Method for predicting phenotype from genotype using categorical distance defined in genome

Abstract

A method for predicting phenotype from genotype according to the present invention captures a probability distribution difference of genotypes of a specific phenotype group and a comparison (control) group at each position of single nucleotide polymorphism (SNPs) as a concept of a mean difference of categorical distances, finds the positions of single nucleotide polymorphism (SNPs) related to phenotypes having differences, accumulates the differences as a concept of total aggregate distance, and accumulates differences at multiple positions even if the differences at individual positions are too small to distinguish, thereby enabling prediction of phenotypes from genotypes. That is, by determining a target reference value of the total aggregate distance for selected positions and comparing a person's total aggregate distance with the target reference value, it is possible to determine whether the person has a corresponding phenotype such as a disease, a drug response, and physical characteristics.

Description

A method of predicting phenotype from genotype using categorical distance defined on genome

[0001] The present invention relates to a technique for classifying people who share phenotypes such as drug reactions, diseases, and physical characteristics, and predicting phenotypes from genotypes by utilizing the difference in probability distribution of genotypes found at the positions of single nucleotide polymorphisms (SNPs) between a specific phenotype group and its comparison (control) group.

[0002]

[0003] Methods have primarily focused on identifying genes associated with specific phenotypes, such as drug responses and diseases, from genotypes. In particular, genome-wide association studies (GWAS) have explored the search for genes or single nucleotide polymorphisms (SNPs) associated with phenotypes across the entire genome. However, some phenotypes involve multiple genes or SNPs, and existing technologies have limitations in combining and analyzing these factors.

[0004]

[0005] The present invention proposes a method for determining whether a person belongs to a group by defining a distance function (categorical distance) at the positions of single nucleotide polymorphisms (SNPs) between a person and a specific phenotypic group, i.e., people who share certain genetic characteristics, and the group.

[0006] To this end, we propose a method to determine whether an individual belongs to a specific phenotypic group by comparing the probability distribution of genotypes at single nucleotide polymorphism (SNP) positions for a specific phenotypic group with the probability distribution of the comparison (control) group, and combining and accumulating the differences at multiple single nucleotide polymorphism (SNP) positions.

[0007]

[0008] In order to achieve the above object, the method for predicting a phenotype from a genotype according to the present invention is characterized by including the steps of accessing DNA sequencing data of people included in a specific phenotype group X and a comparison (control) group Y, the step of calculating a probability distribution of genotypes of the groups X and Y at each position (SNP position) of single nucleotide polymorphisms (SNPs), the step of selecting one or more positions of single nucleotide polymorphisms (SNPs) at which an 'average difference in categorical distance' between the groups X and Y occurs, the step of determining a 'target reference value' of a 'total cluster distance' for the positions (SNP positions), and the step of comparing the total cluster distance of the person for the positions (SNP positions) with the target reference value using DNA sequencing data of the person requiring judgment to determine whether the person is included in the group X.

[0009]

[0010] The present invention allows for determining which individuals belong to a specific phenotypic group, even when the associated genes are unclear and individual single nucleotide polymorphism (SNP) positions differ only slightly from the comparison group. This allows for combining and accumulating differences across multiple positions, even for groups of individuals sharing a specific genetic trait (i.e., a specific phenotypic group). This means that even when the genes associated with a drug's effect are unclear, it is possible to classify individuals who are likely to respond to the drug, demonstrating its wide-ranging utility.

[0011] In particular, it provides a method that can effectively combine and analyze tens, hundreds, or thousands of single nucleotide polymorphism (SNP) positions, and predict specific phenotypic groups from genotypes by accumulating small differences at individual positions.

[0012]

[0013] FIG. 1 is a diagram showing a flowchart of a method for predicting a phenotype from a genotype according to an embodiment of the present invention.

[0014] FIG. 2 is a diagram showing the classification of East Asians and other ethnic groups using the first chromosome according to an embodiment of the present invention. (For people included in the two groups, the horizontal axis represents the total distance, and the vertical axis represents the number of people (converted to probability). The light color on the left represents East Asians, and the dark color on the right represents other ethnic groups. This was done using Python plt.hist(EAS / OTHER, bins=50, density=False, alpha=0.7, histtype='bar').

[0015] FIG. 3 is a diagram showing the classification of East Asian groups and other ethnic groups using chromosomes 1 to 22 according to an embodiment of the present invention. (The dark color on the left indicates East Asians, and the light color on the right indicates other ethnic groups. This was done using Python plt.hist(EAS / OTHER, bins=50, density=False, alpha=0.7, histtype='bar').

