TWI518538B - Predicting hla genotypes using unphased and flanking single-nucleotide polymorphisms in han chinese population - Google Patents

Description

Method and device for predicting Han human leukocyte antigen genotype using single nucleotide polymorphism

The present invention relates to a single nucleotide polymorphism predictive human leukocyte antigen dual gene having ethnic specificity. In particular, the present invention relates to the use of a single nucleotide polymorphism of a Han Chinese to predict a human leukocyte antigen genotype.

The human leukocyte tissue antigen gene group is located on the sixth pair of chromosomes and is divided into major histocompatibility complex class I (HLA-A, HLA-B, and HLA-C) and major histocompatibility complex class II. The alleles of (HLA-DR, HLA-DQ, and HLA-DP), and the polymorphism of multiple dual genes of individual single genes, resulting in graft rejection during tissue or organ transplantation ( Graft rejection) and graft-versus-host diseases. The human leukocyte antigen dual gene also plays an important role in population genetics and immune-related disease status. Furthermore, previous comparative studies have shown that the immune system usually has a strong selective pressure, which may be caused by virus-host interactions. Because of these selective stresses, comparisons between ethnic groups revealed linkage disequilibrium and variable patterns of the dual gene distribution of human leukocyte antigen-to-gene.

The genetic variation of human leukocyte antigen (HLA) is associated with immune function, autoimmune diseases and certain cancers. To date, large-scale studies have been time-consuming and expensive to acquire human leukocyte antigen genes by experiments (via serology or PCR). Therefore, only low-cost single-nucleotide polymorphisms (SNPs) are needed to predict genotypes of leukocyte antigens to save money. Experimental time. However, most of the human leukocyte antigen genotype prediction models are only Caucasian samples, and few studies report samples containing non-Caucasians, and their human leukocyte tissue antigen gene classes are distributed differently among different races.

Zheng et al. in the 2011 BMC genetics journal emphasized that this model cannot be used for genotypes of white blood cell antigens of different races after constructing a predictive model for predicting white blood cell tissue antigens. Therefore, Ayele et al. in the PLOS ONE journal in 2011 have constructed a specific white blood cell tissue antigen prediction model for Africans; however, there is currently no Asian human leukocyte antigen prediction model. Therefore, it is necessary to construct a racially unique white blood cell tissue antigen prediction model, especially the white blood cell tissue antigen prediction model of Han people.

Accordingly, the present invention provides a method for predicting a human leukocyte antigen dual gene, the steps comprising: (a) providing a human nucleic acid sample; and (b) identifying a genotype of a single nucleotide polymorphic collection of the human nucleic acid sample. The single nucleotide polymorphic collection comprises various single nucleotide polymorphisms located on the human leukocyte antigen gene; (c) analyzing a single nucleotide polymorphism in step (b) using a predictive model a genotype to obtain a calculated value, wherein the predictive model uses a single nucleotide polytype genotype to predict a human leukocyte antigen dual gene; and (d) predicts the human based on the calculated value obtained in step (c) The human leukocyte antigen dual genotype of the sample; and wherein the sample is an Asian ethnic group, preferably a Han Chinese ethnic group. Thereby, by discriminating the single nucleotide polymorphism set found by the prediction model in the present invention, a small number of single nucleotide polymorphism positions can be detected at a low cost, and the human leukocyte antigen dual genotype can be accurately predicted.

