KR102269208B1 - Apparatus and method for multitrait genome wide association - Google Patents

Abstract

The present invention relates to multivariate association analysis technology, and more particularly, to a multivariate association analysis apparatus and method in consideration of causal mutations. According to an embodiment of the present invention, when multiple mutations are used, higher power can be obtained than single mutation-associated studies, so it can be used as a tool to understand the mechanisms of various diseases, and furthermore, the prevention, early detection, pharmaceutical, It can be used in treatment techniques.

Description

Multivariate association analysis apparatus and method {APPARATUS AND METHOD FOR MULTITRAIT GENOME WIDE ASSOCIATION}

The present invention relates to multivariate association analysis technology, and more particularly, to a multivariate association analysis apparatus and method in consideration of causal mutations.

The genome (genome) of all living things undergoes spontaneous mutations that cause mutations in the sequence of their offspring in the course of evolution. Mutations may be evolutionarily beneficial or detrimental or neutral. In some cases, the mutation gives a lethal disadvantage and is not passed on to the offspring. In other cases, there is an evolutionary benefit to the species, eventually inserting DNA into most of the species, effectively becoming an ancestral form. In many cases, these ancestral forms and variants survive and coexist among the species. The coexistence of multiple forms of a sequence results in polymorphism. RFLP, STR, VNTR, SNP, and the like are known for such polymorphisms.

RFLP: Restriction Fragment Length Polymorphism

STR: Short Tandem Repeat

VNTR: Variable Number Tandem Repeat

SNP: Single Nucleotide Polymorphism

Among them, Snip (SNP) is a single nucleotide polymorphism, which takes the form of a single nucleotide variation between individuals of the same species. Snip is a representative form of genetic variation. Most DNA is similar to each person, but different parts are called mutations. When the snip affects a phenotype such as a disease, the polynucleotide comprising the snip can be used as a primer or probe to diagnose such a disease. The phenotype refers to height, diabetes level, gene expression level, and the like. Monoclonal antibodies that specifically bind snip may also be used in the diagnosis of disease. However, these snips simply found that single-nucleotide polymorphisms exist on the human genome or cDNA, and did not reveal their effect on the phenotype. Some of these functions are known, but most are unknown.

In the case of the genome association test, which is a study on which mutations among numerous genetic variations in humans are related to which phenotype, for example, height, diabetes level, gene expression rate, etc., snip A univariate association study, which refers to a one-to-one association study between one phenotype and one phenotype, was performed.

When multiple association studies are performed with a gene unit consisting of several snips, studies have been conducted to look at multiple mutations in a gene at once. However, among multiple snips constituting a gene, the phenotype to be studied or the causal mutation that actually controls the gene expression Without analysis, the association of several snips constituting a gene was simply calculated by conducting a single mutation association study, and an approximate method using the average value was used, so accurate analysis results could not be derived.

Background art of the present invention is disclosed in Korean Patent Registration No. 10-1722107.

An object of the present invention is to provide an apparatus and method for analyzing multiple mutation associations for analyzing the relationship between multiple mutations and phenotype or expression rates of genes.

According to one aspect of the present invention, there is provided a genome association research apparatus for the relationship between genetic variation and phenotype or gene expression rate.

A multivariate association analysis apparatus according to an embodiment of the present invention includes a linkage disequilibrium investigation unit that investigates linkage disequilibrium between multiple variants to be analyzed, and a null test quantity that generates a likelihood ratio distribution of a null hypothesis by using linkage disequilibrium. Distribution generator, threshold definition unit that measures the critical likelihood ratio test value corresponding to the significance level from the likelihood ratio distribution of the null test quantity, the alternative test quantity calculator that measures the likelihood ratio test value of the alternative test quantity, and the critical likelihood ratio test It may include a correlation determination unit for determining the association by comparing the value and the test value of the likelihood ratio of the alternative test quantity.

According to another aspect of the present invention, there is provided a multivariate association analysis method and a multivariate association analysis recording medium capable of implementing the same.