[0016] FIG. 4 is a diagram showing the classification of male and female groups using the total distance for positions on the upper body where the absolute value of the average difference in categorical distance is greater than 0.6 according to an embodiment of the present invention. (For people included in the two groups, the total distance is on the horizontal axis, and the number of people (converted to probability) is on the vertical axis, with the dark color on the left representing men and the light color on the right representing women. This was done using Python plt.hist(men / women, bins=50, density=False, alpha=0.7, histtype='bar').

[0017] FIG. 5 is a diagram showing the classification of male and female groups using the total distance for positions on the chromatic body where the absolute value of the average difference in categorical distance is greater than 0.07 and less than 0.21, according to an embodiment of the present invention. (The vertical axis is the number of people (converted to probability), and the dark color on the left is male and the light color on the right is female. This was done using Python plt.hist(men / women, bins=50, density=False, alpha=0.7, histtype='bar').

[0018]

[0019] Throughout this specification, when a part is referred to as "including" a component, this does not exclude other components, but rather implies the inclusion of additional components, unless otherwise specifically stated. Embodiments of the present invention are described in detail below with reference to the accompanying drawings. Furthermore, the contents of KR 10-2024-0009189, the subject matter of which claims priority to this application, are incorporated herein by reference.

[0020] Single nucleotide polymorphisms (SNPs) are highly correlated with phenotypes, and genome-wide association studies (GWAS) utilize genetic variation in SNPs to identify these associations. Individuals who share a specific phenotype, such as susceptibility to a particular disease, drug response, or physical trait (e.g., race), are expected to have different genotype structures than individuals in a comparison group who do not share the phenotype.

[0021] Zhanshan Ma et al. (Defining Individual-Level Genetic Diversity and Similarity Profiles, Scientific Reports (2020)) define entropy in DNA sequences and attempt to identify structural differences using it. However, there is a limitation that a uniform concept such as entropy has limited utility when applied to various phenotypes. On the other hand, it can be assumed that there will be collective structural differences in genotypes when comparing a specific phenotypic group with its comparison group. The problem here is how to capture collective structural differences in genotypes, and the following assumptions can be made.

[0022]

[0023] There is a difference in the probability distribution of genotypes (or bases) at one or more SNP positions between a particular phenotypic group and its comparison (control) group.

[0024] This becomes clearer when the phenotype is determined by the allele of a single gene (Mendelian trait). When Mendel's laws of inheritance are applied, if R is a dominant allele and r is a recessive allele, and the dominant group includes people with genotypes RR and Rr, and if the probabilities of such people within the dominant group are a and 1-a, respectively, then the probability distribution of the genotypes of the dominant group is (RR, Rr, rr) = (a, 1-a, 0). The recessive group only has people with genotype rr, so its probability distribution is (RR, Rr, rr) = (0, 0, 1), which is different from the probability distribution of the dominant group.

[0025] Now, let's explore how to combine the differences in probability distributions across multiple SNP locations using a concept called "categorical distance." Categorical distance uses the probability distribution of genotypes at SNP locations in a specific phenotypic group to measure the distance between an individual and that phenotypic group, thereby determining whether that individual belongs to that phenotypic group.

[0026]

[0027] Category of genotype (or base): A category of genotype refers to a category of possible genotypes at a specific SNP location, such as reference, first variant, and second variant, or in the case of autosomes, a category that collects pairs of possible variants, such as (reference, reference), (reference, first variant), (first variant, reference), (first variant, first variant). The reference and variant can be a base, multiple bases, or indels.

[0028] According to the GRCh38 version of the 1000 Genomes Project (https: / www.internationalgenome.org) data, at position 18212123 of chromosome 13, the reference is A and the first variant is G. At this time, there are three categories of people's genotypes: 0|0, 0|1(=1|0), and 1|1, and these are called category 1, category 2, and category 3. (However, 0|0 = AA, 0|1 = AG, 1|1 = GG) If there were AC, C, and CT, the reference would be AC, the first variant C, and the second variant CT, which are the three cases, and the possible genotype categories would be 0|0, 0|1(=1|0), 1|1, 0|2(=2|0), 1|2(=1|2), 2|2, resulting in six categories.