In the prediction method of the present invention, each single nucleotide polymorphism contained in the mononucleotide polymorphic collection is derived from (1) HLA-A, (2) HLA-B, (3) HLA-C, (4) HLA-DPB1, (5) HLA-DQB1, and (6) HLA-DRB1 gene, wherein the line derived from the (1) HLA-A gene is selected from a first mononucleotide polytype set, a group consisting of a second single nucleotide polymorphic set, a third single nucleotide polymorphic set, and a fourth single nucleotide polymorphic set, wherein (i) the first single Nucleotide polymorphic collections include: rs1633085, rs2254071, rs407238, rs9258881, rs2975046, rs2735096, rs417162, rs9260954, rs6917477, Rs6457144, rs9261394, and rs2523990; (ii) the second single nucleotide polymorphic collection includes: rs4122198, rs16895757, rs1632973, rs9357086, rs11759549, rs3115628, rs3094165, rs2734925, rs2517755, rs2256919, rs11756025, rs7382061, rs6457144, Rs2517646, and rs7744914; (iii) the third single nucleotide polymorphic collection includes: rs3094165, rs9258883, rs3132714, rs1611493, rs2524005, rs2860580, rs12665039, rs6457109, rs3869062, rs3893464, rs5009448, rs2571375, rs7758512, and rs9261394 (iv) the fourth mononucleotide polymorphic collection includes: rs2523409, rs1611133, rs3115628, rs2517859, rs1611732, rs2523998, rs2860580, rs12202296, rs2248153, rs2975046, rs6457109, rs5009448, rs9260932, and rs6457144; 2) The single nucleotide polymorphism of the HLA-B gene is selected from a fifth mononucleotide polytype set, a sixth single nucleotide polytype set, and a seventh single nucleotide set. a group consisting of a type set, and an 8th mononucleotide polytype set, wherein (i) the 5th single nucleotide polytype set includes : rs3130944, rs3130532, rs3130534, rs3134762, rs16899207, rs2524089, rs9366778, rs2524166, rs9295984, rs4394275, rs9378249, rs2523534, rs9266406, rs2844558, rs5022119 rs3099848, rs4081552, rs2848716, rs2596454, and rs2248462; (ii) the sixth single nucleoside The acid polymorphic collection includes: rs11966319, rs2853948, rs6906846, rs9378228, rs2524051, rs9366778, rs16867947, rs4394274, rs4394275, rs2523591, rs9501572, rs7761068, rs2523535, rs9266406, rs5006724, rs13198903, rs9266669, rs9266689, rs3099849, rs2442749, rs1051796, Rs2596464, rs3099836, and rs3131622; (iii) the 7th single nucleotide polymorphic collection includes: rs9264868, rs9264942, rs3094691, rs2156875, rs2523619, rs2442719, rs2596501, rs2523589, rs2523554, rs2844573, rs9266395, rs9266440, rs9295986, Rs2442749, rs2596560, rs3128982, rs2284178, and rs7758090; (iv) the eighth single nucleotide polymorphic collection includes: rs3094691, rs7453967, rs4394274, rs4394275, rs2596509, rs2596501, rs1058026, rs25235 91, rs2523589, rs2523554, rs2523545, rs9501572, rs2844575, rs9266395, rs9266406, rs5006725, rs9295986, rs6933050, rs4959068, rs5022119, rs13198903, Rs9266689, rs2251396, rs1051796, rs3094584, rs9765960, and rs3128982; the single nucleotide polymorphism derived from the (3) HLA-C gene is selected from a ninth single nucleotide polymorphism set, a tenth single a group consisting of a nucleotide polymorphic collection, an 11th single nucleotide polymorphic collection, and a 12th single nucleotide polymorphic collection, wherein (i) the ninth single nucleotide The type of collection includes: rs2073724, rs3130713, rs3130531, rs3095250, rs3130532, rs3130534, rs2844615, rs6906846, rs2524067, rs7382297, rs2394963, rs2524095, rs16899203, rs9366778, rs9295970, and rs2523534; (ii) the 10th single nucleotide The type of collection includes: rs3130712, rs28480108, rs3134762, rs19966319, rs9264523, rs3132488, rs3134745, rs3130693, rs3132486, rs2853948, rs6906846, rs9378228, rs6457372, rs2394963, rs2524057, rs12191877, and rs9366776; (iii) the 11th single nucleoside The acid polymorphic collection includes: rs2516049, rs2858870, rs660895, rs532098, rs3129763, rs1063355, rs9275141, rs9275184, rs7774434, rs7775228, and rs9275224 (iv) the 12th single