A drug repositioning method based on tissue property network

Abstract

The present invention discloses a drug repositioning method based on a tissue-specific network, comprising the following steps: pre-processing cancer expression dataset samples and drug-perturbed gene expression datasets; constructing a tissue-specific network related to rectal cancer based on the KEGG pathway database; calculating the inhibition score of drug-related genes on disease-related genes through the tissue-specific network; and screening out the most likely candidate drugs for treating rectal cancer based on the inhibition score. The drug repositioning method based on a tissue-specific network of the present invention has a highly efficient prediction capability compared to classical methods. The method provided by the present invention is based on drug repositioning of a tissue-specific network, which solves the problems of low overlap between disease-related genes and drug-related genes and imperfect drug targeting relationships. The present invention predicts and screens drug effects based on comparative inhibition scores, and has wide applicability for recommending drugs for cancer.

Description

Technical Field

[0001] The present invention relates to the technical field of pharmaceutical research and development, and in particular to a drug repositioning method based on a tissue characteristic network. Background Art

[0002] Due to the high cost and time required to discover and develop new drugs, repurposing older drugs to treat both common and rare diseases is becoming increasingly attractive. These older drugs, many of which are FDA-approved, offer low risk, high safety, and the potential to reduce the cost and time required to develop new drugs. Numerous researchers have proposed various data-driven and experimental approaches to identify potential disease-related drugs. Most drug repositioning approaches utilize machine learning, complex network analysis, literature mining, and transcriptomic data (omics data on compound processing and disease) to discover patterns. By combining information from knowledge bases spanning chemistry, physics, mathematics, pharmacy, and bioinformatics, drug-disease association prediction models can be developed to identify specific diseases or drug classes for repositioning applications.

[0003] The drug repositioning method based on machine learning mainly integrates drug side effect information, drug chemical structure information, and disease information, and then uses machine learning algorithms to learn relevant features. The optimal model is obtained through continuous training of the training set, and finally the model is applied to the test set for classification or regression prediction.

[0004] Complex network analysis methods integrate drug-target networks, drug-disease similarity networks, disease-disease similarity networks, and protein-protein interaction networks to construct complex networks (bipartite graph networks, multi-level networks). The nodes of these networks include diseases, drugs, and target genes. With the development of high-throughput sequencing technology, data related to commonly used drugs and diseases has increased significantly, necessitating the construction of increasingly complex drug-disease networks to discover potential new drug effects. Furthermore, complex network-based methods require repeated optimization, making large-scale data computations very time-consuming and ineffective.

[0005] Literature mining primarily leverages retrieval and semantic inference techniques from natural language processing to discover new relationships between drugs and diseases by searching biomedical and pharmaceutical information provided in literature or databases. However, a problem with literature mining methods is the large amount of non-homologous and unstructured data that requires preprocessing before subsequent data mining. Furthermore, the unstructured data requires semantic transformation, which is not only time-consuming and labor-intensive, but also results in the loss of important information.

[0006] Transcriptomics-based drug repositioning utilizes transcriptome data from compound treatments for drug screening. Due to the rapid development of high-throughput sequencing technology, a large amount of transcriptome data has become available. The GEO database stores transcriptome data related to diseases or treated with a large number of small molecule drugs. Furthermore, the CMap database collects over 6,100 gene expression profiles of over 1,300 small molecule drugs at different doses in five human tumor cell lines (MCF7, ssMCF7, HL60, PC3, and SKMEL5).

[0007] In summary, existing machine learning, complex network analysis, and literature mining methods are all based on the principle of "similarity." The underlying rationale is that similar drugs act on similar proteins; similar drugs may produce similar side effects; and similar diseases may be treated by similar drugs. However, these existing technologies suffer from problems such as low overlap between disease-associated and drug-associated genes and incomplete drug-target relationships. Summary of the Invention

[0008] The purpose of the present invention is to provide a drug repositioning method based on tissue characteristic networks and a new transcriptomics-based drug repositioning technology solution to solve the problems existing in the existing technology, such as low overlap between disease-related genes and drug-related genes and imperfect drug targeting relationships.

[0009] In order to achieve the above object, the present invention adopts the following technical solutions:

[0010] A drug repositioning method based on tissue property network, comprising:

[0011] S1, obtain the gene expression profile data of cancer and normal samples and the gene expression profile data of drug perturbation;

[0012] S2, preprocessing the gene expression profile data of cancer and normal samples and drug-perturbed gene expression profile data to screen for differentially expressed genes (DEGs); differentially expressed genes (DEGs) include drug-differentially expressed genes and disease-differentially expressed genes;

[0013] S3, obtains signaling pathways related to a specific type of cancer and establishes tissue-specific networks based on signaling pathway integration;

[0014] S4, integrating drug differentially expressed genes and disease differentially expressed genes into tissue-specific networks, calculating the shortest paths between drug differentially expressed genes and disease differentially expressed genes, and finally calculating the inhibition score value of each drug-disease relationship;

[0015] S5, comparing the inhibition score values ​​to obtain candidate drugs.

