KR20190000169A - System and method of biomarker identification for cancer recurrence prediction - Google Patents

Abstract

The present invention present invention provides a method where a system operated by at least one processor discovers a biomarker for cancer recurrence prognosis prediction. The method comprises: a step of collecting large-scale gene expression data of cancer patients of a specific type; a step of analyzing similarity between the large-scale gene expression data to classify the large-scale gene expression data into a plurality of clusters; a step of extracting genes commonly expressed in gene expression data classified into each cluster to determine the genes as a cancer-associated gene set of the corresponding cluster; and a step of selecting at least one gene set among a plurality of cancer-associated gene sets consisting of the cancer-associated gene set of each cluster as a biomarker based on a hazard ratio calculated by cancer recurrence prognosis-related information of each cancer-associated gene set.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a biomarker identification system and a biomarker identification system,

The present invention relates to a biomarker.

Cancer is divided into breast cancer, liver cancer, and lung cancer depending on the location of the cancer. Even cancer of the same kind causes cancer due to mutation of different genes or abnormality of intracellular function. Therefore, even in patients with the same type of cancer, the prognosis, such as the recurrence of cancer or the timing of death, may be different for each patient, and the response may be different for the same drug. Therefore, when predicting the prognosis of cancer patients or establishing a treatment strategy, the need for personalized prediction of recurrence prognosis is increasing. Finally, it is necessary to classify cancer patients first to establish appropriate treatment strategies for cancer diagnosis.

Mammaprint, Prosigna and Oncotype DX have been used as diagnostic kits to predict the recurrence of cancer patients. Mammaprint and Prosigna have been approved by the FDA in 2007 and 2013, respectively. Since the diagnostic kits so far are classified using data having labels, there is a problem that they are overfitted with the data of the diagnostic kits. In addition, the conventional diagnostic kits are difficult to accurately classify the patients because they discriminate the whole gene without considering the characteristics of the subgroup of the patient, and do not consider the mechanism of the disease, Is low.

The problem to be solved by the present invention is to analyze a gene expression data of a large-scale cancer patient, to extract a gene set coexpressing each patient group, and to identify a clinical marker for predicting cancer recurrence in a plurality of extracted gene sets And to a system and method for excavation.

A method for locating a biomarker for predicting recurrence prognosis of a cancer, the system being operated by at least one processor according to an embodiment, comprising the steps of: collecting large scale gene expression data of a specific type of cancer patient; Classifying the large-scale gene expression data into a plurality of clusters by analyzing similarities among the clusters, extracting genes commonly expressed in gene expression data classified into each cluster, and determining the set of cancer-associated genes in the cluster Based on the harzard ratio calculated from the cancer-associated gene set of each cancer-associated gene set and the cancer-relapse prognosis-related information of each cancer-associated gene set, at least one gene set of plural cancer- And selecting the biomarker.

The classification into the plurality of clusters may classify the random gene expression data into at least one cluster by bi-clustering the large-scale gene expression data.

The classification into the plurality of clusters may be performed by bi-clustering the large-scale gene expression data such that each cluster has a Pearson correlation of 0.9 or more and a significance P value of 0.05 or less.

The step of determining the set of cancer-associated genes in the cluster is performed by extracting expression genes from gene expression data classified into each cluster, expressing commonly in gene expression data of a reference number or more among the expression genes of each cluster, Can be determined as the cancer-associated gene set of the cluster.

The step of collecting the large-scale gene expression data may normalize and store the expression amount of each gene expression data.

The cancer recurrence prognosis related information can be obtained from gene expression data of each cancer-associated gene set, including recurrence and survival time for each patient.

In the step of selecting the biomarker, the risk ratio of each cancer-associated gene set is evaluated, and at least one gene set having a significant risk ratio among the plurality of cancer-associated genes may be selected as the biomarker.

A system operating with at least one processor in accordance with another embodiment is a method for locating a biomarker for predicting recurrence prognosis of a cancer using bi-clustering techniques to identify large-scale gene expression data Classifying cancer genes into a plurality of clusters, determining genes that are commonly expressed in each of the plurality of clusters as a cancer-associated gene set of the clusters, setting a plurality of cancer- And determining the biomarker by evaluating the prognostic significance of each cancer-associated gene set determined as the biomarker candidate based on the cancer recurrence prognostic information included in each gene expression data do.