nucleotide polymorphic collection includes: rs9263957, rs9263969, rs3134762, rs11966319, rs2248880, rs9264532, rs2524099, rs2074488, rs2395471, rs5010528, rs13207315, rs3132488, rs3130693, rs9391714, rs4386816, rs2524057, Rs16899205, and rs9295970; the single nucleotide polymorphism derived from the (4) HLA-DPB1 gene is selected from a 13th single nucleotide polymorphism set, a 14th single nucleotide polytype set, a group consisting of a 15th single nucleotide polymorphic set and a 16th single nucleotide polymorphic set, wherein (i) the 13th single nucleotide polymorphic collection comprises: rs3128955, Rs3130588, rs9277194, rs9348904, rs9296073, rs2856816, rs3135021, rs1431403, rs3128963, rs3117229, rs7763822, rs2295120, rs3117242, rs6937034, and rs1003979; (ii) the 14th single nucleotide polymorphic collection includes: rs9296068, rs9277183, Rs3135402, rs9348904, rs2856830, rs9296073, rs2071350, rs1431402, rs1431403, rs9277550, rs3128963, rs3117229, rs9277567, rs3128918, and rs6937034; (iii) the 15th Nucleotide polymorphic collections include: rs206769, rs6920606, rs375912, rs1431399, rs987870, rs3135021, rs9277535, rs9277554, rs10484569, Rs2281390, rs3128917, rs2281388, rs3130215, and rs2269346; (iv) the 16th single nucleotide polymorphic collection includes: rs2216264, rs423639, rs3097669, rs987870, rs1431402, rs1431403, rs9277378, rs9277535, rs9277550, rs9277554, rs9277565, Rs2281390, rs2281388, rs3130215, rs6937034, rs6937061, and rs2395357; the single nucleotide polymorphism derived from the (5) HLA-DQB1 gene is selected from a 17th single nucleotide polymorphism set, an 18th single a group consisting of a nucleotide polymorphic collection, a 19th single nucleotide polymorphic collection, and a 20th single nucleotide polymorphic collection, wherein (i) the 17th single nucleotide The type of collection includes: rs9269186, rs9270986, rs615672, rs3129768, rs9272219, rs9272346, rs6908943, rs9275134, rs9469220, rs6457617, rs2647046, rs2858308, and rs9275418; (ii) the 18th single nucleotide polymorphic collection includes: Rs2647073, rs502055, rs3129768, rs9272535, rs9272723, rs34485459, rs3129716, rs7775228, rs6469219, rs5000634, rs6457617, and rs9275418; (iii) the 19th single nucleotide polymorphism The collection system includes: rs2516049, rs2858870, rs660895, rs532098, rs3129763, rs1063355, rs9275141, rs9275184, rs7774434, rs7775228, and rs9275224; (iv) the 20th single nucleotide polymorphism collection includes: rs17533090, rs9272219, rs17211510, Rs41269947, rs34485459, rs1063355, rs9275141, rs3129716, rs7774434, rs9405119, rs9469219, rs9469220, and rs9275224. The single nucleotide polymorphism derived from the (6) HLA-DRB1 gene is selected from a 21th single nucleotide polymorphism. a group consisting of a sex collection, a 22nd mononucleotide polytype set, a 23rd mononucleotide polytype set, and a 24th single nucleotide polytype set, wherein (i) the group The 21st mononucleotide polymorphic collection includes: rs9268831, rs9268861, rs7747521, rs9268877, rs9269186, rs2027852, rs615672, rs3129768, rs9272219, rs9272346, rs9275134, rs7775228, rs9469220, rs6457617, rs2647046, and rs2858308; (ii) The 22nd mononucleotide polymorphic collection includes: rs9268877, rs4410767, rs7749092, rs17210980, rs2647073, rs615672, rs674343, rs502771, rs399787 2, rs9271367, rs9271720, rs2187668, rs34485459, rs3129716, and rs9405119; (iii) the 23rd mononucleotide polymorphic collection includes: rs9405098, rs3129871, rs13209234, rs9268832, rs6903608, rs602875, rs660895, rs9271366, rs3129769, rs17211510, rs2187668, rs9275141, rs9275184, rs9275383, rs2856717, rs2858305, rs13192471 And rs3104405; (iv) the 24th single nucleotide polymorphic collection includes: rs2395175, rs9405035, rs9268831, rs6903608, rs9268877, rs9269186, rs7749092, rs2027852, rs17210980, rs2516049, rs615672, rs660895, rs674313, rs502771, rs3997872 , rs9271366, rs2187668, rs34485459, rs9275141, rs7755224, rs3129716, and rs3104404.