In the method and recording medium according to an embodiment of the present invention, in a genomic association research method for the association between genetic variation and phenotype or gene expression rate, linkage disequilibrium investigation for calculating the degree of linkage disequilibrium with a correlation coefficient between snips to be analyzed Step, a null test amount distribution generation step of generating the likelihood ratio distribution of the null hypothesis by using the linkage disequilibrium matrix, a threshold value definition step of deriving a critical likelihood ratio test value of significance level from the null test amount likelihood ratio distribution, The allele test amount calculation step, which calculates the likelihood ratio test value of the allele test amount using the expression rate of the gene to be analyzed, and the critical likelihood ratio test value corresponding to the significance level and the likelihood ratio test value of the allele test amount are compared It may include a linkage determination step of determining whether the gene is related.

According to an embodiment of the present invention, it can be utilized to find genes whose expression is regulated by proximity snips in gene tissue expression studies.

In addition, according to an embodiment of the present invention, when Snip is used for genome-wide association research that is applied to disease-related research, it can be utilized to discover genes related to phenotypes and diseases.

In addition, according to an embodiment of the present invention, when multiple mutations are used, higher power can be obtained than single mutation-associated studies, so it can be used as a tool to understand the mechanisms of various diseases and furthermore, prevention, early detection, and restriction of various diseases. , can be used in treatment techniques.

1 is a view for explaining a multivariate association analysis apparatus according to an embodiment of the present invention.
2 is a block diagram illustrating a multivariate association analysis apparatus according to an embodiment of the present invention.
3 is a block diagram illustrating a multivariate association analysis apparatus according to an embodiment of the present invention.
4 is a graph showing that the multivariate association analysis apparatus according to an embodiment of the present invention well controls Type1 errors.
5 is a graph showing that the multivariate association analysis apparatus according to an embodiment of the present invention has higher power than the conventional single mutation method.
6 is a graph in which the multivariate association analysis apparatus according to an embodiment of the present invention finds a genome that is related to gene expression.

Since the present invention can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and will be described in detail through detailed description. However, this is not intended to limit the present invention to specific embodiments, and it should be understood to include all modifications, equivalents and substitutes included in the spirit and scope of the present invention. In describing the present invention, if it is determined that a detailed description of a related known technology may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted. Also, as used herein and in the claims, the terms "a" and "a" and "a" are to be construed to mean "one or more" in general, unless stated otherwise.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings, and in the description with reference to the accompanying drawings, the same or corresponding components are given the same reference numerals, and the overlapping description thereof will be omitted. do it with

1 is a diagram for explaining a multivariate association analysis apparatus according to an embodiment of the present invention;

Referring to FIG. 1 , the apparatus for analyzing multiple mutation associations according to an embodiment of the present invention determines the association between a gene and multiple mutations by using the gene expression rate extracted from the population and information on related snips. In the present invention, gene X is a gene to be analyzed, and the N snips are linked snips for analyzing multiple mutations, not single mutations.

Conventionally, since it is difficult to find a causal mutation due to linkage disequilibrium in genomic data, the genome was analyzed in an approximate way without considering the causal mutation. Here, the linkage disequilibrium refers to a phenomenon in which snip B always appears next to snip A when snip A is present. Conventionally, snip A is a causative mutation that regulates the expression rate of gene X. However, if snip A and snip B always appear together due to linkage imbalance, snip B is also seen as the expression rate of gene X, so the actual cause mutation is snip A, snip Because there was confusion about whether it was B or both, when an approximate genome analysis method was used without considering the causal mutation, the effect did not appear properly in the real data.

The present invention introduces the recently developed Fine Mapping technology to find an efficient and accurate causal variability, and grafts it to the Likelihood Ratio method to model the causal variability for the first time. it's about how Here, the fine mapping technology is to identify the true causative mutation among several snips that are candidates for the causative mutation. In the above example, among Snip A and Snip B, which snip is the cause of gene X? way to cover it up. Likelihood Ratio is one of the methods to determine which hypothesis is correct between the null hypothesis and the alternative hypothesis. In the present invention, the null hypothesis assumes that any snip does not affect the gene, and the alternative hypothesis is a hypothesis that assumes that a specific snip affects the gene. The likelihood ratio test method is used to determine which hypothesis is correct.

