset experimental condition data may include at least one of the following data:

[0128] Duration of compound perturbation, compound dose, duration of gene knockdown, and cell type.

[0129] The Pearson correlation coefficient and experimental condition data are input into the classification model. In this embodiment, the classification model consists of a fully connected hidden layer (for extracting input features) and an output layer (for classifying and discriminating whether there is a compound-protein target interaction), and the score of the probability of interaction between the compound and the target gene output by the classification module is obtained.

[0130] In one embodiment, the preset experimental condition data includes at least one of the following data:

[0131] Duration of compound perturbation, compound dose, duration of gene knockdown, and cell type.

[0132] In this example, heterogeneous experimental condition information is integrated, so that the influence of cell line background, dose and time dependence on differential gene expression and drug target reasoning prediction can be considered, further improving the accuracy of the prediction.

[0133] In one embodiment, Figure 3 As shown, when there are multiple types of target genes, the method can also be implemented as the following steps S31-S33:

[0134] In step S31, the scores of the probability of interaction between each type of target gene and compound are obtained respectively;

[0135] In step S32, the scores corresponding to each type of target gene are sorted;

[0136] In step S33, it is determined whether the target gene corresponding to the highest score value interacts with the compound.

[0137] In this embodiment, when there are multiple target genes, the scores of the probability of interaction between each target gene and the compound are obtained respectively, that is, each target gene performs the aforementioned steps S11-S14 once, calculates the scores of the probability of interaction between each target gene and the compound, and then sorts the calculated scores to determine whether the target gene corresponding to the highest score value interacts with the compound. That is, the target gene corresponding to the highest score value is the target of the drug corresponding to the compound.

[0138] Figure 4 This is a block diagram of an information processing device for predicting drug targets according to an embodiment of the present application, the device comprising the following modules:

[0139] A first acquisition module 41 is used to acquire a compound perturbation spectrum corresponding to the compound;

[0140] The second acquisition module 42 is used to obtain the gene perturbation spectrum corresponding to the target gene acted by the compound;

[0141] A determination module 43, for determining the degree of correlation between the compound perturbation spectrum and the gene perturbation spectrum;

[0142] The prediction module 44 is used to predict the probability that the compound can act on the target gene according to the correlation degree and the preset experimental condition data.

[0143] In one embodiment, Figure 5 As shown, the determination module 43 includes:

[0144] A first input submodule 51 is used to input the compound perturbation spectrum and the gene perturbation spectrum into a feature extraction network to perform feature extraction on the compound perturbation spectrum and the gene perturbation spectrum;

[0145] A first acquisition submodule 52, used to acquire a first vector corresponding to the compound perturbation spectrum and a second vector corresponding to the gene perturbation spectrum output by the feature extraction network;

[0146] A second input submodule 53, used for inputting the first vector and the second vector into the calculation module;

[0147] The second acquisition submodule 54 is used to acquire the Pearson correlation coefficient of the first vector and the second vector output by the calculation module.

[0148] In one embodiment, Figure 6 As shown, the prediction module 44 includes:

[0149] The third acquisition submodule 61 is used to acquire preset experimental condition data;

[0150] A third input submodule 62, for inputting the Pearson correlation coefficient and experimental condition data into the classification module;

[0151] The fourth acquisition submodule 63 is used to obtain the score of the probability of interaction between the compound and the target gene output by the classification module.

[0152] The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and protection scope of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present application.

