KR20120014512A - Method and apparatus for generating model for prognostic prediction using snps - Google Patents

Abstract

PURPOSE: A prognosis prediction model creation method using a SNP(Single Nucleotide Polymorphism) method and apparatus thereof are provided to create a prognosis prediction model by collecting information for a dominant model, a recessive model, and an additive model. CONSTITUTION: An SNP determination unit(1102) determines the SNP of a test target person by using hereditary models which statistically examine the hereditary feature of the SNP. A hereditary model determination unit(1103) determines similar hereditary model for the SNP. A standardization unit(120) standardizes coding values of SNP hereditary types based on the determined hereditary model. A model creation unit(130) creates a prognosis evaluation model which displays a correlation by analyzing a statistical evaluation algorithm.

Description

Method and apparatus for generating model for prognostic prediction using SNPs

A method and apparatus are provided for generating a model for prognostic prediction using genotype data of subjects for SNPs.

As the technology of analyzing genes of individuals has been developed after the discovery of DNA, research has been conducted to analyze the genotype of mutations and to reveal the polymorphism. Among the types of polymorphism, the polymorphism most commonly found in the human genome is Single Nucleotide Polymorphism (SNP). Human genetic factors are also associated with all human diseases, and humans also differ in their resistance, sensitivity and degree of disease to their genetic factors. In particular, SNPs correlate with human disease expression, so that the base sequences of specific positions representing SNPs in a group of patients with specific diseases differ from those of the control or normal group at the same positions. Turned out. Thus, diagnosis, prescription, and prevention of diseases are possible based on the difference in bases found through DNA sequences. Recently, research has been conducted to create a model that accurately predicts the prognosis of diseases such as individual susceptibility to diseases from base sequences associated with SNPs.

The technical problem to be achieved by at least one embodiment of the present invention is to provide a method and apparatus for generating a prognosis prediction del using SNP. The technical problem to be solved by this embodiment is not limited to the above-described technical problems, and other technical problems may exist.

According to one aspect, the method for generating a prognostic model predicts each of the SNPs obtained from the subjects with a plurality of genetic models that statistically test the genetic characteristics of the SNP, thereby making the subjects of the SNPs a patient group and a control group. Determining SNPs to classify automatically; Determining the most significant genetic model of the genetic models used in the assay for each of the determined SNPs; Normalizing encoded values of genotypes of each of the determined SNPs based on the determined genetic model; And generating a prognostic prediction model indicating a correlation between the genotypes of the determined SNPs and the prognosis of the subjects by analyzing the standardized results with a statistical prediction algorithm.

According to another aspect, a computer-readable recording medium having recorded thereon a program for executing the prognostic prediction model generation method on a computer is provided.

According to another aspect, the apparatus for generating a prognostic model predicts each of the SNPs obtained from the subjects with a plurality of genetic models that statistically test the genetic characteristics of the SNPs. An SNP determiner for determining significantly classified SNPs; A genetic model determiner for determining the most significant genetic model of the genetic models used in the assay for each of the determined SNPs; A normalizer for normalizing encoded values of genotypes of each of the determined SNPs based on the determined genetic model; And a model generator for generating a prognostic prediction model indicating a correlation between the genotypes of the determined SNPs and the prognosis of the subjects by analyzing the standardized results by a statistical prediction algorithm.

As described above, since one prognostic model is generated using all of the SNPs corresponding to a plurality of genetic models such as a dominant model, a recessive model, and an additive model, It is possible to generate an accurate prognostic model rather than generating a prognostic model using only some SNPs corresponding to one particular genetic model, for example a dominant model. That is, it is possible to generate one prognostic model that reflects all of the dominant, recessive, and additive different genetic characters of SNPs.