[0029]

[0030] Definition of categorical distance: SNP locations Some people in The categorical distance for a set X of people is defined as follows:

[0031] ( ) = -

[0032] (However, the location For the categories of possible genotypes in , the probability distribution for each category is (probability of category 1, probability of category 2, ..., probability of category n) = ( , , ... , ) and, = max( , , ..., ), person location of The genotype in is the kth category. For reference, if there are N people in the set X, the number of people corresponding to the first category is ×N people)

[0033] If, at position 100, (0|0, 0|1, 1|1) = (0.81, 0.18, 0.01), and the person If the genotype is 0|1, ( ) = - = max(0.81, 0.18, 0.01) - = 0.81 – 0.18 = 0.63. Categorical distance measures how common a person's genotype is within a group at a specific SNP location. It converts the "letter" genotype into a number, allowing the calculation of the group average.

[0034] Moreover, categorical distance is a more useful concept than frequency. For example, if A: (0|0, 0|1, 1|1) = (0, 1, 0), B: (0|0, 0|1, 1|1) = (0.5, 0, 0.5), then sets of people A and B have the same frequency of the first variant, 50%. However, there are many differences when looking at the categorical distance. In other words, categorical distance can distinguish two groups even when it is difficult to distinguish them based on frequency.

[0035]

[0036] Mean of categorical distance: location In , the mean of the categorical distances of people included in set X to set X ( ) is calculated = - ( + + ... + ) is this. And, the location In , the mean of the categorical distances of people included in set Y to set X ( ) is calculated = - ( + + ... + ) is. (However, the location The probability distribution of the genotypes of set X in is ( , , ..., ), the probability distribution of the genotypes of the set Y is ( , , ..., ) and, = max( , ,..., ) am.)

[0037] The above assumption that “there is a difference in the probability distribution of genotypes (or bases) at one or more SNP positions between a specific phenotypic group and its comparison group” is based on the assumption that there are certain SNP positions ( , ,..., ) ≠ ( , ,..., ) means.

[0038] The average categorical distance allows us to identify SNP locations where the probability distribution differs between a specific phenotypic group and its comparison group. If a certain location at ≠ On the other hand, ( , ,..., ) ≠ ( , ,..., ) is. Therefore, the 'average difference of categorical distances ( )', ≡ - By finding the locations where ≠ 0, we can find the locations where the genotypic probability distributions of the two groups differ. | | The larger the value, the greater the difference in probability distribution.

[0039] In general ≤ as, > can be. For example, if for X (0|0, 0|1, 1|1)=(0.81, 0.18, 0.01) and for Y (0|0, 0|1, 1|1)=(1, 0, 0), = - ( + + ) = 0.81 - 0.6886 = 0.1214, = - ( + + ) = 0.81 - 0.81 = 0, so > am.

[0040]

[0041] Collective distance: Modify the categorical distance to obtain the collective distance as follows: Defines .

[0042] location at, = - If ≤ 0 = is defined as,

[0043] > If 0 = 1 - is defined as

[0044] (step, is the categorical distance, i.e. ( ) = - , = max( , , ..., ), ( , , ..., ) is the location Probability distribution of genotypes in humans location of The genotype in the k category)

[0045] The arithmetic mean of the set distances of sets X and Y for each set X , When said, all positions About ≤ , and the 'absolute value of the mean difference of categorical distances' is | | = - It becomes.

[0046] The aggregation distance makes the mean of the aggregation distance of phenotypic group X smaller than the mean of the comparison group Y at all locations. This transformation is important because adding aggregation distances at multiple locations has the effect of amplifying small differences in means at a single location. This becomes clearer when we understand the distribution of aggregation distances statistically. If we create a new random variable by adding 100 independent random variables that follow a normal distribution, its mean will be 100 times larger and its standard deviation will be 10 times larger. The aggregation distance of a set X at a single SNP location is normally distributed. , and the set Y is Let's say that the mean of set X is 0.05 and the standard deviation is 0.1, so it is not easy to decide which set a given value belongs to. However, there are 100 locations, not just one location, and the set distance of each location is all X. , Y is Assuming that they are independent while following , the sum of the set distances at these locations is respectively , will follow. (The mean is 100 times, the standard deviation is 10 times). In this case, the difference in the means is 5 and the standard deviation is 1, so it is easy to determine which group it belongs to when a certain value is given. In other words, even if it is difficult to distinguish sets X and Y with a single specific location, the members of the two groups can be distinguished by calculating the distance by adding the set distances at multiple locations. In relation to the selection of locations, all locations where the average of the set distances differ can be selected. However, it can be more useful to select important locations where the difference in the means is large. This is because the parts that are virtually noisy can be ignored.