Claims

1. An information processing method for predicting drug targets, characterized in that: include: Obtaining a compound perturbation spectrum corresponding to the compound, including: co-incubating the selected small molecule compound with specific cells, setting positive and negative control groups, analyzing differentially expressed genes using sequencing technology, and obtaining the compound perturbation spectrum; or obtaining the compound perturbation spectrum by searching an existing database; Obtaining a gene perturbation spectrum corresponding to the target gene acted by the compound, wherein the gene perturbation spectrum is used to characterize the difference between the expression spectrum after the cell gene is knocked down and the expression spectrum of the cell in a normal state; The perturbation spectrum characteristics of the compound are represented by a 978-dimensional vector, denoted as , , for any i (i=1-978), represents the differential expression value of gene i after compound perturbation, that is, the difference between the gene expression profile after drug addition and the gene expression profile of the cell in the normal state; the gene perturbation profile feature is represented by a 978-dimensional vector, recorded as , , for any i (i=1-978), Represents the differential expression value of gene i after gene knockdown, that is, the difference between the expression profile of the cell after gene knockdown and the expression profile of the cell in the normal state; Based on an algorithm program or a feature extraction network, the Pearson correlation coefficient of the compound perturbation spectrum and the gene perturbation spectrum is determined as the degree of correlation, including: passing the compound perturbation spectrum and the gene perturbation spectrum through a feature extraction network, the feature extraction network is a spectrum-based graph neural network, and two parallel spectrum-based graph neural networks are constructed to extract features from the compound perturbation spectrum and the gene perturbation spectrum respectively, that is, to extract key features and achieve dimensionality reduction. After feature extraction, the feature extraction network outputs a first vector corresponding to the compound perturbation spectrum and a second vector corresponding to the gene perturbation spectrum; the first vector and the second vector are obtained by reducing the dimensionality of the corresponding 978-dimensional feature vectors, and the dimensions of the first vector and the second vector are less than 978 dimensions, and the first vector and the second vector are input into a calculation module; and the Pearson correlation coefficient of the first vector and the second vector output by the calculation module is obtained; According to the correlation degree and the preset experimental condition data, the probability that the compound can act on the target gene is predicted based on an algorithm program or a classification model, including: inputting the Pearson correlation coefficient and the experimental condition data into the classification model, the classification model is composed of a fully connected hidden layer and an output layer, and obtaining the score of the probability of interaction between the compound and the target gene output by the classification model; the hidden layer is used to extract input features, and the output layer is used to classify and discriminate whether there is a compound-protein target interaction; The preset experimental condition data includes at least one of the following data: compound perturbation duration, compound dosage, gene knockdown duration and cell type.

2. The method according to claim 1, characterized in that When there are multiple types of target genes, the method further comprises: Obtain scores for the probability of interaction between each type of target gene and the compound; Sort the scores corresponding to each type of target gene; It is determined that the target gene corresponding to the highest score value interacts with the compound.

3. An information processing device for predicting drug targets, characterized in that: include: A first acquisition module, used for acquiring a compound perturbation spectrum according to the method of claim 1 corresponding to the compound; A second acquisition module is used to obtain the gene perturbation spectrum described in the method of claim 1 corresponding to the target gene acted upon by the compound; A determination module, used to determine the Pearson correlation coefficient of the compound perturbation spectrum and the gene perturbation spectrum as the degree of correlation; The prediction module is used to predict the probability that the compound can act on the target gene according to the correlation degree and the preset experimental condition data.

4. The device according to claim 3, characterized in that The determining module comprises: A first input submodule, used for inputting the compound perturbation spectrum and the gene perturbation spectrum into a feature extraction network to perform feature extraction on the compound perturbation spectrum and the gene perturbation spectrum; A first acquisition submodule, used to acquire a first vector corresponding to the compound perturbation spectrum and a second vector corresponding to the gene perturbation spectrum output by the feature extraction network; A second input submodule, used for inputting the first vector and the second vector into a calculation module; The second acquisition submodule is used to acquire the Pearson correlation coefficient between the first vector and the second vector output by the calculation module.

5. The device according to claim 4, characterized in that The prediction module comprises: The third acquisition submodule is used to obtain preset experimental condition data; A third input submodule, used for inputting the Pearson correlation coefficient and the experimental condition data into the classification model; The fourth acquisition submodule is used to obtain the score of the probability of interaction between the compound and the target gene output by the classification model.