1 is a block diagram of an apparatus for generating a prognostic prediction model according to an embodiment of the present invention.
2 is a table showing genotype data of SNPs of subjects received by a data receiving unit according to an embodiment of the present invention.
Figure 3 is a table showing the data input according to the genotype in the SNP analysis unit according to an embodiment of the present invention.
4 is a table showing a result of standardizing values encoded by genotypes in a standardization unit according to an embodiment of the present invention.
5A and 5B are tables showing results simulated in a prognostic model generation apparatus according to an embodiment of the present invention.
6 is a flowchart of a method of generating a prognostic model according to an embodiment of the present invention.

Hereinafter, with reference to the drawings will be described embodiments of the present invention;

1 is a block diagram of a prognostic prediction model generating device 1 according to an embodiment of the present invention. Referring to FIG. 1, the apparatus for generating a prognostic prediction model according to the present embodiment 1 includes a processor 10, a data receiver 20, a storage unit 30, and an output unit 40. The processor 10 includes an SNP analyzer 110, a standardizer 120, a model generator 130, and a predictor 140. Here, the SNP analysis unit 110 is composed of an assay unit 1101, an SNP determination unit 1102, and a genetic model determination unit 1103. The processor 10 may be implemented as an array of a plurality of logic gates, or may be implemented as a combination of a general purpose microprocessor and a memory in which a program that may be executed in the microprocessor is stored. It will be appreciated by those skilled in the art that the present invention may be implemented in other forms of hardware. In this specification, only hardware components related to the present embodiment will be described in order to prevent blurring the features of the present embodiment. However, it will be understood by those skilled in the art that other general hardware components may be included in addition to the hardware components illustrated in FIG. 1.

Referring to FIG. 1, the prognosis prediction model generating device 1 of the present embodiment may analyze gene information analyzed from a genetic analysis device (not shown) that analyzes gene samples of test subjects responding to a DNA chip such as a microarray or the like. Create a prognostic model. Here, the genetic information used by the prognostic prediction model generating device 1 is genotype data for a plurality of SNPs of a plurality of subjects. Therefore, it will be understood by those skilled in the art that the gene analysis device may be any device as long as it is a gene analysis device that analyzes such genetic information.

Single Nucleotide Polymorphism (SNP) refers to genetic changes or variations that show differences in the nucleotide sequence (A, T, G, C) at any position in the DNA sequence. Form. In general, SNPs are known to occur at a frequency of about every 1,000 bp (base pair) in humans.

When SNPs occur in the nucleotide sequence of a coding region, defective or mutated proteins may be expressed, causing disease and have no effect. SNPs may also occur in the nucleotide sequence of the non-coding region. In other words, SNP is a genetic factor associated with human disease, and due to differences in SNPs, humans have different resistance, sensitivity, and degree of disease. Therefore, it is possible to diagnose, prescribe, and prevent disease through correlation with SNP and disease sensitivity. Prognostic prediction model generation device 1 of the present embodiment generates a prognostic prediction model for accurately predicting the prognosis such as individual sensitivity to the disease by using the SNP for the diagnosis, prescription and prevention of such human diseases.

The data receiver 20 receives genotype data of a plurality of SNPs of the subjects from a genetic analysis device (not shown). In FIG. 1, it is assumed that the genetic analysis device and the prediction model generation device 1 of the present embodiment are separate devices. However, the present invention is not limited thereto, and the prognostic prediction model generating device 1 may be implemented as a device embedded in the genetic analysis device. Hereinafter, as shown in FIG. 1, it will be described on the assumption that the genetic analysis device and the prognostic prediction model generating device 1 of the present embodiment are separate devices.

The processor 10 includes an SNP analyzer 110, a standardizer 120, a model generator 130, and a predictor 140.

The SNP analysis unit 110 selects significant SNPs based on a predetermined standard among the SNPs obtained by analyzing genotype data of SNPs obtained from the subjects, and also selects a genetic model most suitable for the selected SNPs. Decide Such genotype data is received through the data receiver 20.

2 is a table showing genotype data of SNPs of subjects received by the data receiving unit 110 according to an embodiment of the present invention.