[0047] Phenotype prediction: = { , ,..., } Let be the selected SNPs positions. Total collective distance cast = + +...+ is defined as. If, some threshold Once the value is determined, for some person z ( ) ≤ If so, we can determine that z belongs to a specific phenotypic group X. The threshold can also be determined using well-known metrics such as precision, recall, and accuracy.

[0048] Precision is the ratio of the actual True values ​​classified by the model as True (= TP / (TP+FP)), recall is the ratio of the actual True values ​​predicted by the model as True (= TP / (TP+FN), and accuracy is the ratio of the model's correct predictions (=(TP+TN) / (TP+FN+FP+TN)). (Note that TP: predicts True values ​​when they are actually True, FP: predicts True values ​​when they are actually False, FN: predicts False values ​​when they are actually True, and TN: predicts False values ​​when they are actually False.)

[0049] When the mean and standard deviation of the total distance of set X are M and A, respectively, and set Y is m and a, then they are normally distributed. , One way is to treat it as a point and select the point where the two curves meet (with maximum accuracy).

[0050]

[0051] A processor can perform computations, steps in a programmed method.

[0052] It should be understood that the processor may include a computing device, which may be implemented to perform the method steps described herein. The processor may be coupled to a communication interface operable to be accessed remotely (e.g., via an intranet, the Internet, etc.). The processor may further include human input-output signals to enable a human input device (e.g., a mouse, a trackpad, a keyboard, etc.) to issue commands to the processor or to provide human readable results to a human output device (e.g., a screen, a terminal, a monitor, a printer, etc.). The processor may access data storage for reading / writing data, and optionally for retrieving computer instructions that, when read by the processor, cause the processor to perform the method steps described herein. The data storage may be a storage device that is locally accessible to the processor (e.g., an internal drive, a register, a memory chip, etc.) or a storage device that is remotely accessible (e.g., an external drive, a network storage device, an intranet / "cloud" storage, etc.). That is, the processor can access programming instructions and data to perform specific method steps described herein, and output the results to an output device to produce human-readable output in the storage device.

[0053]

[0054] Now, the composition of the present invention will be described.

[0055] The present invention may include the steps of accessing DNA sequencing data of people included in a specific phenotypic group X and a comparative (control) group Y that share genetic characteristics such as susceptibility to a particular disease, drug response, and physical characteristics; calculating a probability distribution of genotypes of the group X and the group Y at each position (SNP position) of single nucleotide polymorphisms (SNPs); and selecting positions of one or more single nucleotide polymorphisms (SNPs) at which an 'average difference in categorical distances' of the group X and the group Y occurs.

[0056] In addition, the present invention may further include a step of determining a 'targeted threshold' of the 'total cluster distance' for the above locations (SNP locations); and a step of comparing the total cluster distance of the person for the locations (SNP locations) with the target threshold using DNA sequencing data of the person requiring judgment, thereby determining whether the person is included in the group X. The embodiment of Fig. 1 is a diagram showing a flowchart of the above steps. (DNA data access step, probability distribution calculation step, SNP location selection step, target threshold determination step, judgment step)

[0057]

[0058] The steps for accessing DNA sequencing data of people belonging to a specific phenotypic group X and its comparison group Y are described in more detail as follows.

[0059] A specific phenotypic group X refers to people who share a genetic characteristic, and the comparison group (control group) refers to a group of people who do not have that phenotype. For example, a group of people who respond to a specific anticancer drug could be called X, and a group of people who do not respond to the drug could be called Y. Furthermore, East Asians could be called group X, and people of other races could be called comparison group Y.

[0060] DNA sequencing data from the 1000 Genomes Project

[0061] (https: / / www.internationalgenome.org) may include the meaning of a database that stores sequencing results in various file formats. Furthermore, it may refer not only to the entire genome, but also to a portion of the entire genome, such as chromosome 1 or a specific gene-related portion. Accessing may include making it available for use.

[0062] In addition, the step of accessing 'DNA sequencing data' may include a concept that includes a DNA sequencing step of extracting human genome data from blood, hair, etc. At this time, DNA sequencing may use next-generation sequencing (NGS) technology. The data generated in this way may be created as a database, such as the 1000 Genomes Project. The data may be in various formats, including VCF (Variant Call Format). In addition, the step of accessing 'DNA sequencing data' may be performed by a processor (or computer).

