KR102361615B1 - Method for drug repositioning based on drug responding gene expression features - Google Patents

Abstract

The drug re-creation method using drug response gene expression characteristics according to an embodiment of the present invention comprises the steps of selecting analyzable drug response gene expression data from a public drug response gene expression DB, and a significant gene in the selected drug response gene expression data. Selecting a drug response gene showing expression, selecting a drug response gene set showing a significant correlation from the selected drug response gene expression data, and the significant association with the drug response gene showing the significant gene expression and outputting a drug similarity value by evaluating similarity between drugs using a drug response gene set showing

Description

Method for drug repositioning based on drug responding gene expression features

The present invention relates to a drug re-creation method using drug response gene expression characteristics. More specifically, a method that enables comparative analysis of the gene expression amount and gene expression pattern for drug response in drug re-creation, minimizes errors due to individual differences, and compares the functional similarity of drugs together It relates to a new drug re-creation method that can more accurately predict the new functions and efficacy of existing drugs and failed drugs by developing

In certain conditions of the cell, the gene expression profile provides information at the transcriptional level. This is a popular method in biological analysis, including the identification of genes with expression patterns associated with some phenotypes.

The identification of molecular markers can be used to diagnose disease states. In addition, it can be used to infer the regulatory relationship of genes suitable for biological procedures.

A representative method for analyzing gene expression patterns is to identify differentially expressed gens (DEGs) or co-expressed genes related to a specific phenotype. Another method is to compare the expression patterns of one modality to those of other modalities. This comparison also provides another way to query large gene expression databases, such as gene expression (Omnibus and ArrayExpress). In order to add persuasiveness to the comparison of these expression patterns, various expression similarity study methods have been devised.

With the advent of the connectivity map, the use of gene expression similarity studies that provide a large-scale library of expression patterns corresponding to human cell lines treated with various small molecules has been utilized for drug discovery and development.

The Connectivity map also provides a comparative method to measure the similarity between a gene feature provided as input and a drug response reference gene expression profile. In order to measure similarity, a query gene characteristic consisting of genes that are highly or less expressed from the user is required.

Connectivity map calculates a score for each drug response reference profile by estimating whether the upregulated gene is at the top or bottom of the drug response reference gene expression profile. The usefulness of data sets and comparative approaches have been demonstrated by conducting case studies involving drug repurposing, elucidation of uncharacterized molecular mechanisms of action, and mimicry of specific biological states.

Innovative approaches lead to the development of various methods for the large-scale analysis of drug response patterns. Although these variant methods employ different patterns of matching approaches to measure similarity, the main concept of similarity measurement for the use of highly or underexpressed genetic features from query modalities is still the most widely used.

The set of co-expressed genes can be identified by clustering the drug response pattern. In cluster analysis, it is assumed that the expression pattern of the set of co-expressed genes is related to all conditions. Alternatively, double colonization is used to find related gene sets in some conditions. A set of co-expressed genes newly obtained or significantly disrupted from drug response patterns would be more suitable for describing the expression status of cells treated with a particular drug. Several studies have identified differentially co-expressed genes known to be involved in cancer, even if their expression levels do not change significantly. For this reason, genes with differential co-expression patterns can be a characteristic of different value that characterizes not only drug treatment patterns, but also differentially expressed genes (DEGs).

Meanwhile, in the related art, Korean Patent Registration No. 10??1292843 discloses a technology for generating related disease information for a drug. However, this prior art discovers disease information related to a drug by comparing the target substance and biological network information obtained from drug information, but does not reflect condition-specific information in the case of biological network information.

Therefore, the technical task of the present invention is to collect drug response gene expression data obtained from cancer cell lines to which drugs were administered and cancer cell lines to which drugs were not administered from a connectivity map to minimize errors due to differences between individuals. To provide a drug re-creation method that enables discovery of drugs applicable to specific diseases by selecting genes significantly expressed in .

According to one feature of the present invention, the drug re-creation method comprises the steps of selecting analyzable drug response gene expression data from a public drug response gene expression DB, a drug response showing significant gene expression in the selected drug response gene expression data Selecting a gene, selecting a drug response gene set showing a significant correlation from the selected drug response gene expression data, and a drug response gene showing a significant gene expression and a drug response gene showing a significant correlation and outputting a drug similarity value by evaluating similarity between drugs using the set.