Referring to FIG. 2, data on genotypes of 300 subjects from SNP 1 to SNP 100,000 are shown. Since each subject has different responsiveness to various factors such as sensitivity to disease, appearance, and longevity, each SNP may not exhibit the same genotype. Therefore, a prognostic prediction model can be made by analyzing and modeling the correlation between SNPs representing normal or abnormality for a specific factor and a specific factor.

Referring back to FIG. 1, the SNP analyzer 110 analyzes SNPs using the Max Test method. Hereinafter, the SNP analysis unit 110 will be described for the process of analyzing the SNPs using the Max Test method, but the detailed description of the Max Test method will be described below with the general knowledge in the art. If you grow up, you can understand.

The test unit 1101 tests the significance of the SNPs of each subject using each of the genetic models. The plurality of genetic models are models that statistically test the genetic characteristics of the SNP, and include a dominant model, a recessive model, and an additive model. The significance here includes the significance of whether each SNP is significantly classified into a patient group and a control group for a particular condition, and the significance of each of the genetic models used for the verification of the genetic characteristics of each SNP.

Based on the Max Test method in the test unit 1101, the method of testing each of SNPëë¤ using genetic models is as follows.

As described with reference to the table of FIG. 2, there may be tens or hundreds of thousands of SNPs obtained from the subjects. However, among these SNPs, there may be SNPs which do not act as genetic factors such as disease. That is, among the SNPs of the subjects, there may be SNPs that have nothing to do or have little influence on the prognosis for a certain factor when generating a prognostic model. Therefore, such SNPs need not be reflected in the prognosis prediction model generation. Therefore, SNPs that have nothing to do with the prognosis or have little influence on the prognosis can be selected by arranging the genotype data for any SNP as shown in Table 1 below and analyzing them statistically.

SNP No. 1 AA AB BB Total Response x ₀ x ₁ x ₂ x No response n ₀ -x ₀ n ₁ -x ₁ n ₂ -x ₂ n-x Total n ₀ n ₁ n ₂ n

In Table 1, each of Response and No Response refers to a case in which there is a response to a specific condition and a case in which there is no response to SNP genotype 1. In more detail, classification of response and no response means that SNP genotype 1 is classified into a case group and a case group for a specific condition, respectively, according to a general case-control study. It means classified. Each of x ₀ to x ₂ represents the number of AA, AB and BB genotypes present among genotype data of the subjects corresponding to the patient group (Response). And, each of n ₀ to n ₂ represents the total number of each of the AA, AB and BB genotypes among the genotype data of the subjects. Accordingly, the number of AA, AB, and BB genotypes among genotype data of the subjects corresponding to the control group (No Response) corresponds to n ₀ -x ₀ , n ₁ -x _1, and n ₂ -x ₂ , respectively.

However, in order to test based on the Max Test method in the test unit 1101, Table 1 is slightly modified and tested.

In more detail, first, genotypes of SNPs are encoded under different conditions according to genetic models such as dominant model, recessive model, and additional model. For example, the significance test using the dominant model encodes the genotypes AA, AB, and BB of 0, 1, and 1, respectively. Thus, according to the dominant model, AB and BB are represented by the same value. Next, for example, the significance test using the recessive model encodes S, genotypes AA, AB, and BB as 0, 0, and 1, respectively. Thus, according to the recessive model, AA and AB are represented by the same value. Lastly, for example, the significance test using the additional model encodes the genotypes AA, AB, and BB of 0, 1, and 2, respectively. Therefore, according to the additional model, each genotype is represented by a different value. In other words, they are encoded as shown in Table 2 below.

Genotype of SNP Genetic model sign (AA, AB, BB) Dominant (0, 1, 1) (AA, AB, BB) Recessive (0, 0, 1) (AA, AB, BB) Additive (0, 1, 2)

Genetic models such as dominant model, recessive model, and additional model are models modified from Table 1 by encoding as shown in Table 2.