[0063]

[0064] The steps for calculating the probability distribution of the genotypes of the group X and the group Y at each position (SNP position) of the single nucleotide polymorphism (SNP) are described in more detail as follows.

[0065] "Each position of a single nucleotide polymorphism (SNP)" refers to a chromosome number and a numerical position (POS) within that chromosome, using data from the 1000 Genomes Project. SNP positions can be assigned IDs such as rs6054257.

[0066] As explained above, the 'probability distribution of genotypes' is the probability distribution of each category of possible SNP genotypes at a specific SNP location for a set X of people, i.e., (probability of the first category, probability of the second category, ..., probability of the n-th category) = ( , , ... , ) is said. (This probability can be obtained by counting the number of people included in each category. If there are N people included in the set, the number of people corresponding to the first category is The probability distribution of genotypes for set Y is (probability of the first category, probability of the second category, ..., probability of the n-th category) = ( , ,..., ) am.

[0067] The above genotype categories can be combined into a single category, or certain categories can be excluded, as needed. These categories are designed to elicit differences in the probability distribution of genotypes at that location, and can be adjusted and modified to suit the purpose.

[0068] A step of calculating a probability distribution of genotypes of the group X and the group Y at each position (SNP position) of single nucleotide polymorphisms (SNPs) may also be performed by a processor (or computer).

[0069]

[0070] The step of selecting the position of one or more single nucleotide polymorphisms (SNPs) at which the 'average difference in categorical distance' between the above groups X and Y occurs is described in more detail as follows.

[0071] In the above 'each position of single nucleotide polymorphisms (SNPs)', when the 'probability distribution of genotypes' is mentioned, the average of the categorical distances for the set X refers to the average value of the categorical distances for the set X of people included in a certain set. As above, the position In , the mean of the categorical distances of people included in set X to set X ( ) is calculated = - ( + + ... + ) , the mean of the categorical distances for the set X of people included in the set Y ( ) is calculated = - ( + + ... + ) (step, =max( , ,..., )) is. At this time, the 'average difference of categorical distances ( )'Is ≡ - is defined as

[0072] Since the probability distribution of the genotypes of people included in the comparison set Y is different from that of the set X, there are positions where the mean value of the categorical distance is different. 'Mean difference of the categorical distance ( ) can be positive or negative, and the larger the absolute value, the greater the difference in probability distribution. In this case, the absolute value of the mean difference of the categorical distance | | It is advisable to choose this large location.

[0073] In one way, | | You can select the locations in this order of largest to smallest. You can select as many as you need. Alternatively, | |You can select a location that is greater than a certain value. For example, | | > You can also select positions where 0.1 is the maximum.

[0074] A step of selecting the positions of one or more single nucleotide polymorphisms (SNPs) at which the 'average difference in categorical distances' between the group X and the group Y occurs may also be performed by the processor (or computer).

[0075]

[0076] The steps for determining the 'targeted threshold' of the 'total distance' for the above locations (SNP locations) are described in more detail as follows.

[0077] ={ , ,..., } are the SNP positions selected in the above step. At each position, the 'average difference in categorical distance' is = - When you say,

[0078] collective distance is defined as follows:

[0079] location at, If ≤ 0 = is defined as,

[0080] > If 0 = 1 - is defined as (however, is categorical distance)

[0081] Selected locations (SNPs locations) 'Total collective distance' for cast = + +...+ is defined as

[0082] A "target threshold" can be established as a criterion for classifying two sets. Here, the "target threshold" can be a threshold for the total set distance, determined by considering the target value among achievable indicators such as precision, recall, and accuracy. Predicting phenotypes such as fetal developmental disorders can take into account their specificity, such as minimizing the probability of non-developmental disorders being diagnosed as developmental disorders.

[0083] A step of determining a 'targeted threshold' of the 'total distance' for the above locations (SNP locations) may also be performed by the processor (or computer).

[0084]

[0085] The step of determining whether the person is included in the group X by comparing the total distance of the person for the positions (SNP positions) with the target reference value using the DNA sequencing data of the person requiring judgment is described in more detail as follows.

[0086] A person who requires judgment When there is, his DNA sequencing data can be accessed and used. Accessing and using DNA sequencing data may include the concept of DNA sequencing step of extracting the person's genetic data from blood, hair, etc. In this case, DNA sequencing may use next-generation sequencing (NGS) technology. In addition, not the entire data, but the above locations ( ={ , ,..., }) is desirable to use data for.