In the multivariation association research method according to an embodiment of the present invention, an N:1 association between multiple (N) snips and a specific phenotype of interest is studied, but in this case, which snip is the causal variant of the phenotype, the model is considered. In addition, in many cases, the present invention can be applied not only to the case where only summary statistics (a value indicating the degree of 1:1 association between a single snip and a single phenotype) are valid even without raw data that is not provided in many cases, but also the linkage disequilibrium structure of the data However, accurate analysis is possible without being limited by the number and type of causal mutations.

2 is a block diagram illustrating a multivariate association analysis apparatus according to an embodiment of the present invention.

Referring to FIG. 2 , the multivariate association analysis apparatus according to an embodiment of the present invention includes a linkage disequilibrium investigation unit 110 , a null test quantity distribution generation unit 120 , a threshold value definition unit 130 , and an alternative test quantity calculation unit. 140 and the association determining unit 150 .

The linkage disequilibrium investigation unit 110 calculates the degree of linkage disequilibrium among the N mutations (multiple mutations) to be analyzed as a correlation coefficient between the multiple mutations. The linkage disequilibrium investigation unit 110 measures a linkage disequilibrium structure between snips of the mutations to be analyzed using the correlation of the mutations.

The null test amount distribution generating unit 120 generates a test statistic (null test amount) likelihood ratio distribution of the null hypothesis by using linkage imbalance. The null test amount distribution generating unit 120 obtains the null hypothesis distribution by obtaining the likelihood ratio test values of the test statistics generated under the null hypothesis using the correlation matrix obtained by the linkage imbalance investigation unit 110 .

The threshold value measurement unit 130 derives a threshold value of a significance level from the null test quantity likelihood ratio distribution. The significance level is a reference value used in statistical hypothesis testing.

The allele test amount calculation unit 140 is a likelihood ratio test value of the allele test amount obtained from a given phenotype or gene expression rate.

The association determination unit 150 compares the threshold value corresponding to the significance level with the likelihood ratio test value of the alternative test amount to determine whether association is present. N used in the experiment when the likelihood ratio test value according to the gene expression rate is large by comparing the likelihood ratio test value of the allele test amount calculated by the allele test amount calculating unit 140 and the threshold value obtained by the threshold value measuring unit 130 It is defined that the dog snip and the corresponding gene are related.

3 is a block diagram illustrating a multivariate association analysis method according to an embodiment of the present invention.

Referring to FIG. 3 , in step S310, the multivariate association analysis apparatus calculates the degree of association disequilibrium using the correlation coefficient between N snips. Here, the correlation coefficient is a value calculated with respect to how similar one snip is to another snip as the similarity of snips. The multivariate association analysis apparatus measures the linkage disequilibrium structure between snips of the mutations by using the correlation of the mutations. Here, snip has a binary number of 0/1 for a minor allele and a major allele, or a real number representing each probability. Also, the snip correlation is a numerical value indicating how similar the snip is. For example, (000) and (000) are exactly the same, so the correlation is 1, and (000) and (111) are the opposite, so the correlation is corresponds to 0. On the other hand, the association disequilibrium is a structure of the correlation of each pair of snips to be analyzed. For example, when two snips are tested, the association disequilibrium takes the form of a 2 x 2 matrix.

The multivariate association analysis apparatus may derive an N x N size correlation matrix (linkage disequilibrium matrix) between the N mutations as a result value when N mutation sequences of M names are input. In more detail, the input mutant sequence is converted into a binary form (0 for allele, 1 for major allele) or a real value expressed as a probability of allele and major allele. After transformation, a correlation matrix of size N x N is derived from a matrix of size M x N having 0/1 or real values.

Input: N mutant sequences of M humans

(e.g. 10 variant sequences of 100 humans)

person 1. ACCCTCCACG

person 2. CTCCTAGACG

…

100 people. CCCTAACTGC

Result: correlation matrix of size N x N between N variants (linkage disequilibrium matrix)

In step S320, the multivariate association analysis apparatus generates a probability ratio distribution of the null test quantity of the null hypothesis by using the association imbalance matrix.