Taking SNP 1 as an example, the dominant model, recessive model, and additional model to be applied to SNP 1 are shown in Tables 3, 4, and 5, respectively, below.

SNP No. 1 0 One Response x ₀ x ₁ No response y ₀ y ₁

Table 3 is a table for testing the significance of dominant models for SNP No. 1 genotypes. As described above, in the dominant model, genotypes of AB and BB were encoded as 1, and genotypes of AA were encoded as 0. And, similar to Table 1 described above, x ₀ and x ₁ for the SNP No. ₁ refers to the number of the encoded value is 0 and the number of 1 when corresponding to the patient group (Response) for a specific condition, respectively. And, y ₀ and y ₁ for the SNP No. ₁ refers to the number of the encoded value is 0 and the number of 1, respectively, when corresponding to the control (No Response) for a specific condition.

SNP No. 1 0 One Response x ₀ x ₁ No response y ₀ y ₁

Table 4 is a table for testing the significance of the recessive models for the genotype of SNP No. 1. As described above, in the recessive model, the genotype of BB was encoded as 1, and the genotypes of AA and AB were encoded as 0. And, x ₀ and x ₁ for the SNP No. ₁ refers to the number of the encoded value is 0 and the number of 1, respectively, when corresponding to the patient group (Response) for a specific condition. In addition, y ₀ and y ₁ with respect to SNP No. 1 mean a number having a coded value of 0 and a number of 1, respectively, when the response corresponds to a control (No Response) for a specific condition.

SNP No. 1 0 One 2 Response x ₀ x ₁ x ₂ No response y ₀ y ₁ y ₂

Table 5 is a table to test the significance of the additional models for the genotype of SNP No. 1. As described above, in the additional model, the genotype of AA was encoded as 0, the genotype of AB was encoded as 1, and the genotype of BB was encoded as 2. In addition, x ₀ , x _1, and x ₂ for SNP No. 1 mean the number of encoded values 0, the number 1, and the number 2, respectively, when corresponding to a patient group (Response) for a specific condition. In addition, y ₀ , y _1, and y ₂ with respect to SNP No. 1 mean the number of encoded values 0, the number 1, and the number 2, respectively, when the response corresponds to a control (No Response) for a specific condition.

The test unit 1101 tests significances with an dominant model as shown in Table 3, for example SNP # 1, tests significances with a recessive model as shown in Table 4, and tests significances with an additional model as shown in Table 5. That is, each of the three genetic models for one SNP is tested.

The test unit 1101 tests statistically summarized data as shown in Tables 3 to 5 using a significance test method such as a chi-squared test.

Chi-square test is a generally well-known statistical test method, and when applied to the above dominant model Table 3, a p-value indicating whether Response and No Response are classified significantly in SNP # 1 can be obtained. Moreover, each p-value can also be obtained about Table 4 which is a recessive model, and Table 5 which is an additional model.

Here, the p-value corresponds to a value indicating how significant the chi-square test is. The user may preset a threshold value, which is a criterion for determining the significance of the p-value, and test the significance by comparing the threshold value with each p-value.

The assay unit 1101 of this embodiment may test using the Cochran-Armitage trend test in the chi-square test method. Since the Cochran-Armitage trend test is obvious to those skilled in the art, detailed description thereof will be omitted.

The assay unit 1101 applies the same assay method to other SNPs in the same manner. That is, the genetic models shown in Tables 3 to 5 are also applied to other SNPs such as SNPs 2 to 100,000 of FIG.

Three p-values according to the test result of the three genetic models for each SNP in the assay unit 1101 are transmitted to the SNP determination unit 1102.

The SNP determiner 1102 determines SNPs that significantly classify the subjects among the SNPs of the subject into a patient group and a control group. In more detail, significant SNPs are determined based on a comparison result of threshold values set by a user and respective p-values obtained as a result of testing the respective genetic models for one SNP.