[0087] And, the target standard value determined in the above step When,

[0088] That person For the above locations (as calculated in the above step) )for The total distance of ( ) after calculating, ( )≤ On the other hand, is a person belonging to a specific phenotypic group X, ( ) > On the other hand, can be judged as a person included in comparison group Y.

[0089] A step may be performed to determine whether the person is included in the group X by comparing the total cluster distance of the person for the positions (SNP positions) with the target reference value using DNA sequencing data of the person requiring judgment by a processor (or computer).

[0090]

[0091] The invention according to the present invention may be performed from data including positions of one or more single nucleotide polymorphisms (SNPs), a formula for calculating 'aggregate distances' for the positions (SNP positions), a 'target reference value' for the aggregate distances, and DNA sequencing data of a person requiring a determination. That is, the method may be performed by a step of accessing data of positions of one or more single nucleotide polymorphisms (SNPs) at which an 'average difference in categorical distances' occurs between people included in a specific phenotype group X and a comparison group Y thereof, a formula for calculating 'aggregate distances' for the positions (SNP positions), and a 'target reference value' for the aggregate distances; and a step of comparing, from the DNA sequencing data of the person requiring a determination, the aggregate distances of the person for the positions (SNP positions) with the target reference value to determine whether the person is included in the group X. The above steps may be performed by a processor.

[0092]

[0093] As one embodiment of the present invention, a method of predicting a phenotype from a genotype is proposed, wherein East Asians are regarded as a specific phenotypic group X, and the remaining races are regarded as a comparison group Y.

[0094] We use the GRCh38 version of the 1000 Genomes Project (https: / www.internationalgenome.org) data. You can download the data or use data that has already been downloaded. We use single nucleotide polymorphisms (SNPs), which are known to be the main cause of mutations in human DNA sequences. The data includes 515 East Asians (EAS) and 2,032 people from other races (AFR, EUR, AMR, SAS) (excluding one person who falls into two race categories). A specific phenotypic group X is East Asians, and the remaining races are the comparison group Y. We use data from chromosome 1.

[0095] Calculate the probability distribution of genotypes of group X and group Y at all SNP locations on chromosome 1, and calculate the absolute value of the 'average difference of categorical distances' (| |) is greater than 0.6, i.e. | | > Select 367 positions where 0.6 is present. If we list the 10 positions (POS corresponding number, absolute value of mean difference of categorical distance) with the largest absolute value of 'mean difference of categorical distance', ('155053638', 0.7274306004484381), ('16064561', 0.726624894329061), ('1438102', 0.7042619251371032), ('16060971', 0.6993427286147627), ('16060974', 0.6965705781675491), ('62992403', 0.6932967867128118), ('75786650', 0.6896632557559581), ('75792833', 0.6894950728953206), ('75795749', 0.6894950728953206), ('43218705', 0.6825462706972166). Selecting positions where the absolute value of the mean difference of the categorical distance exceeds 0.6 and selecting 367 positions in descending order of the absolute value of the mean difference of the categorical distance are the same. You can select fewer or more positions than 367. These selected positions are useful in themselves, such as finding related genes or finding genes associated with related SNPs. Additionally, there are cases where the absolute values ​​of the mean differences in categorical distances at different locations are the same ('75792833' and '75795749' above), which allows us to infer that these locations are combined and interact with certain genes.

[0096] Now, we calculate the total set distance for the 367 locations of people included in set X and people included in set Y. The example of Fig. 2 shows the distribution of frequencies according to the total set distance for East Asians (set X) and people of other races (set Y), with the total set distance on the horizontal axis and the number (converted to probability) of how many people are included in the range divided into 50 on the vertical axis. (The light color on the left is East Asian, and the dark color on the right is other races.) In order to achieve 99% accuracy, if the target standard value is 100, 510 East Asians are below the standard value, and 20 people of other races are below the standard value. At this time, the accuracy is 99%, the precision is 99%, and the recall is 96.2%.

[0097] When a person requires identification, their DNA sequencing data is accessed. Accessing data at the 367 locations mentioned above is particularly efficient, especially when DNA data is obtained directly from blood or other sources. If the total distance across the 367 locations is 100 or less, the person is identified as East Asian. If it exceeds 100, the person is identified as belonging to another race. An accuracy of approximately 99% can be expected.