In more detail, the multivariate association analysis apparatus obtains the null hypothesis distribution of the likelihood ratio test value of the test statistic using the correlation matrix obtained in step S310. That is, the multivariate association analysis device calculates the distribution of the likelihood ratio test value of the summary statistics when the null hypothesis that none of the N snips affects the gene expression rate is true. The multivariate association analysis apparatus uses the fact that the test statistic follows a multivariate normal distribution, and extracts the test statistic by using the association disequilibrium matrix obtained in step S310 as variance-covariance. That is, the test statistic is a multivariate multivariate with a mean value of 0 and a variance value of the correlation matrix structure obtained in step S310 under the assumption that the null hypothesis (the hypothesis that none of the N snips affect the expression rate of a gene) is true. It is calculated by extracting from a multivariate normal distribution.

는 연관 불균형 조사부(110)에서 구한 연관불균형 구조를 나타낸다.

는 분석하는 변이들의 각자 원인변이 상태를 나타낸다. 검정 통계량의 조건부 확률과 원인변이 확률은 각각 다변수정규분포에서 도출한다.S is the test statistic, and C is the actual causal variation status. Since it is the null hypothesis, the mean is 0.

represents the linkage disequilibrium structure obtained by the linkage disequilibrium investigation unit 110 .

represents the state of each causal mutation of the mutations to be analyzed. The conditional probability and the causal variance probability of the test statistic are derived from the multivariate normal distribution, respectively.

The calculated test statistic is calculated using the following Equation 2 likelihood ratio test value formula.

Here, C is a binary value representing Snip's causal mutation state and represents all possible causal mutation states. Co is a form of C and represents a state in which there is no causal mutation.

는 가능한 C에서 Co를 제외한 경우의 집합으로 적어도 하나이상의 원인변이가 존재하는 경우를 나타낸다. 즉 우도비율검정값식의 분자는 테스트하는 N개의 스닙 중 적어도 한 개 이상이 원인변이일 경우의 사후확률 합을 나타내며 분모는 N개의 스닙 중 원인변이가 하나도 없는 경우의 사후확률을 나타낸다.

is a set of possible C except Co, and represents a case in which at least one causal mutation exists. That is, the numerator of the likelihood ratio test value formula represents the sum of posterior probabilities when at least one of the N snips to be tested is a causal variation, and the denominator represents the posterior probability when there is no causal variation among the N snips to be tested.

By repeating the process of extracting the test statistic and calculating the likelihood ratio test value using the method described above, the likelihood ratio distribution of the null test quantity of the null hypothesis is generated.

In step S320, the input values and result values for obtaining the test value of the likelihood ratio of the null test quantity are as follows.

Input: Linkage disequilibrium correlation matrix obtained in step S310

Result: The likelihood ratio distribution of the null test assuming the linkage disequilibrium correlation matrix obtained in step S310

In step S330, the multivariate association analysis apparatus defines a threshold value for deriving the critical likelihood ratio test value of the significance level from the likelihood ratio distribution of the null test quantity. Here, the significance level is expressed as alpha(α) as a reference value used in statistical hypothesis testing, and the significance level is 5% if 95% reliability is the standard, and the significance level value is 0.05. In the case of genome research, since more than one association study is tested, alpha(α)=5 * 10^-8 is usually used instead of 0.05 considering that it is a multiple hypothesis test.

In step S330, the input value and the result value defining the threshold value for deriving the critical likelihood ratio test value of the significance level from the likelihood ratio distribution of the null test quantity are as follows.

Input: Likelihood Ratio Tests

Result: One likelihood-ratio test corresponding to the threshold

For example, the multivariate association analysis apparatus extracts the test statistic in step S320 and repeats the process of calculating the likelihood ratio test value about 100000000 times, assuming that there are 100,000,000 likelihood ratio test values, 10^8 * 0.05th likelihood A ratio test value can be set as a threshold value.

In step S340, the multivariate association analysis apparatus measures the likelihood ratio test value of the allele test amount using the likelihood ratio test value calculation formula using the expression rate of the gene X. Here, the likelihood ratio test value may be calculated using the likelihood ratio test value calculation formula of Equation 2 used in step S320 described above.

In step S350, the multivariate association analysis apparatus compares the critical likelihood ratio test value corresponding to the significance level and the likelihood ratio test value of the allele test amount to determine whether the snip and the gene are related.

In more detail, the multivariate association analysis device compares the likelihood ratio test value of the allele test value with the critical likelihood ratio test value, and when the likelihood ratio test value of the allele test amount according to the gene expression rate is large, N snips used in the experiment and the gene are defined as being related.