[0016] Preferably, the S1 includes:

[0017] Obtain expression profile data of cancer and normal samples from the GEO database;

[0018] Obtain drug-induced gene expression profile data from the CMap database.

[0019] Preferably, the S2 includes:

[0020] The limma package in R language was used to preprocess the expression profile data of cancer and normal samples and the gene expression profile data of drug perturbations to screen for differentially expressed genes.

[0021] Preferably, the screening conditions for the drug differentially expressed genes and the disease differentially expressed genes are both FDR<0.01.

[0022] Preferably, the S3 includes:

[0023] The signaling pathways associated with the specified type of cancer were obtained from the KEGG database.

[0024] Preferably, the S4 includes:

[0025] The suppression score S is calculated using the following formula:

[0026]

[0027] Where m represents the number of differentially expressed genes affected by drug perturbations, n represents the number of differentially expressed genes associated with diseases, log2(fc i ) represents the log value of the fold change of drug-related differentially expressed gene i, log2(fc j ) represents the logarithmic value of the fold change of disease-related differentially expressed gene j. If drug-perturbed gene i is upstream of disease gene j, then gene i has an effect on gene j, otherwise it has no effect. ij Represents the shortest path from gene i to gene j. Coefficient β ij Indicates the type of effect of gene i on gene j. If the effect type is activation or induction, then β ij =1; if it is inhibition, then β ij =-1. sign(log2(fc j)) indicates that if the fold change of disease-related differentially expressed gene j is greater than 0, the gene is upregulated in the disease state and the value is 1; otherwise, the value is -1. Drug effect on disease-related gene j represents the sum of the effects of m drug-perturbed genes on gene j. If the sum is greater than 0, it means that the drug upregulates the disease gene j; otherwise, it downregulates it. In addition, if the disease gene is upregulated (downregulated) and the drug downregulates (upregulates) the gene, the inhibition score is +1; otherwise, the inhibition score is -1; if the drug effect on the gene is 0, the inhibition score is 0.

[0028] Preferably, the S5 includes:

[0029] The most effective drugs are screened as candidate drugs by comparing the inhibition score values; a larger inhibition score value indicates a better potential effect of the drug, otherwise the drug effect is worse.

[0030] The beneficial effects of the present invention are:

[0031] 1. The present invention's tissue-specific network-based drug repositioning method offers highly predictive power compared to conventional methods. This method addresses the issues of low overlap between disease-associated and drug-associated genes and incomplete drug-target relationships. Furthermore, this method approaches drug repositioning from a signaling pathway network perspective and, based on comparative inhibition scores, has broad applicability for recommending cancer drugs.

[0032] 2. The technical solution adopted by the present invention is a method based on transcriptome data (omics data of compound treatment and disease), which is a completely different idea from the existing technology. It defines new effects of drugs through gene expression spectrum data, and screens the most effective drugs as candidate drugs by calculating and comparing inhibition scores, providing a new technical concept for drug research and development. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The present invention is further described with reference to the accompanying drawings. However, the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative effort.

[0034] Figure 1 , which is a flow chart of a drug repositioning method based on a tissue characteristic network according to an embodiment of the present invention.

[0035] Figure 2 , which is a schematic diagram of the prediction effect of the drug repositioning method based on tissue-specific network in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The following describes embodiments of the present invention in detail. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present invention and are not to be construed as limiting the present invention.

[0037] See also Figure 1 The drug repositioning method based on tissue characteristic network provided by the embodiment of the present invention includes the following steps:

[0038] S1, obtain the expression profile data of cancer and normal samples and the gene expression profile data of drug perturbation; obtain the expression profile data of cancer and normal samples from the GEO database; obtain the gene expression profile data of drug perturbation from the CMap database.

[0039] S2, preprocessing the gene expression profile data of cancer and normal samples and drug-perturbed gene expression profile data to screen for differentially expressed genes (DEGs); differentially expressed genes (DEGs) include drug-differentially expressed genes and disease-differentially expressed genes;

[0040] Specifically, the expression profile data of cancer and normal samples and the gene expression profile data of drug perturbations are preprocessed using the limma software package in the R language to screen for differentially expressed genes; the screening conditions for the drug differentially expressed genes and the disease differentially expressed genes are both FDR<0.01.

[0041] S3, obtains signaling pathways related to a specific type of cancer and establishes tissue-specific networks based on signaling pathway integration;

[0042] Specifically, signal pathways related to a specified type of cancer are obtained from the KEGG database.