[0098]

[0099] There are a total of 4,888 SNP positions on chromosomes 1 to 22 where the absolute value of the average difference in categorical distance exceeds 0.6. When the target standard value of the total distance is set to 1000, the accuracy is 100%, as shown in the example of Figure 3. This shows that East Asians can be defined by the function (total distance) defined at these positions.

[0100] That is, East Asians can be defined as people whose total clustering distance at the above 4,888 SNP locations is less than 1000. (It is an important achievement of the present invention to have revealed that East Asians can be defined as a function defined in the genome.)

[0101]

[0102] As another embodiment of the present invention, a method is proposed for predicting a phenotype from a genotype of an autosome (excluding a sex chromosome), with a male as a specific phenotypic group X and a female as a comparison group Y.

[0103] The GRCh38 version data of the 1000 Genomes Project (https: / www.internationalgenome.org) consists of 1,252 males and 1,296 females. After excluding 300 individuals from the front of the list for testing, 952 males (set X) and 996 females (set Y) were used, and the probability distribution of genotypes of group X and group Y at all SNP locations of the autosomes was calculated, and the absolute value of the 'average difference in categorical distance' (| |) is greater than 0.6, i.e. | | > 0.6, listing the 6 positions (chromosome number, POS corresponding number, absolute value of the mean difference of categorical distances), (No. 13, '18212123', 0.6444051752522412), (No. 17, '26867060', 0.9637934550148906), (No. 17, '26867146', 0.9250370594729741), (No. 17, '26866764', 0.8976834266662015), (No. 17, '26866745', 0.822476758244049), (No. 17, '26867019', 0.6064387484263727), and the steps to access the data for these 6 locations and the calculation formula (method) for the total distance for these locations and the target reference value set to 3 can be performed.

[0104] From the DNA sequencing data of the six positions of the person requiring judgment, if the total aggregation distance for the six positions is less than or equal to the target reference value of 3, the person is judged as a man (group X), and if it exceeds 3, the person is judged as a woman (group Y). The embodiment of Fig. 4 shows the distribution of frequencies according to the total aggregation distance for 300 men and women each for testing who were excluded above, and the number (converted to probability) of how many people are included in the range where the total aggregation distance is divided into 50 on the horizontal axis is shown on the vertical axis. (The dark color on the left is a man, and the light color on the right is a woman.) 299 men have 3 or less, and 0 women have 3 or less. At this time, the accuracy is 99.8%.

[0105]

[0106] To understand the effect of the present invention in more detail, the absolute value of the 'average difference of categorical distances' (| |) Let's try to analyze by selecting relatively smaller locations.

[0107] 0.07 < | | < 0.21 There are 51 positions in total, but it is not easy to distinguish between men and women at individual positions. However, if the total set distance for the 51 positions is calculated and the target reference value is set to 30.5, 285 men and 13 women are below the target reference value, and the accuracy is 95.3%, and the two groups can be easily distinguished, as shown in the example of Fig. 5. This case well demonstrates the effect of the present invention that phenotypes can be distinguished by collecting small differences in genotypes even if there is no position where the probability distribution difference between the phenotype set and its control group is large. This analysis was conducted to argue that it can also be useful in distinguishing other phenotypes (other than men and women) where the difference at individual SNP positions is not large. The 51 positions are as follows. Chromosome 1: 5 [('242917670', 0.20716295908866728), ('242886341', 0.11797898642093263), ('242887048', 0.11397115543595182), ('242911992', 0.10097135948638841), ('202378896', 0.07369052333903087)] Chromosome 2: 9

[0108] [('88773387', 0.12181343125485489), ('88773398', 0.12181343125485489),

[0109] ('95938800', 0.11903290728055926), ('87883796', 0.1132570941204264),

[0110] ('97168870', 0.10380622837370235), ('87905256', 0.10024715059659584),

[0111] ('87905255', 0.1000480277156931), ('87882867', 0.07631820446108994),

[0112] ('87870325', 0.07146208057992864)] Chromosome 3: 4 [('16438051', 0.11841176052277752), ('16544144', 0.09581148192799444),

[0113] ('16373342', 0.09115776005653853), ('16551924', 0.08878082593774572)]

[0114] 1 copy of chromosome 5 [('10346475', 0.09500626083176)] 1 copy of chromosome 9

[0115] [('31120122', 0.11553006213116007)] Chromosome 10; 4 [('38788107', 0.11762254512010323), ('38879848', 0.08158431693017393), ('38879852', 0.08158431693017393), ('38879885', 0.07604762315982383)] Chromosome 13; 4