For example, based on a comparison between the smallest p-value of the three p-values obtained for SNP # 1 and the threshold, if the threshold is greater than the smallest p-value, the subjects were compared with the patient group. It is determined that it corresponds to the SNP that is classified as a control significantly. Similarly, based on the comparison of the p-values and thresholds of other SNPs, it is determined that SNPs with p-values less than the threshold value correspond to SNPs that are significantly classified into patient and control groups.

By selecting only such significant SNPs, a more accurate prognostic model can be generated with only SNPs having a high correlation in prognosis.

Genetic model determination unit 1103 determines the most significant genetic model of the genetic models used in the assay for each of the determined SNPs. That is, the genetic model determiner 1103 determines the genetic model only for the SNPs determined by the SNP determiner 1102.

In more detail, the genetic model having the lowest p-value among the p-value for the dominant model, the p-value for the recessive model, and the p-value for the additional model obtained for each of the determined SNPs is selected. Determined by the genetic model of This is because the lower the p-value, the more significant it is. The genetic model determined for an SNP reflects the genetic properties of that SNP, which means that the SNP has the genetic properties of either dominant, recessive or additive.

For example, if SNP 1 and SNP 2 are included among the SNPs determined by the SNP determiner 1102, the genetic model determiner 1103 may include the genetic models having different p-values for the dominant model of SNP 1. SNP 1 is determined to correspond to the dominant model when it is lowest compared to p-values, and SNP 2 is determined when the p-value for the recessive model of SNP 2 is lowest compared to p-values of other genetic models. Determine to correspond to the recessive model. That is, the genetic model determiner 1103 determines each genetic model for all of the SNPs determined by the SNP determiner 1102.

3 is a table showing the results analyzed by the SNP analysis unit 110 according to an embodiment of the present invention. Referring to FIG. 3, after testing the significance of each SNP in the assay unit 1101, a genetic model determined for each of the SNPs determined by the SNP determiner 1102 and the SNPs determined by the genetic model determiner 1103. Is shown. Each genotype of the determined SNPs is represented by a value encoded according to the determined genetic model. The encoded values are based on Table 2 discussed above.

As shown in Table 2, it can be seen that the conditions for encoding genotypes are different according to each genetic model. That is, only genotypes of SNPs corresponding to the dominant model have a value of 0, and AB and BB have a value of 1. However, genotypes of SNPs corresponding to the recessive model have AA and AB values of 0, and only BB have a value of 1. Furthermore, of the genotypes of SNPs corresponding to the additional model, BB has a value of 2 which is not found in other genetic models.

In other words, the encoding conditions are different for each of the determined SNPs according to the determined genetic model. Therefore, the values of 0, 1, and 2 are valid for only one SNP, but the determined SNPs are not valid because the respective SNPs are different from each other and thus cannot be used for generating a prognostic prediction model. Therefore, there is a need for a method that can apply the same criterion to each of the SNPs to uniformly display each value as values for the same criterion.

In the related art, the above problems cannot be solved, and only prognostic prediction models are generated by selecting only SNPs corresponding to one genetic model. For example, a prognostic model was generated using only genotypes of SNPs corresponding to the dominant model. However, the prognosis for a particular condition affects not only SNPs corresponding to the dominant model, but also SNPs corresponding to other genetic models, the recessive model and the additional model, so that the prediction model can be generated by using only one genetic model. Will fall.

Therefore, the prognostic prediction model generating apparatus 1 of the present embodiment uniformly displays each value as values for the same reference by standardizing values encoded by genotypes of each SNP for each SNP. For this reason, the prognostic model of the present embodiment is not generated by using only one genetic model but reflects all of the determined genetic models, that is, the results using the dominant model, the recessive model, and the additional model.

The standardization unit 120 normalizes the encoded values of genotypes of each of the SNPs determined based on the determined genetic model. That is, the standardization unit 120 normalizes values encoded by different conditions for each SNP based on the type of the genetic models determined to values having the same reference for each of the determined SNPs.