The classification into the plurality of clusters may classify random gene expression data among the large-scale gene expression data into at least one cluster.

The cancer recurrence prognostic information may include a recurrence rate and a survival time for each patient.

Wherein the step of selecting the biomarker comprises the steps of: determining a harzard ratio indicating a degree of influence of the cancer-associated gene set on the basis of information on cancer recurrence prognosis of the arbitrary cancer-associated gene set determined as the biomarker candidate; And determine whether to select the arbitrary cancer-associated gene set as the biomarker based on the calculated significance of the risk ratio.

According to another embodiment of the present invention, there is provided a system for identifying a biomarker for predicting cancer recurrence prognosis, which comprises classifying the cancer patients based on large-scale gene expression data of a specific cancer patient, A cancer-associated gene set extracting apparatus, and a plurality of gene sets each consisting of a patient group-specific gene set, the prognostic significance of the gene set is evaluated based on information on the cancer recurrence prognosis of each gene set, And a biomarker selection device for selecting at least one of the biomarkers as a biomarker.

The cancer-associated gene set extracting apparatus may classify the large-scale gene expression data into a plurality of patient groups by bi-clustering the large-scale gene expression data.

The cancer recurrence prognosis information includes the recurrence rate and the survival time of each patient. The biomarker selection device is a device for determining the risk of cancer gene deletion based on information on cancer recurrence prognosis of each gene set, It is possible to calculate the harzard ratio and decide whether to select the gene set as the biomarker based on the significance of the calculated hazard ratio.

The biomarker may be a combination of genetic sets having a prognostic significance value of more than a reference among genes set commonly expressed in each patient group.

According to the embodiment of the present invention, the problem of overexposure to the data labeled in the diagnostic kit can be solved conventionally, and the cancer-associated gene set can be extracted according to the gene characteristics of each cancer patient. Therefore, the biomarker according to the embodiment of the present invention is composed of a set of cancer-related genes extracted from gene characteristics of various patients, so that a personalized cancer relapse prognosis can be predicted.

The diagnostic agent or diagnostic kit developed using the biomarker according to the embodiment of the present invention can generate a profit that is equal to that of a new drug due to high profitability in relation to investment. BT-based biomarker content related to molecular biological information such as genomics, proteomics, and epigenomics is combined with IT-based hardware such as medical equipment related to healthcare and web-based healthcare service through the biomarker according to the embodiment of the present invention It can contribute to the development of healthcare and pharmaceutical industry.

1 is a configuration diagram of a biomarker excavation system according to an embodiment of the present invention.
2 is a view for explaining a biomarker discovery method according to an embodiment of the present invention.
3 and 4 are flowcharts of a biomarker excavation method according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

Throughout the specification, when an element is referred to as "comprising ", it means that it can include other elements as well, without excluding other elements unless specifically stated otherwise. Also, the terms " part, "" module, "and the like, which are described in the specification, refer to a unit for processing at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software have.

A biomarker is a unique molecular information derived from proteins or genes that can be used to detect biologic changes. It is also used to diagnose various intractable diseases such as cancer, stroke, and dementia, to predict recurrence, and to develop new drugs.

Among the various cancers, breast cancer belongs to a cancer that has been studied in a subtype based on molecular characteristics. Traditionally, three proteins (estrogen receptor, progesterone receptor, human epidermal growth factor receptor 2 (HER2) Luminal A, Luminal B, HER2-overexpressed, and Basal-like (Triple-negative) For example, the HER2-overexpressed subtype is administered the Trastuzumab drug to suppress HER2. However, even with this targeted treatment strategy, Trastuzumab shows about 26% of drug reactivity, and when used with drugs such as Doxorubicin or Docetaxel, it also has about 65% drug reactivity. Because of the heterogeneity of these cancers, diagnostic kits are used to predict the recurrence of cancer patients. Diagnostic kits, such as Mammaprint, Prosigna, and Oncotype DX, use biomarkers to predict recurrence prognosis using the expression of mRNA extracted from breast cancer tissues in breast cancer patients.

However, existing diagnostic kits include analyzed genes (genes) that are related to specific cancers, namely, 70 labeled gene data (Mammaprint), 50 gene data (Prosigna) and 21 gene data (Oncotype DX) There is a limit to be classified only by the known gene, since the patient is classified through. In addition, existing diagnostic kits are discriminated as whole genes related to specific cancers, and thus there is a limit that can not reflect the gene characteristics of each patient group.