4 is a graph showing that the multivariate association analysis apparatus according to an embodiment of the present invention rejects data generated under the null hypothesis as much as the significance level specified, that is, that the Type1 error is well controlled.

Referring to FIG. 4 , the multivariate association study method of the present invention was applied to 10 randomly selected genes. Obtain 10^8 null test likelihood ratio test values through step S320, and test the critical likelihood ratio corresponding to significance levels of 0.1 and 0.05 using 10^8/2 null test likelihood ratio test values corresponding to half of them This is a graph depicting the ratio of the null hypothesis rejected when the values are calculated and the remaining 10^8/2 null test values are used as the alternative test values likelihood ratio test values. When a significance level of 0.1 is applied, it can be seen that the ratio of rejection of the null hypothesis for all 10 genes is close to 0.1, and when a significance level of 0.05 is applied, it can be seen that the ratio of rejection of the null hypothesis for all 10 genes is 0.05. That is, it can be seen that the multivariate association analysis device rejects the data generated under the null hypothesis as much as the significance level specified (controls the Type 1 error well).

5 is a graph showing that the multivariate association analysis apparatus according to an embodiment of the present invention has higher power than the existing single-variation analysis method.

Referring to FIG. 5 , it is possible to confirm the degree of improvement in the degree (power) of gene association found using 10^8 allele test amounts for five different linkage disequilibrium structures ((LD1 to LD5). It can be seen that the mutation association analysis device found more correlation as the power increased by ~40% (0.4) than the existing method, where the alternative test amount was derived from the multivariate normal distribution and the mean was 2, 2.5, 3, 3.5, Each case can be obtained in the case of 4, 4.5, 5, 5.5, 6. Here, the existing simple multivariate association analysis method studies the association of several snips constituting a gene without analyzing the causal mutation, each of which is a single mutation association study. The method of taking the maximum value after performing the calculation was used.

6 is a graph in which the apparatus for analyzing multiple mutation associations according to an embodiment of the present invention finds genomes related to gene expression better than the conventional method.

로 표기)를 찾은 것을 볼 수 있다.In more detail, FIG. 6 shows, for example, the degree of association between the gene expression rate and the snip constituting the gene when testing 50 genes. The multivariate association analysis method of the present invention (represented as MultiTest on the graph) and the conventional This is a graph showing the comparison of the results when a simple multivariate association analysis (indicated as univariate on the graph) was performed. Referring to FIG. 6 , the y-axis indicates the degree of correlation, and the higher the number, the higher the correlation. Here, the horizontal dotted line indicates the threshold, which indicates the critical likelihood ratio test value of the significance level. If the correlation is higher than this, it can be considered that eGene, that is, proximity snip, is an expression change gene that controls the expression rate of genes. When using the present invention, six new expression change genes (on the graph) that were not found by the existing multiple mutation method

) can be found.

Using the present invention, it is possible to find a larger number of expression-altering genes than in the existing method.

So far, the present invention has been looked at focusing on the embodiments thereof. Those of ordinary skill in the art to which the present invention pertains will understand that the present invention can be implemented in a modified form without departing from the essential characteristics of the present invention. Therefore, the disclosed embodiments should be considered in an illustrative rather than a restrictive sense. The scope of the present invention is indicated in the claims rather than the foregoing description, and all differences within the scope equivalent thereto should be construed as being included in the present invention.

100: multivariate association analysis device
110: linkage imbalance investigation unit
120: null test quantity distribution generating unit
130: threshold definition unit
140: alternative test amount calculation unit
150: association judgment unit

Claims (12)