The standardization unit 120 generally uses a standardization method such as Equation 1 below. Equation 1 is applied for each determined SNPs and used to standardize.

Referring to Equation 1, z _i is a standardized value, Z _i is a value encoded by current genotypes, μ is an average of values encoded by current genotypes, and σ is a standard deviation of values encoded by current genotypes. .

를 이용하여 계산된다. 그리고, 현재 유전자형이 부호화된 값들의 표준편차인 σ는

를 이용하여 계산된다. 그러므로, 이와 같은 수학식들을 이용하면, (AA, AB, BB)가 부호화된 (Z₁, Z₂, Z₃)의 표준화된 값인 (z₁, z₂, z₃)를 얻을 수 있다. In more detail, when genotypes of any one SNP are expressed as (AA, AB, BB), each genotype is encoded with a value of (Z ₁ , Z ₂ , Z ₃ ). And, the probability that each genotype exists in this SNP is (f ₁ , f ₂ , f ₃ ). Then, μ, the average of the values encoded by the current genotype,

It is calculated using And, the standard deviation of the current genotype encoded values

It is calculated using Therefore, using these equations, it is possible to obtain (z ₁ , z ₂ , z ₃ ), which is a standardized value of (Z ₁ , Z ₂ , Z ₃ ) encoded with (AA, AB, BB).

로 계산된다. 결국, 수학식 1에 각각의 값들을 대입하면, 이 SNP가 부호화된 (Z₁, Z₂, Z₃)=(0, 1, 2)의 값은

로 표준화된다.For example, if a determined SNP corresponds to an additional model, (AA, AB, BB) of this SNP is encoded with a value of (0, 1, 2). Here, it can be assumed that the probability of each genotype is (0.25, 0.5, 0.25). Therefore, μ, which is the average of the encoded values of the current genotype, is calculated as μ = E (Z) = 0 × 0.25 + 1 × 0.5 + 2 × 0.25 = 1. Then, E (Z ² ) = 0 ² × 0.25 + 1 ² × 0.5 + 2 ² × 0.25 = 1.5. Thus, σ, the standard deviation of the values encoded by the current genotypes,

. As a result, when each value is substituted into Equation 1, the value of (Z ₁ , Z ₂ , Z ₃ ) = (0, 1, 2) in which the SNP is encoded is

Is standardized.

The standardization unit 120 normalizes values encoded by genotypes of SNPs determined using Equation 1 as described above. Those skilled in the art may understand that the standardization unit 120 may standardize values encoded by genotypes using a standardization method other than the standardization method as described above.

4 is a table illustrating a result of normalizing values encoded by genotypes in the standardization unit 120 according to an embodiment of the present invention. Referring to FIG. 4, values encoded by genotypes in the table of FIG. 3 are normalized by Equation 1 in the standardization unit 120. As a result, values encoded under different conditions according to genetic models are normalized to the same reference and displayed as values for the same reference.

Referring back to FIG. 1, the model generator 130 generates a prognostic prediction model indicating a correlation between genotypes of SNPs determined by prognostics of the subjects by analyzing the standardized results with a statistical prediction algorithm. Here, the prediction algorithm includes at least one of a classification algorithm, a machine learning algorithm, and a regression analysis algorithm.

Classification algorithms include a G-LASSO algorithm (Gradient lasso algorithm), LASSO algorithm, etc. Machine learning algorithms (Machine Learning Algorithm), such as a support vector machine (SVM). That is, one of ordinary skill in the art may understand that the model generator 130 of the present embodiment may use any algorithm that can generate a predictive model statistically or mathematically from the standardized results.

The model generator 130 statistically or mathematically analyzes the standardized values using a prediction algorithm. The model generator 130 generates a prognostic prediction model expressed as a kind of equation, as shown in Equation 2 below.