The step of outputting the drug similarity value comprises:

Performing drug similarity quantification based on the drug response gene based on the individual Pearson correlation coefficient (PCC) between each drug response gene obtained from the two drugs to be compared, based on the drug response gene set obtained from the two drugs to be compared performing drug similarity quantification based on the drug response gene set to calculate the average value of differentially coexpressed gene sets based similarity score (DCS), and the value obtained by performing the drug response gene-based drug similarity quantification and the drug response gene set based drug After adjusting the size of the similarity quantification value, outputting a final drug similarity value by taking each average.

In addition, the step of selecting the drug response gene,

After performing the conversion of the gene expression level for each gene from the drug response gene expression data, a significant gene that is less than or equal to a predefined threshold as a result of the significance evaluation is selected as the drug response gene,

The step of selecting the drug response gene set,

From the drug response gene expression data, a gene network is constructed by selecting a gene pair in which the difference in inertia values calculated for each drug-treated and non-drug-treated condition meets a predefined threshold, and a gene forming a maximum click in the gene network. The three may be selected as the drug response gene set.

According to an embodiment of the present invention, the accuracy of the similarity between drugs is increased by integrating the gene expression amount and correlation, and the difference between individuals in selecting genes significant for the drug response based on the gene expression amount through modification of the statistical analysis method errors can be minimized.

In addition, it is possible to significantly reduce drug development time and cost by applying a novel similar drug obtained by a drug similarity comparison method to a disease in which a specific drug is used.

1 is a block diagram of a drug re-creation system using drug response gene expression characteristics according to an embodiment of the present invention.
FIG. 2 is a flowchart illustrating the operation of the drug response gene expression data selection module of FIG. 1 .
3 is a flowchart illustrating the operation of the drug response gene selection module of FIG. 1 .
4 is a flowchart illustrating the operation of the drug response gene set selection module of FIG. 1 .
5 is a flowchart illustrating the operation of the drug-to-drug similarity evaluation module of FIG. 1 .

Hereinafter, with reference to the accompanying drawings, the embodiments of the present invention will be described in detail so that those of ordinary skill in the art to which the present invention pertains can easily implement them. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part "includes" a certain component, it means that other components may be further included, rather than excluding other components, unless otherwise stated.

In addition, the terms “…unit” and “…module” described in the specification mean a unit that processes at least one function or operation, which may be implemented as hardware or software or a combination of hardware and software.

Hereinafter, a drug re-creation method using the drug response gene expression characteristic according to an embodiment of the present invention will be described in detail with reference to the drawings.

1 is a block diagram of a drug re-creation system using drug response gene expression characteristics according to an embodiment of the present invention.

1, the drug re-creation system 100 includes a drug response gene expression data selection module 101, a drug response gene selection module 103, a drug response gene set selection module 105, and a drug similarity evaluation module ( 107).

The drug response gene expression data selection module 101 selects analyzable drug response gene expression data from a public drug response gene expression DB (not shown).

Here, the open drug response gene expression DB (not shown) is a connectivity map that provides gene expression data through treatment/non-treatment of small molecules including drugs to cancer cell lines (https://www. .broadinstitute.org/cmap/) information was secured and constructed.

Drug response data included expression patterns of 6,100 genes in 5 cell lines treated with 1,309 small molecules. These expression patterns are grouped by batch ID, and then each group is normalized by a robust multi-array average. After extracting expression values from each experiment, they are merged into a single matrix.

The drug response gene selection module 103 selects a drug response gene showing significant gene expression from drug response gene expression data under drug administration and non-administration conditions.

Since the drug response gene expression data provided by the drug response gene expression data selection module 101 includes those measured in five cancer cell lines, a conversion of gene expression levels for each gene is performed to correct for differences by condition. At this time, the correcting equation is the following equation (1).

은 약물비투여 조건에서의 전환된 유전자 발현양을 의미하고,

은 약물투여 조건에서의 전환된 유전자 발현양을 의미하며,

은 약물 처리 조건에서의 초기 유전자 발현양을 의미하며,

은 약물비처리 조건에서의 초기 유전자 발현양을 의미한다. here,

means the converted gene expression amount in the drug non-administration condition,

means the converted gene expression amount under drug administration conditions,

denotes the initial gene expression level under drug treatment conditions,

denotes the initial gene expression level in the non-drug treatment condition.

A simple correction that shifts the initial gene expression level based on the average value can maximize the difference in gene expression level in drug-treated/untreated conditions. Using the corrected values, t??test and FDR (false discovery rate) multiple testing corrections were used to evaluate gene significance to select genes with P??value <0.05 in drug-treated/untreated conditions. carry out A drug response gene is provided for each drug by selecting the top 100 genes and the bottom 100 genes based on the expression level of significant genes.