[0043] Specifically, the nodes in the tissue-specific network represent genes or proteins, and the edges between them represent signal transduction between genes or proteins, such as activation or inhibition.

[0044] S4, integrating drug differentially expressed genes and disease differentially expressed genes into tissue-specific networks, calculating the shortest paths between drug differentially expressed genes and disease differentially expressed genes, and finally calculating the inhibition score value of each drug-disease relationship;

[0045] Specifically, the suppression score S is calculated using the following formula:

[0046]

[0047] Where m represents the number of differentially expressed genes affected by drug perturbations, n represents the number of differentially expressed genes associated with diseases, log2(fc i) represents the log value of the fold change of drug-related differentially expressed gene i, log2(fc j ) represents the logarithmic value of the fold change of disease-related differentially expressed gene j. If drug-perturbed gene i is upstream of disease gene j, then gene i has an effect on gene j, otherwise it has no effect. ij Represents the shortest path from gene i to gene j. Coefficient β ij Indicates the type of effect of gene i on gene j. If the effect type is activation or induction, then β ij =1; if it is inhibition, then β ij =-1. sign(log2(fc j )) indicates that if the fold change of disease-related differentially expressed gene j is greater than 0, the gene is upregulated in the disease state and the value is 1; otherwise, the value is -1. Drug effect on disease-related gene j represents the sum of the effects of m drug-perturbed genes on gene j. If the sum is greater than 0, it means that the drug upregulates the disease gene j; otherwise, it downregulates it. In addition, if the disease gene is upregulated (downregulated) and the drug downregulates (upregulates) the gene, the inhibition score is +1; otherwise, the inhibition score is -1; if the drug effect on the gene is 0, the inhibition score is 0.

[0048] S5, comparing the inhibition scores to obtain candidate drugs; specifically, the most effective drugs are screened by comparing the inhibition scores, and the drugs with the highest scores are selected as candidate drugs; the larger the inhibition score, the better the potential effect of the drug, otherwise the drug effect is poorer.

[0049] The present invention is described in detail below with reference to specific embodiments:

[0050] 1. Data selection and data processing

[0051] The present invention selected the gene expression profile datasets of rectal cancer from the public database GEO: GSE8671, GSE9348, and GSE23878. The specific sample and tissue information of the three datasets are shown in Table 1. And the drug perturbation gene expression data from the CMap database.

[0052] Table 1 Rectal cancer expression dataset

[0053]

[0054] Data preprocessing: GEO database colorectal cancer datasets (GSE8671, GSE9348, and GSE23878) and drug perturbation expression datasets were used to obtain differentially expressed genes in each dataset using the limma package. Each disease dataset was then divided into cancer samples and normal samples based on sample name label information.

[0055] 2. Constructing cancer-related tissue-specific networks

[0056] Signaling pathways closely related to rectal cancer were downloaded from the KEGG pathway database, including colorectal cancer, Wnt signaling pathway, PI3K-Akt signaling pathway, Apoptosis, Cell cycle, MAPK signaling pathway, TGF-b signaling pathway, ErbB signaling pathway, mTORsignaling pathway and p53 signaling pathway. These pathways were integrated to form a specific network related to rectal cancer.

[0057] 3. Calculating the inhibitory score of drugs on rectal cancer

[0058] To investigate the relationship between drug-perturbed genes and colorectal cancer-related genes within tissue-specific networks, we applied inhibition scores based on signaling pathway networks to quantitatively assess drug-disease relevance. Specifically, we calculated the shortest distance between drug-perturbed genes and disease-related genes within tissue-specific networks, and finally calculated the drug-disease inhibition score to demonstrate the drug's effect on the disease.

[0059] IV. Screening of drug candidates

[0060] The above steps can obtain the inhibition scores of multiple drugs on rectal cancer, so drugs with higher inhibition scores are more likely to be potential therapeutic drugs for rectal cancer. The results show the top 30 drugs, as shown in Table 2. The comparison results with the classical method show that the method based on signal pathway network has a good prediction effect. Figure 2 shown.

[0061] Table 2 Top 30 drug candidates for rectal cancer

[0062] Drug name Inhibition score Ranking Drug name Inhibition score Ranking Trichostatin.A 720 1 Emetine 584 16 Anisomycin 712 2 Ciclopirox 563 17 Cicloheximide 679 3 Ouabain 562 18 Vorinostat 679 4 Ellipticine 555 19 Lanatoside.C 668 5 Methotrexate 528 20 Digoxigenin 658 6 Monobenzone 520 21 Azacyclonol 657 7 Astemizole 486 22 Digoxin 655 8 Mebendazole 481 23 Thioridazine 648 9 Niclosamide 477 24 Tanespimycin 638 10 Gossypol 471 25 Helveticoside 618 11 Digitoxigenin 467 26 Allantoin 616 12 Mefloquine 456 27 Lycorine 604 13 Camptothecin 453 28 Sirolimus 599 14 Amiodarone 441 29 Cephaeline 599 15 Suloctidil 440 30

[0063] The most effective drug was screened by comparing the inhibition scores, and Trichostatin A with the highest score was selected as a candidate drug.