In the prognostic prediction model such as Equation 2, SNP 1, SNP 2, SNP 3, (omitted), SNP 99,876, etc. correspond to the determined SNPs, and α _i correlates with the prognosis for a specific condition for each of the determined SNPs. Corresponds to the coefficients representing the relationship.

However, since the prediction algorithm used for generating the prognostic model in the model generator 130 is not limited to any one, various types of prognostic models different from those of Equation 2 may be generated. Those of ordinary skill in the art can understand.

The prognostic model generated in this way is not generated using only one genetic model, but is generated by reflecting all the results using several genetic models. Thus, the prediction model is generated using only one genetic model. The accuracy is much higher.

Referring back to FIG. 1, the prediction unit 140 predicts the prognosis for the subject by using the generated prognostic prediction model. That is, the prognosis for the subject is predicted by using the prognostic prediction model shown in Equation 2. In more detail, if a subject wants to predict the prognosis of a particular condition, the genetic analysis device (not shown) is the genotype for the SNPs used in the prognostic model, such as Equation 2, among the SNPs of the subject. We grasp from city). In other words, SNP 1, SNP 2, SNP 4, (omitted), and SNP 99,876 were used in the prognostic model, such as Equation 2, so that the genotypes of SNP 1, SNP 2, SNP 4, (omitted), and SNP 99,876 of the subject were used. Figure out. Thereafter, the predictor 140 substitutes standardized values corresponding to each genotype of the examinee into a prognostic model.

5A and 5B are tables showing the results performed in the prognostic model generating apparatus 1 according to an embodiment of the present invention. 5A and 5B are results performed for 1,000 SNPs of 250 persons, and assume three prognostic SNPs showing genetic characteristics of dominant, recessive and additive respectively. And, resampling was performed 50 times.

The predictor 140 predicts the prognosis by comparing the threshold value with a predetermined threshold as a result of substituting the standardized values corresponding to each genotype of the subject into the generated prognostic model. The result calculated by substituting into the prognostic prediction model shown in Equation 2 represents Y in Equation 2. In other words, when the value calculated by substituting a prognostic model such as Equation 2 (Y in Equation 2) exceeds a certain threshold, the prognosis is predicted by a certain condition, for example, that the cancer is more likely to develop, The prognosis is predicted to be less likely to develop cancer if the threshold is not exceeded.

The storage unit 30 stores genotype data received by the data receiver 20, data processed by the processor 10, for example, the generated model 502, and the like.

The output unit 40 outputs the result of predicting the prognosis of the subject in the predictor 140 to the subject. The output unit 40 may be a device for displaying visual information (eg, a display, an LCD screen, an LED, a scale display device, etc.) for reporting information to a user, or a device for displaying auditory information (for example, Speaker, etc.).

6 is a flowchart of a method of generating a prognostic model according to an embodiment of the present invention. Referring to FIG. 6, the method for generating a prognostic model according to the present embodiment includes steps that are processed in time series in the apparatus 1 for predicting predictive model shown in FIG. 1. Therefore, even if omitted below, the above descriptions of the prognostic model generation apparatus 1 shown in FIG. 1 also apply to the method of generating the prognostic model according to the present embodiment.

In step 601, the SNP determiner 1102 tests each of the SNPs obtained from the subjects with a plurality of genetic models that statistically test the genetic characteristics of the SNP, thereby significantly classifying the subjects among the SNPs into a patient group and a control group. Determine SNPs.

In operation 602, the genetic model determiner 1103 determines the most significant genetic model among the genetic models used in the assay for each of the determined SNPs.

In step 603, the standardization unit 120 normalizes values encoded by genotypes of each of the SNPs determined based on the determined genetic model.

In operation 604, the model generator 130 generates a prognostic prediction model indicating a correlation between the genotypes of the SNPs determined by the statistical prediction algorithm and the prognosis of the test subjects.