Another object of the present invention is to provide a device for predicting a human leukocyte antigen dual gene comprising no more than 200 nucleotide probes, wherein the probe can detect the above-described single nucleotide polymorphism; It is fixed to the device.

The present invention constructs a human leukocyte antigen genotype prediction model with ethnic specificity for Asian ethnic groups, comprising 437 Han Chinese blood samples with Affymetrix 5.0 and Illumina 550K single nucleotide polymorphism, of which 214 samples are also found in Affymetrix 6.0. Single nucleotide polymorphism data. All individuals were typed on a 6 human leukocyte antigen locus (loci) to a 4-digit resolution and used in the human leukocyte antigen genotype prediction model as a training and testing set. The results of the present invention show that a larger number of samples and a higher single nucleotide polymorphism density generally result in more accurate predictions. Furthermore, the flanking region optimized for the human leukocyte antigen dual gene in the Asian species of the present invention is generally shorter than the flanking region of the donor. In the most accurate model, the flanking region is 20-200 kb (median 70 kb) of the human leukocyte antigen dual gene across different wafer data sets. When the human leukocyte antigen dual gene is shorter, the flanking region increases, and the human leukocyte antigen dual gene density increases. The best model of the invention provides accurate predictions among Asian species. In addition, the present invention also provides practical recommendations for a population-specific human leukocyte antigen genotypic prediction model for dual gene regions, wafers, and Imputation. The present invention requires only about 20 mononuclear acid polymorphisms to correctly predict a leukocyte antigen genotype, so that a white blood cell antigen genotype information can be obtained at a price of only 1/10.

The embodiments of the present invention will be further described in conjunction with the following drawings, which are set forth to illustrate the invention and not to limit the scope of the present invention. The scope of protection of the present invention is defined by the scope of the appended claims, and the scope of the invention is defined by the scope of the appended claims.

Figure 1 shows the test accuracy associated with the size of different flanking regions; for each of the 6 human leukocyte antigen dual genes (lines of different colors), the test accuracy rate increases as the size of the flanking region increases; Affy 6.0 in the figure The data for the wafer is expressed as unfilled single nucleotide polymorphism.

Figure 2 is the test accuracy of the optimized model generated by each gene-formed wafer; the graph shows the test accuracy and interpretation rate of the 6 human leukocyte antigen dual genes (the threshold of credibility is 0); Filling (A) and unfilled (B) single nucleotide polytypes of the three genetically shaped wafers and the combined wafers of the three genetically shaped wafers (different color strips).

definition

The terminology used in the specification refers to the general meaning in the field. For the purposes of the following discussion in this specification, some terms are indicated in a special font format for convenience, for example, in italics and/or parentheses. The use of these font formats does not affect the scope and meaning of the term itself. Whether or not marked in a special font format, the scope and meaning of the term itself are the same. Therefore, the use of any equivalent language or synonym is not intended to change its meaning. The use of one or more synonyms does not exclude the use of other synonyms. The use of any terms in the embodiments of the present invention is merely illustrative and is not intended to limit the scope and meaning. Likewise, the scope of the invention is not limited only by the embodiments that are present.

Unless specifically defined, all technical and scientific terms appearing herein have the ordinary meaning recognized by those of ordinary skill in the art.

The terms "about" and "about" as used in the present invention mean 20%. Within the scope of the present invention, it is preferably within the range of 10%, and more preferably within the range of 5%. The numbers provided herein are approximate and, if not explicitly indicated, are intended to have an approximate or approximate meaning.

Example

The rsID of all single nucleotide polymorphisms (SNPs) provided by the present invention, the sequence and the position of the single nucleotide variation contained therein and the base of the variation thereof are disclosed in the National Biotechnology of the United States prior to the application of the present invention. Single Nucleotide Polymorphism Database (SNP database, dbSNP) of the National Center for Biotechnology Information (NCBI).

The apparatus, the device, the method, the related results, and the like according to the embodiments of the present invention described below are for illustrative purposes only and are not intended to limit the scope of the present invention. The names in the examples or their sub-names are for convenience of reading and are not intended to limit the scope of the invention. Further, the theory disclosed herein, whether or not it is inaccurate, is not intended to limit the scope of the invention.

Research design

An estimation equation for human leukocyte antigen genotypes using an unphased genotype was established using an estimating equation approach. For each dual gene, the human leukocyte antigen genotype prediction method is performed in two stages. The first stage is to construct a predictive model, and the second stage is to verify the model produced by the first stage. In this first phase, a set of confusing genotypes is selected to establish a predictive model. This selection is assessed using an objective function, which is a negative log-likelihood of the human leukocyte antigen-specific gene-specific confounding genotype (based on Akaike Information Criterion). Next, the genotype selection is carried out by a method of forward-selection and backward-elimination. Start with a genotype associated with a human leukocyte antigen dual gene and add one genotype one by one. This second phase uses a separate set of samples to validate the predictive model of the first phase. Unphased genotypes and non-confusing human leukocyte antigen pairs (phased HLA alleles) were provided as separate samples. According to the parsimonious rule, the best predictive model requires the least possible franking region and the least probable single nucleotide polymorphism to produce the most accurate predictions. The sample used in the present invention is a blood sample of 437 Han Chinese living in Taiwan obtained from the Taiwan Han Chinese Cell and Genome Bank.