[0064] This invention focuses on a tissue-specific network-based drug repositioning method, which offers significantly higher predictive power than conventional methods. This method addresses the issues of low overlap between disease-associated and drug-associated genes and incomplete drug-target relationships. Furthermore, this method approaches drug repositioning from a signaling pathway network perspective and, based on comparative inhibition scores, has broad applicability for recommending cancer drugs.

[0065] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

[0066] It should be noted that the functional units / modules in the various embodiments of the present invention may be integrated into a single processing unit / module, each unit / module may exist physically separately, or two or more units / modules may be integrated into a single unit / module. The aforementioned integrated units / modules may be implemented in the form of hardware or software functional units / modules.

[0067] Through the description of the above embodiments, it will be clear to those skilled in the art that it should be understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code, or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following units: an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a processor, a controller, a microcontroller, a microprocessor, other electronic units designed to implement the functions described herein, or a combination thereof. For software implementation, part or all of the processes of the embodiments can be completed by instructing the relevant hardware through a computer program.

[0068] When implemented, the above program can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any media that facilitates the transmission of computer programs from one place to another. The storage medium can be any available medium that can be accessed by a computer. Computer-readable media can include, but are not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer.

Claims

1. A drug repositioning method based on tissue property network, characterized in that: The steps include: S1, obtain the gene expression profile data of cancer and normal samples and the gene expression profile data of drug perturbation; S2, data preprocessing of the gene expression profile data of cancer and normal samples and drug-perturbed gene expression profile data, and screening of differentially expressed genes DEGs; Differentially expressed gene DEGs include drug differentially expressed genes and disease differentially expressed genes; S3, obtain signaling pathways related to a specified type of cancer from the KEGG database and establish a tissue-specific network based on signaling pathway integration; S4, integrating drug differentially expressed genes and disease differentially expressed genes into tissue-specific networks, calculating the shortest paths between drug differentially expressed genes and disease differentially expressed genes, and finally calculating the inhibition score value of each drug-disease relationship; S5, comparing the inhibition scores and selecting the drug with the highest score as a candidate drug; The S4 includes: The suppression score S is calculated using the following formula: Where m represents the number of differentially expressed genes affected by drug perturbations, n represents the number of differentially expressed genes associated with diseases, log2(fc i ) represents the log value of the fold change of drug-related differentially expressed gene i, log2(fc j ) represents the log value of the fold change of disease-related differentially expressed gene j; β ij is the coefficient, L ij represents the shortest path from gene i to gene j, where i represents the drug-perturbed gene.

2. The drug repositioning method based on tissue property network according to claim 1, characterized in that: When calculating the suppression score S: If drug-perturbed gene i is upstream of disease-related differentially expressed gene j, then gene i has an effect on gene j, otherwise it has no effect; L ij represents the shortest path from gene i to gene j; coefficient β ij Indicates the type of effect of gene i on gene j. If the effect type is activation or induction, then β ij =1; if it is inhibition, then β ij =-1; sign(log2(fc j )) indicates that if the fold change of disease-related differentially expressed gene j is greater than 0, the gene is an upregulated gene in the disease state and the value is 1, otherwise the value is -1; Disease-related differentially expressed gene j represents the sum of the effects of m drug-perturbed genes on gene j. If the sum is greater than 0, it means that the drug upregulates the disease-related differentially expressed gene j, otherwise it downregulates it. If the state of the disease-related differentially expressed gene is up-regulated or down-regulated, and the drug affects the down-regulation or up-regulation of the gene, the inhibition score is +1, otherwise the inhibition score is -1; if the drug affects the state of the gene to 0, the inhibition score is 0.

3. The drug repositioning method based on tissue property network according to claim 1, characterized in that: Said S1 comprises: Obtain expression profile data of cancer and normal samples from the GEO database; Obtain drug-induced gene expression profile data from the CMap database.

4. The drug repositioning method based on tissue property network according to claim 1, characterized in that: The S2 includes: The limma package in R language was used to preprocess the expression profile data of cancer and normal samples and the gene expression profile data of drug perturbations to screen for differentially expressed genes.

5. The drug repositioning method based on tissue property network according to claim 4, characterized in that: The screening conditions for the drug differentially expressed genes and the disease differentially expressed genes are both FDR<0.01.

Images