[0116] [('19389753', 0.1391035950170288), ('18211952', 0.08587694468290497),

[0117] ('18212061', 0.07842812000774801), ('19394927', 0.07470229409978024)]

[0118] Chromosome 16: 1 [('33741748', 0.07981159331202359)]

[0119] Chromosome 17: 17 [('21863032', 0.12462705670503493), ('21863043', 0.12450347786173299), ('26866572', 0.12274027257962017), ('21881265', 0.1212904279358803), ('26863513', 0.12078728550243634), ('26854337', 0.1129062539260921), ('26854338', 0.1129062539260921), ('26854346', 0.1129062539260921), ('26854349', 0.1129062539260921), ('26854355', 0.1129062539260921), ('21903582', 0.10617408728197164), ('21903588', 0.10617408728197164), ('21903589', 0.10617408728197164), ('21929921', 0.10408699269927668), ('21933986', 0.08478170680036723), ('21893345', 0.08072059440360124), ('21863754', 0.07334183673469397)]

[0120] Chromosome 21: 5 [('10735603', 0.09106878045335787), ('10735583', 0.08236309229574179), ('10735592', 0.08236309229574179), ('10735593', 0.08236309229574179), ('10735586', 0.08077269710151957)]

[0121]

[0122] The method according to the embodiment may be implemented in the form of various program commands and recorded on a computer-readable medium, and may be executed by an electronic medium such as a server. Examples of the computer-readable medium include a hard disk, a magnetic medium, a magneto-optical medium, and a ROM, a RAM, a flash memory, etc. In addition, the computer-readable medium may include program commands, data files, data structures, etc. The program commands may be specially configured for the embodiment or may be publicly available. Examples of the program commands may include not only machine language codes but also high-level language codes that can be executed using an interpreter, etc. The hardware device may be configured to operate as one or more software modules to perform the operations of the embodiment.

[0123]

[0124] While the present invention has been described in detail through preferred embodiments, those skilled in the art will appreciate that various modifications and variations are possible without departing from the technical spirit of the present invention. For example, appropriate results can still be achieved even if the embodiments are performed in a different order than described above, components are combined or combined in different forms, or components are substituted with other components or equivalents.

[0125] Accordingly, the scope of protection of the present invention should be determined by the interpretation of the claims described below, and the scope equivalent to or equivalent to the claims is also included in the scope of rights of the present invention.

[0126]

[0127] The present invention can classify people who are responsive to a drug, for example, people who are responsive to a specific anticancer drug, thereby doubling the usefulness or effectiveness of the drug.

[0128] The present invention can be used for diagnosis or disease prevention by classifying people who are likely to develop diseases related to their genome, such as people who are likely to develop dementia.

Claims

1. A step of accessing DNA sequencing data of people included in a specific phenotypic group X and its comparison (control) group Y; A step of calculating the probability distribution of genotypes of the group X and the group Y at each position (SNPs position) of single nucleotide polymorphisms (SNPs); A step of selecting the positions of one or more single nucleotide polymorphisms (SNPs) at which the 'average difference in categorical distances' between the above groups X and Y occurs; Method for finding locations of single nucleotide polymorphisms (SNPs) associated with a phenotype characterized by including 2. A step of accessing DNA sequencing data of people included in a specific phenotype group X and its comparison (control) group Y; A step of calculating the probability distribution of genotypes of the group X and the group Y at each position (SNPs position) of single nucleotide polymorphisms (SNPs); A step of selecting the positions of one or more single nucleotide polymorphisms (SNPs) at which the 'average difference in categorical distances' between the above groups X and Y occurs; A step of determining the ‘target reference value’ of the ‘total distance’ for the above locations (SNPs locations); A step of comparing the total cluster distance of the person for the positions (SNPs positions) with the target reference value using the DNA sequencing data of the person requiring judgment to determine whether the person is included in the group X; A method for predicting a phenotype from a genotype, characterized in that it comprises 3. A step in which a processor accesses data of positions of one or more single nucleotide polymorphisms (SNPs) at which an 'average difference in categorical distance' occurs between people included in a specific phenotype group X and a comparison (control) group Y, a calculation formula for a 'total distance' for the positions (SNP positions), and a 'target reference value' of the total distance; A step in which a processor compares the total cluster distance of the person for the positions (SNPs positions) from the DNA sequencing data of the person requiring judgment with the target reference value to determine whether the person is included in the group X; A method for predicting a phenotype from a genotype, characterized in that it comprises

Images