In the following, a system and method for discovering biomarkers for solving such conventional problems will be described in detail.

FIG. 1 is a configuration diagram of a biomarker excavation system according to an embodiment of the present invention, and FIG. 2 is a diagram for explaining an exemplary biomarker excavation method according to an embodiment of the present invention.

Referring to FIG. 1, the biomarker excavation system (10) analyzes gene expression data of a large-scale cancer patient and finds out a gene set expressing a specific cancer patient group. Thus, a clinical feature Branches uncover biomarkers. To this end, the biomarker search system 10 includes a cancer-associated gene set extraction apparatus 100 and a biomarker selection apparatus 200 for predicting the cancer recurrence prognosis (hereinafter simply referred to as "biomarker selection apparatus") 200. The biomarker search system 10 may obtain necessary gene expression data in conjunction with at least one open gene expression database (e.g., Gene expression omnibus) 300.

The cancer-associated gene set extraction apparatus 100 extracts a plurality of candidate gene sets having a possibility of biomarker, and clusters gene expression data of large-scale cancer patients to extract gene sets that are commonly expressed for each patient group. The cancer-associated gene set extraction apparatus 100 operates as at least one processor and includes a gene expression data collection unit 110, a gene expression data clustering unit 130, and a cancer-associated gene set determination unit 150 for a specific cancer. .

The gene expression data collection unit 110 acquires gene expression data of a cancer patient of a certain kind from the gene expression database 300 and normalizes and stores the expression amount of each data. For example, the gene expression data collection unit 110 may acquire gene expression data extracted from the breast cancer tissue of the patient in the gene expression database 300 using keywords such as breast, tumor, cancer, and neoplasm. The gene expression data collection unit 110 can perform preprocessing by quantile normalizing the amount of expression of each data.

The gene expression data clustering unit 130 analyzes gene expression data of all patients and clusters gene expression data having similarity. The gene expression data clustering unit 130 generates a cluster (patient group) based on the similarity between the data, so that each patient can be classified into at least one cluster. For example, referring to FIG. 2, the gene expression data of patient P1 may be classified into a plurality of clusters (C1, C2, C3) It is predicted to have similarity of gene expression data with another patient P3 belonging to cluster C2 with similarity. At this time, since the patient P2 and the patient P3 do not belong to the same cluster, the patient P2 and the patient P3 have the same kind of cancer, but the gene expression data may be different.

,

가 있을 때 계산되는 피어슨 상관계수 r을 계산하는 식이다. 수학식 2는 두 벡터 X₁={X₁₁,X₁₂,..X_1n}, X₂={X₂₁,X₂₂,..X_2n}와 평균값

,

, 각 벡터의 분산값 s²으로 계산된 표준편차인 s_p가 있을 때, t 값을 계산하는 식이다.The gene expression data clustering unit 130 may classify gene expression data (samples) into a plurality of clusters using a bi-clustering technique. Each cluster may consist of data with a Pearson correlation of 0.9 or greater and a P value of 0.05 or less. The gene expression data clustering unit 130 can obtain the P value based on the expression difference calculated based on the t-test and the false discovery rate. The Pearson correlation coefficient (r) can be calculated as shown in Equation (1), and the t-test can be calculated as Equation (2). Equation (1) are the two vectors _{X = {X 1, X 2} , .. X n}, Y = {Y 1, Y 2, .. Y n} , and the average value

,

Is calculated by the following equation. Equation 2 can be expressed by two vectors X ₁ = {X ₁₁ , X ₁₂ , X _1n }, X ₂ = {X ₂₁ , X ₂₂ , .. X _2n }

,

, And s _{p, which} is the standard deviation calculated by the variance value s ² of each vector, is calculated.

[Equation 1]

&Quot; (2) "

The cancer-associated gene set determining unit 150 extracts the genes commonly expressed in each cluster as a set of genes of the cluster (patient group). Referring again to FIG. 2, the cancer-associated gene set determining unit 150 extracts a cancer-associated gene set (for example, a C1 gene set, a C2 gene set, a C3 gene set, etc.) for each cluster.