The drug response gene set selection module 105 selects a drug response gene set showing a significant correlation in drug response gene expression data under drug administration and non-administration conditions.

The drug-to-drug similarity evaluation module 107 evaluates the similarity between drugs using the gene selected from the drug response gene selection module 103 and the gene set selected from the drug response gene set selection module 105 .

In order to compare the similarity of two drugs, for example, drug A and drug B using the drug response gene, the top 100 and bottom 100 drug response genes respectively obtained from the two drugs to be compared are input. In order to express the 200 gene values of drug A as one, the drug-treated expression amount/drug-untreated gene expression amount is calculated and 200 values are assigned. For 200 genes selected in drug A, calculate the individual correlation coefficient (PCC) between two drugs by calculating one value in drug B, and calculate PCC with drug A based on drug B in the same way Thus, the drug response gene-based drug similarity quantification is performed by taking the average of the two PCC values.

In addition, in order to compare the similarity of two drugs, for example, drug A and drug B using the drug response gene set, the drug response gene set obtained from the two drugs to be compared is input. The gene set input from the two drugs is divided into a high correlation gene set and a low correlation gene set by the drug response gene set selection module.

For all gene pairs in the highly correlated gene set, the association difference value is calculated by subtracting the PCC value in the drug-treated condition from the PCC value in the non-drug treatment condition. The formula is:

은 유전자쌍의 연관성이 약물처리 조건에서는 없다가 약물비처리 조건에서 나타나는 경우의 차등 공동발현 값을 의미하고,

은 유전자쌍의 약물비처리 조건에서의 PCC 값을 의미하며,

은 유전자쌍의 약물처리 조건에서의 PCC 값을 의미한다.Here, DC is a differentially coexpression value.

denotes the differential co-expression value when the gene pair association is absent in the drug-treated condition but appears in the non-drug-treated condition,

is the PCC value in the non-drug treatment condition of the gene pair,

denotes the PCC value in the drug treatment condition of the gene pair.

For all gene pairs in the low association gene set, the association difference value is calculated by subtracting the PCC value in the drug-treated condition from the PCC value in the drug-treated condition. The formula is:

은 유전자쌍의 연관성이 약물비처리 조건에서는 없다가 약물처리 조건에서 나타나는 경우의 차등 공동발현 값을 의미한다.here,

denotes the differential co-expression value in the case where the gene pair association does not exist in the drug-treated condition but appears in the drug-treated condition.

Based on the drug response gene set selected in drug A, the same gene pair in drug B is performed as above. The correlation value of one drug response gene set selected in drug A with drug B is calculated by calculating the DC value of the possible gene pair corresponding to the type of drug response gene set in drug A from both drug A and drug B. Calculate by subtracting the DC value of the gene pair in drug B from the DC value of , and take the average value. The formula is:

은 약물 A의 차등 공동발현 값이고,

은 물 B의 차등 공동발현 값을 의미한다.Here, DCM is a differentially coexpression module, and n is the number of combinations of all gene pairs of a gene set to be calculated among a plurality of drug response gene sets selected for drug A.

is the differential co-expression value of drug A,

denotes the differential co-expression value of water B.

To calculate the similarity quantification between drug A and drug B, all DCMs are calculated for all drug response gene sets selected from drug A and averaged. The formula is:

Here, DCS is a differentially coexpressed gene sets based similarity score. N is the number of drug response gene sets selected from drug A. In the same way, the gene set-based DCS with drug A is calculated based on drug B, and the drug response gene set-based drug similarity quantification is performed by taking the average of the two DCS values.

In order to integrate the drug response gene set-based similarity, scaling is performed with a value of 0 to 1, which converts the score to a value between 0 and 1. The conversion formula is as follows.

는 약물반응 유전자셋 기반 얻어진 유사성 값이며,

는 약물반응 유전자 기반 얻어진 유사성 값이다.

is the similarity value obtained based on the drug response gene set,

is the similarity value obtained based on the drug response gene.

The final drug similarity value is provided by taking the average of the scaled drug response gene-based drug similarity value and the drug response gene set-based drug similarity value.

FIG. 2 is a flowchart illustrating the operation of the drug response gene expression data selection module of FIG. 1 .

2, the drug response gene expression data selection module 101 normalizes the expression level of the drug response gene expression data obtained from the public drug response gene expression DB (not shown) (S101). That is, the drug response gene expression data selection module 101 performs quantile normalization in order to remove the difference for each condition in the drug response gene expression data.