Genotype analysis

Three commercial wafers were used in the present invention: 1) Affymetrix Genome-Wide Human SNP Array 5.0 wafer (Affy 5.0); 2) Affymetrix Genome-Wide Human SNP Array 6.0 wafer (Affy 6.0); and 3) Illumina's HumanHap550 Genotyping BeadChip wafer (Illumina 550) Among them, 437 white blood cell DNA samples were genotyped using Affy 5.0 and Illumina 550 wafers, and 214 of 437 samples were also genotyped using Affy 6.0 wafers. The human major histocompatibility complex (MHC) located on the short arm of chromosome 6 is also known as the human leukocyte antigen (HLA) gene group region. The Affy 5.0 wafer has 1,406 single nucleotide polymorphisms (SNPs). The Affy 6.0 wafer has 2,203 single nucleotide polytypes; the Illumina 550K wafer has 1,939 single nucleotide polytypes (as shown in Table 1), while the intra-MHC region is located at the centromere The centromeric end of the HLA-A dual gene (position on chromosome 6: 30, 018, 310-30, 021, 632; NCBI build 36.3) and the HLA-DPB1 dual gene located at the telomeric end (position on chromosome 6: 33, 151, 738-33, 162, 954) is bounded. This region includes class I loci ( HLA-A , HLA-B , HLA-C ) and class II loci ( HLA-DRB1 , HLA-DQB1 ). For HLA-A, -B, -C, - DQB1 and - DRB1 allele, the system using the Dynal RELI SSO typing kit (Dynal Biotech Ltd., UK) genotyping; for HLA-DPB1 alleles, the system using Genotyping was performed on the Gold SSP HLA-DPB1 High Resolution kit (Invitrogen, Calif.). All genotyping was performed by the National Research Center for Genetic Medicine of the Academia Sinica, and the single-nucleotide polymorphic call rate was greater than 98%.

For genome-wide association studies (GWASs), the present invention evaluates the utility of genotype imputation in constructing a human leukocyte antigen genotype prediction model. For the consistency of data and the best filling performance, the present invention uses the dataset of MaCH software and Hanren Beijing (CHB) and Tokyo, Japan (JPT) as a reference, which is used to fill the single nucleotide polymorphism of the present invention. Genotypes from the International Human Genome HapMap Project. The present invention examines all single nucleotide polymorphisms in the MHC region, and typically evaluates and screens for single nucleotide polymorphism to control its quality prior to genotyping, when severely violating Hardy-Weinberg At equilibrium (p<10 ^-4 ), the single nucleotide polymorphism rate (call rate) <0.95, and the minority-allele frequency <0.01, excludes the single nucleotides. Type. Furthermore, the filled single nucleotide polymorphisms of the present invention each have a posterior probability > 0.8 for the results of the analysis of the MaCH software, a discrimination rate of > 0.95, and a minority of the dual gene frequencies > 0.01.

On the other hand, in order to test the repeatability and uniformity of the single nucleotide polymorphism between the wafers, the present invention compares the single nucleotide polymorphism data of the two wafers. Judging the consistency of genotype data is calculated by Cohen's kappa coefficient, and a Kappa value greater than 0.9 usually means that the data of the two wafers has high consistency. The present invention also compares the differences between selected genotypes in the process of constructing a human leukocyte antigen genotype prediction model for each of the two wafers to determine whether the selected genotype is specific to the wafer, i.e., unique. This difference is defined as , plat _i and plat _j are two different wafers; plat ( plat _i , plat _j ) is a union of single nucleotide polymorphisms of two different wafers; and ∩ ( plat _i , plat _j ) is the intersection of single nucleotide polymorphisms of two different wafers.

Frequency distribution of human leukocyte antigen dual gene and size of flanking region among different ethnic groups

There is a substantial difference between the human leukocyte antigen dual gene and its dual distribution among different ethnic groups, which is the evolution history of the response population in recent times. Furthermore, the human leukocyte antigen dual gene is distributed in different regions on the sixth pair of chromosomes, including several single nucleotide polymorphisms. The present invention explores the dual gene frequency distribution of Asian samples and Caucasian samples in the International Human Genome HapMap Project. For each human leukocyte antigen dual gene, the present invention uses a chi-square and Fisher's exact test to determine whether the human leukocyte antigen dual gene differs between the two populations. The present invention constructs the human leukocyte antigen genotype prediction model by extending the flanking region of ±10 kb to ±400 kb. In the Han Chinese, the most suitable flanking region for each human leukocyte antigen dual gene is the most simplified Then decide. In addition, the present invention also compares the size of the Asian flanking region (Affy 5.0 wafer) with the size of the known Caucasian flanking region.