Methods for extracting common expression genes can be varied. For example, the cancer-associated gene set determining unit 150 extracts expression genes from gene expression data belonging to each cluster. The cancer-associated gene set determiner 150 is commonly expressed in a plurality of gene expression data (for example, five or more genes) among the expression genes of each cluster, and the cancer-associated gene set determiner 150 determines whether the Pearson correlation coefficient is equal to or greater than a reference value Genes can be determined as the cancer-associated gene set of the cluster.

The cancer-associated gene set determination unit 150 extracts a plurality of cancer-associated gene sets composed of cluster-specific cancer-associated gene sets as biomarker candidates.

The biomarker selection apparatus 200 evaluates the prognostic significance of a plurality of cancer-associated gene sets extracted from the cancer-associated gene-set extraction apparatus 100 to determine at least one gene set that is statistically significant among a plurality of cancer- Biomarkers for prediction of recurrence prognosis are selected.

The biomarker selection apparatus 200 confirms the cancer relapse prognosis-related information from the gene expression data obtained in the gene expression database 300. Because gene expression data is extracted from the patient's tissues, it may include information on cancer recurrence prognosis, including the recurrence of individual patients, survival time information, and the like.

The biomarker selection apparatus 200 calculates a prognostic index of each cancer-associated gene set based on information on cancer relapse prognosis. The prognostic indicator can be a hazard ratio. That is, the biomarker selection apparatus 200 calculates a risk ratio for each gene set (cluster). The risk ratio of a cancer-associated gene set can be calculated as a Cox proportional hazards model, which analyzes the influence of factors influencing survival, and can be calculated from Equations (3) through (5).

Equation 3 is a risk function h (t), T is a random variable for survival time, and t is a specific value for a random variable T. h ₀ is a reference risk function, where X ₁ = 0 and X ₂ = 0.

&Quot; (3) "

&Quot; (4) "

&Quot; (5) "

The biomarker selection device 200 calculates a P value based on a log-rank test to evaluate the significance of the risk ratio of each gene set. The biomarker selection device 200 determines the gene set having the lowest P value of the risk ratio as a biomarker of a specific cancer. In some cases, the biomarker selection apparatus 200 can determine a plurality of gene sets having a P value of the risk ratio equal to or less than a reference value as a biomarker of a specific cancer.

The biomarker selection apparatus 200 can perform tens or hundreds of thousands of significance calculations according to the amount of collected information, and can perform parallel processing using distributed computing for high-speed calculation.

3 and 4 are flowcharts of a biomarker excavation method according to an embodiment of the present invention.

Referring to FIG. 3, the cancer-associated gene set extraction apparatus 100 collects gene expression data of a cancer patient of a certain kind (S110).

The cancer-associated gene set extraction apparatus 100 analyzes the similarity of the collected gene expression data and classifies the gene expression data into a plurality of clusters (S120).

The cancer-associated gene set extraction apparatus 100 extracts the genes commonly expressed in each cluster as a cancer-associated gene set of the cluster (S130).

The cancer-associated gene set extraction apparatus 100 extracts a plurality of cancer-associated gene sets composed of cluster-specific cancer-associated gene sets as biomarker candidates (S140).

Referring to FIG. 4, the biomarker selection apparatus 200 receives a plurality of cancer-associated gene sets composed of cancer-associated gene sets per cluster as biomarker candidates (S210).

The biomarker selection apparatus 200 calculates the risk ratio of each cancer-associated gene set based on information on the cancer recurrence prognosis of gene expression data (S220).

The biomarker selection apparatus 200 evaluates whether the risk ratio of each cancer-associated gene set is significant (S230).

The biomarker selection apparatus 200 selects at least one gene set having a significant risk ratio among a plurality of cancer-associated gene sets as a biomarker of a specific cancer (S240). The biomarker contains the set of genes with the lowest P value for the hazard ratio. The biomarker can be composed of a plurality of sets of genes whose P value of the risk ratio is below a reference.

As described above, according to the embodiment of the present invention, the gene set common to each patient group is identified through analysis of gene expression data of a large-scale cancer patient, so that a gene set reflecting various gene characteristics can be obtained, Biomarkers that can predict the recurrence of cancer in a patient can be identified. A biomarker according to an embodiment of the present invention can be used to diagnose cancer and ultimately establish a therapeutic strategy suitable for the patient's condition.