In the multivariate association analysis apparatus for studying the N:1 association between N snips and a specific phenotype of interest in the association study in the genomic association study on the association between genetic variation and gene expression rate,
Linkage disequilibrium investigation unit that investigates linkage disequilibrium between mutations to be analyzed;
a null test quantity distribution generating unit that generates a null test quantity likelihood ratio distribution of the null hypothesis;
a threshold value definition unit for measuring a threshold likelihood ratio test value;
an alternative test amount calculation unit for calculating a likelihood ratio test value of the alternative test amount; and
Including an association determination unit for determining whether the snip and the gene are related,
The linkage imbalance investigation unit,
Using the correlation of the snips to calculate the linkage disequilibrium structure between snips with the correlation coefficient between the N snips,
The correlation coefficient is
The similarity of the snips is a value calculated by how similar one snip and the other snip are,
Obtain a linkage disequilibrium matrix by deriving a correlation matrix between snips in N * N size from N variant sequences of M people,
The null test quantity distribution generating unit,
The test statistic of the null hypothesis is calculated using [Equation 1] using the correlation matrix, which is the linkage imbalance structure obtained by the linkage imbalance investigation unit,
A distribution is generated by calculating the test value of the likelihood ratio of the null test statistic of the test statistic through [Equation 2],
The threshold value definition unit,
Deriving the critical value of the significance level from the distribution of the likelihood ratio of the null test quantity,
The threshold value is a test value corresponding to the significance level of 0.05 among the test values of the likelihood ratio of the null test value,
The alternative test amount calculation unit
Calculate the likelihood ratio test value of the allele test amount from the gene expression rate,
The association determination unit,
By comparing the critical likelihood ratio test value and the likelihood ratio test value of the allele test amount, the association between snip and the gene is determined,
Fine mapping and modeling techniques by introducing a mutation caused by combining the likelihood ratio test how mutations associated with multiple analysis devices.
[Equation 1]

S: test statistic
C: causal mutation state

: Linkage Disequilibrium Structure

: State of each causal mutation of the mutations to be analyzed

[Equation 2]

Likelihood Ratio Test (LRT); Likelihood Ratio Test Value
C: Binary value representing the causal mutation state of Snip
C ₀ : A form of C, in which there is no causal mutation.

: A set of possible C _{except C 0} (at least one causal mutation exists)
delete delete delete delete delete In the multivariate association analysis method, a multivariate association analysis apparatus for studying the N:1 association between N snips and a specific phenotype of interest in a genome association study on the association between genetic variation and gene expression rate,
Linkage disequilibrium investigation step of examining linkage disequilibrium between multiple mutations to be analyzed;
a null test quantity distribution generating step of generating a null test quantity likelihood ratio distribution of the null hypothesis using a linkage imbalance matrix;
a threshold definition step of measuring a critical likelihood ratio test value of the significance level from the likelihood ratio distribution of the null test quantity;
an alternative test amount calculation step of measuring the likelihood ratio test value of the alternative test amount; and
Including an association determination step of determining whether the snip and the gene are related,
The linkage imbalance investigation step is,
using the correlation of the snips to calculate the linkage imbalance structure between snips with the correlation coefficient between the N snips,
The correlation coefficient is
The similarity of the snips is a value calculated by how similar one snip and the other snip are,
The step of generating the null test quantity distribution comprises:
The test statistic of the null hypothesis is calculated using [Equation 1] using the correlation matrix, which is the linkage disequilibrium structure obtained in the linkage disequilibrium investigation step,
A distribution is generated by calculating [Equation 2] for the null test value likelihood ratio test value of the test statistic,
The correlation matrix is
Obtain a linkage disequilibrium matrix by deriving a matrix between snips in N * N size from N mutated sequences of M people,
The threshold definition step is,
Deriving the critical value of the significance level from the likelihood ratio distribution of the null test quantity obtained in the step of generating the null test quantity distribution,
The alternative test amount calculation step is,
Calculate the likelihood ratio test value of the allele test amount from the gene expression rate,
The association determination step is,
Comparing the critical likelihood ratio test value obtained in the threshold definition step and the likelihood ratio test value of the allele test amount obtained in the allele test amount calculation step to determine whether the snip and the gene are related,
A multivariate association analysis method that models causal variability by incorporating the likelihood ratio test method by introducing a precise mapping technique.
[Equation 1]

S: test statistic
C: causal mutation state

: Linkage Disequilibrium Structure

: State of each causal mutation of the mutations analyzed

[Equation 2]

Likelihood Ratio Test (LRT); Likelihood Ratio Test Value
C: Binary value representing the causal mutation state of Snip
C ₀ : A form of C, in which there is no causal mutation.

: A set of possible C _{except C 0} (at least one causal mutation exists)
A computer program recorded on a computer-readable recording medium for executing the method for analyzing the multivariate association of claim 7 .
delete delete delete delete

Images