The drug response gene expression data selection module 101 selects 10 or more drugs from the normalized data according to each condition, that is, drug administration and drug non-administration (S103). That is, for each drug, 10 or more drugs for each drug administration/non-administration condition are selected.

The drug response gene expression data selection module 101 provides the selected drug response gene expression data (S105). That is, it provides a list of selected drugs for similarity between final drugs and their gene expression data.

3 is a flowchart illustrating the operation of the drug response gene selection module of FIG. 1 .

Referring to FIG. 3 , the drug response gene selection module 103 converts the gene expression level of each gene from the drug response gene expression data provided from the drug response gene expression data selection module 101 ( S201 ).

The drug response gene selection module 103 performs gene significance evaluation (S203).

The drug response gene selection module 103 determines whether the gene significance evaluation result (p) is less than 0.05 (S205).

As a result of the judgment, if it is less than 0.05, the top 100 genes and the bottom 100 genes based on the expression level among the significant genes are selected (S207). And the selected drug response gene is provided (S209).

4 is a flowchart illustrating the operation of the drug response gene set selection module of FIG. 1 .

4, the drug response gene set selection module 105 randomly sets 10,000 gene pairs from the drug response gene expression data provided from the drug response gene expression data selection module 101 (S301).

For the gene pair set in step S301, an association difference distribution is configured as a PCC-based association difference value of drug treatment/non-treatment conditions (S303).

The difference between the association values of all gene pairs in the drug-treated/untreated condition is calculated (S305).

In the correlation difference distribution, a gene pair having a correlation value of the top 5% of the gene pair is selected (S307).

In addition, with respect to the gene pair set in step S301, a PCC-based association distribution of each drug treatment/non-treatment condition is configured (S309).

A PCC-based association value is calculated for all gene pair combinations in each drug-treated/untreated condition (S311).

In the correlation value distribution, one or more top 5% gene pairs are selected under drug-treated or non-treated conditions of each gene pair (S313).

A gene network is constructed by connecting the gene pairs selected under the two conditions (S315).

A gene set constituting a maximal clique in the gene network is selected (S317).

The selected gene set is provided as a drug response gene set (S319).

5 is a flowchart illustrating the operation of the drug similarity evaluation module of FIG. 1 .

Referring to FIG. 5 , the drug-drug similarity evaluation module 107 receives each drug response gene obtained from two drugs to be compared ( S401 ). And the drug response gene-based drug similarity is quantified (S403).

In addition, each drug response gene set obtained from two drugs to be compared is input (S405). And the drug response gene set-based drug similarity is quantified (S407).

The digitized values in steps S403 and S407 are scaled to a value of 0 to 1 (S409).

The average value of each drug similarity value is calculated (S411) to provide a drug similarity value (S413).

The embodiment of the present invention described above is not implemented only through the apparatus and method, and may be implemented through a program for realizing a function corresponding to the configuration of the embodiment of the present invention or a recording medium in which the program is recorded.

Although the embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto. is within the scope of the right.

Claims (3)

Selecting analyzable drug response gene expression data from the public drug response gene expression DB;
selecting a drug response gene showing significant gene expression from the selected drug response gene expression data;
Selecting a drug response gene set showing a significant correlation from the selected drug response gene expression data, and
Comprising the step of evaluating the similarity between drugs using the drug response gene showing the significant gene expression and the drug response gene set showing the significant association, and outputting a drug similarity value,
The step of selecting the drug response gene set,
From the selected drug response gene expression data, a gene network is constructed by selecting a gene pair in which the difference in the correlation value calculated for each drug treatment and non-drug treatment condition satisfies a predefined threshold, and the maximum click is formed in the gene network. selecting the gene set as the drug response gene set,
The step of outputting the drug similarity value comprises:
performing drug similarity quantification based on drug response gene based on the individual Pearson correlation coefficient (PCC) between each drug response gene obtained from two drugs to be compared;
performing drug similarity quantification based on the drug response gene set to calculate the average value of the differentially coexpressed gene sets based similarity score (DCS) based on the drug response gene set obtained from the two drugs to be compared; and
After adjusting the size of the drug response gene-based drug similarity quantification value and the drug response gene set-based drug similarity quantification value, taking each average and outputting a final drug similarity value
A drug re-creation method comprising a. delete The method of claim 1,
The step of selecting the drug response gene,
A drug re-creation method in which a significant gene, which is less than or equal to a predefined threshold as a result of significance evaluation, is selected as the drug response gene after performing conversion of the gene expression level for each gene from the drug response gene expression data.

Images