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

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, It belongs to the scope of right.

Claims (15)

A method for locating a biomarker for predicting recurrence prognosis of a cancer, the system being operated by at least one processor,
Collecting large-scale gene expression data of certain types of cancer patients,
Classifying the large-scale gene expression data into a plurality of clusters by analyzing similarities among the large-scale gene expression data,
Extracting genes commonly expressed in gene expression data classified into each cluster, and determining the cancer-associated gene set of the cluster; and
Based on the harzard ratio calculated from the cancer recurrence prognosis information of each cancer-associated gene set, at least one gene set of plural cancer-associated gene sets composed of cancer-associated genes in each cluster is used as a biomarker Selection step
The method comprising: The method of claim 1,
The step of classifying into the plurality of clusters
And bi-clustering the large-scale gene expression data to classify the random gene expression data into at least one cluster. 3. The method of claim 2,
The step of classifying into the plurality of clusters
Clustering the large-scale gene expression data such that each cluster has a Pearson correlation of 0.9 or more and a significance P value of 0.05 or less. The method of claim 1,
The step of determining the cancer-associated gene set of the cluster
Expression genes are extracted from gene expression data classified into respective clusters, and genes having a Pearson correlation coefficient equal to or higher than a reference value are commonly expressed in gene expression data of a reference number or more among the expression genes of each cluster, How to find biomarkers to determine. The method of claim 1,
The step of collecting the large scale gene expression data
Wherein the expression amount of each gene expression data is normalized and stored. The method of claim 1,
Wherein the cancer recurrence prognosis related information includes recurrence and survival time of each patient and is obtained from gene expression data of each cancer-associated gene set. The method of claim 1,
The step of selecting the biomarker
Determining a risk ratio of each cancer-associated gene set to be significant, and selecting at least one gene set having a significant risk ratio among the plurality of cancer-associated gene sets as the biomarker. A method for locating a biomarker for predicting recurrence prognosis of a cancer, the system being operated by at least one processor,
Classifying large-scale gene expression data of a specific type of cancer patient into a plurality of clusters using bi-clustering technology,
Determining genes that are commonly expressed in each of the plurality of clusters as a cancer-associated gene set of the cluster,
Determining a plurality of cancer-associated gene sets composed of cancer-associated genes in each cluster as candidate biomarkers, and
Selecting the biomarker by evaluating the prognostic significance of each cancer-associated gene set determined as the biomarker candidate on the basis of cancer relapse prognostic information included in each gene expression data
Wherein the gene expression data comprises at least one of: 9. The method of claim 8,
The step of classifying into the plurality of clusters
And sorting random gene expression data among the large-scale gene expression data into at least one cluster. 9. The method of claim 8,
Wherein the cancer recurrence prognostic information includes a recurrence rate and a survival time of each patient. The method of claim 9,
The step of selecting the biomarker
Based on the information on the cancer recurrence prognosis of the arbitrary cancer-associated gene set determined as the biomarker candidate, calculates a harzard ratio indicating the degree of survival of the arbitrary cancer-associated gene set, Determining whether the selected cancer-associated gene set is a biomarker based on the significance of the cancer-associated gene set. A system for locating biomarkers for predicting cancer recurrence prognosis,
A cancer-associated gene set extraction apparatus for classifying cancer patients based on large-scale gene expression data of a specific type of cancer patient and extracting a gene set commonly expressed for each patient group, and
The prognostic significance of the set of genes is evaluated based on the information on the cancer recurrence prognosis of each gene set among a plurality of gene sets composed of gene sets for each patient group and at least one of the plurality of gene sets is selected as a biomarker Biomarker selection device
/ RTI > The method of claim 12,
The cancer-associated gene set extracting apparatus
Wherein the large-scale gene expression data is classified into a plurality of patient groups by bi-clustering the large-scale gene expression data. The method of claim 12,
The cancer recurrence prognostic information includes the recurrence rate and the survival time of each patient,
The biomarker selection device
Based on the information on cancer recurrence prognosis of each gene set, a harzard ratio, which indicates the degree to which the gene set affects survival, is calculated, and based on the calculated significance of the risk ratio, The system determines whether to select. The method of claim 12,
The biomarker
Wherein the prognostic significance of the gene sets common to all patient groups is greater than or equal to a reference.

Images