Meanwhile, the above-described embodiments of the present invention can be written as a program that can be executed in a computer, and can be implemented in a general-purpose digital computer that operates the program using a computer-readable recording medium. In addition, the structure of the data used in the above-described embodiment of the present invention can be recorded on the computer-readable recording medium through various means. The computer-readable recording medium includes a storage medium such as a magnetic storage medium (e.g., ROM, floppy disk, hard disk, etc.), optical reading medium (e.g., CD ROM,

So far I looked at the center of the preferred embodiment for the present invention. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. Therefore, the disclosed embodiments should be considered in an illustrative rather than a restrictive sense. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the scope will be construed as being included in the present invention.

1: prognostic model generation device 10: processor
20: data receiving unit 30: storage unit
40: output unit 110: SNP analysis unit
120: standardization unit 130: model generation unit
140: prediction unit 1101: test unit
1102: SNP determination unit 1103: genetic model determination unit

Claims (16)

Testing each of the SNPs obtained from the subjects with a plurality of genetic models statistically testing the genetic characteristics of the SNPs to determine SNPs that significantly classify the subjects into a patient group and a control group;
Determining the most significant genetic model of the genetic models used in the assay for each of the determined SNPs;
Normalizing encoded values of genotypes of each of the determined SNPs based on the determined genetic model; And
Generating a prognostic prediction model indicating a correlation between the genotypes of the determined SNPs and the prognosis of the subjects by analyzing the standardized results by a statistical prediction algorithm. Way. The method of claim 1,
The normalizing may include normalizing values encoded by different conditions for each of the SNPs to values having the same reference for each of the determined SNPs based on the determined types of genetic models. The method of claim 1,
The determined genetic models include at least one of a dominant model, a recessive model, and an additive model,
The prognostic model is generated by reflecting the results using the determined genetic models. The method of claim 1,
Before each determining the SNPs, further testing the significance for each of the SNPs using each of the genetic models,
Determining the SNPs and determining the genetic model are based on an assay result of the significances. The method of claim 4, wherein
Wherein said assaying, determining said SNPs, and determining said genetic model use a Max Test method. The method of claim 1,
Predicting a prognosis for the subject using the generated prognostic model. The method of claim 1,
Wherein the prediction algorithm comprises at least one of a classification algorithm, a machine learning algorithm and a regression analysis algorithm. The method of claim 1,
The method of generating a prognostic model is performed by a computing device including a processor. A computer-readable recording medium storing a program for causing a computer to execute the method according to any one of claims 1 to 7. An SNP determination unit that determines each of the SNPs obtained from the subjects by a plurality of genetic models statistically testing the genetic characteristics of the SNP, and determines the SNPs of which the subjects are significantly classified into a patient group and a control group;
A genetic model determiner for determining the most significant genetic model of the genetic models used in the assay for each of the determined SNPs;
A normalizer for normalizing encoded values of genotypes of each of the determined SNPs based on the determined genetic model; And
And a model generator for generating a prognostic model that represents a correlation between the genotypes of the determined SNPs and the prognosis of the subjects by analyzing the standardized results by a statistical prediction algorithm. Generating device. The method of claim 10,
And the normalization unit normalizes values encoded by different conditions for each of the SNPs to values having the same reference for each of the determined SNPs based on the determined types of genetic models. The method of claim 10,
The determined genetic models include at least one of a dominant model, a recessive model, and an additive model,
The prognostic model is generated by reflecting the results of using the determined genetic models. The method of claim 10,
And further including an assay for respectively testing the significances of the SNPs using each of the genetic models before determining the SNPs.
And the SNP determiner and the genetic model determiner each determine based on a test result of the significances. The method of claim 13,
And the assay unit, the SNP determination unit, and the genetic model determination unit use a Max Test method. The method of claim 10,
And a predictor for predicting prognosis for the subject using the generated prognostic model. The method of claim 10,
And the prediction algorithm comprises at least one of a classification algorithm, a machine learning algorithm, and a regression analysis algorithm.

Images