Cross validation (Corss-validation)

Prior to the initiation of human leukocyte antigen prediction analysis, the present invention divided the data into groups for cross-validation (CV). Taking a 10-fold cross-validation as an example, the data is divided into a training data set (9/10 of data) and a testing data set (1/10 of data). For each cross-validation subset, the accuracy of the test set is calculated and defined as , where T _v is the correct prediction of the number of samples in the test set, and N _v is the total number of samples in the test set. The mean test accuracy is the average of 10 cross-validated subsets, indicating the performance of the constructed model in predicting the human leukocyte antigen dual gene. Prediction of human leukocyte antigens may not be cross-validated, however performing cross-validation may avoid over-fitting of the predictive model and may save time and cost in obtaining an independent sample set for evaluation. The present invention constructs a human leukocyte antigen genotype prediction model, and therefore, uses ten-fold cross-validation.

Confidence threshold

For each sample in the test set, its probability value is assigned to each possible human leukocyte antigen pair pair of genes for a particular haplotype. These values are based on the provided confusing genotype and the non-confusing human leukocyte antigen pair pair. After the probability is assigned, if the probability exceeds a pre-specified threshold of confidence, the human leukocyte antigen pair pair with the greatest probability is selected. In general, the confidence threshold is set to 0, indicating that the call rate is 100% (ie, all samples are predicted). If the confidence threshold is set to 0.5 (or any value greater than 0), only samples with a maximum predicted probability that exceeds the confidence threshold will be used. The present invention sets the confidence threshold to 0, 0.5, or 0.9 to assess the impact of the credibility threshold on constructing a human leukocyte antigen genotype prediction model.

result

The present invention uses 214 samples to generate six representative human leukocyte antigens ( HLA-A, HLA-B , HLA-C, HLA ) from three different wafers (Affy 5.0, Affy 6.0, and Illumina 550K wafers) . -DRB1 , HLA-DQB1, HLA-DPB1 ) Frequency distribution of the dual gene. The present invention also analyzes 180 Caucasian samples obtained from the International Human Genome HapMap Project, but without the HLA-DPB1 data for these samples. The human leukocyte antigen locus is most likely HLA-B . In the Han Chinese, 44 pairs of genes were observed to cross the HLA-B region, while 32 pairs of genes were observed across the HLA-B region in Caucasian humans in the International Human Genome HapMap Project. The chi-square test and Fisher accuracy shown in Caucasian and Chinese human, HLA-A, -B, -C , -DQB1, and - having a significant difference (all p values DRB1 allele frequency of allele distribution <0.0001; HLA-A, -B, -C, -DQB1 , and - DRB1 degrees of freedom are 29, 62, 23, 16, and 35, respectively, which shows that human leukocyte antigen dual gene The frequency distribution varies greatly among different races. That is to say, the human leukocyte antigen genotype prediction model constructed by a group of human leukocyte antigen dual genes may produce poor predictions when predicting different ethnic groups.

Different wafers not filled

The use of a single genetic typing technique may bias the prediction of human leukocyte antigen dual genes. In order to overcome this problem, the 214 Han Chinese samples of the Taiwanese region of the present invention were genetically shaped with three wafers (Affy 5.0, Affy 6.0, and Illumina 550K). The results of each wafer and the results of the combined wafers of the three wafers were used to construct a model for human leukocyte antigen prediction. Finally, the present invention evaluates whether predictive models derived from these three data sets yield comparative predictions.

There is little overlap of data between pairs of wafers (as shown in Table 1). Affy 6.0 has the most single nucleotide polymorphism in the human MHC region, while Affy 5.0 is the least (as shown in Table 1). Table 2 shows the consistency coefficients between pairs of wafers for the observed genotype. Comparing the two Affymetrix arrays, the genotypes present in both arrays have a consensus coefficient of up to 0.9926. This high degree of agreement indicates high quality genotyping, which is supported by comparing genotypes between different wafers.

Table 1. International Human Genome HapMap Project and Extended MHC between Three Genotyping Wafers

In general, combined wafers produce more accurate predictions of human leukocyte antigen dual genes than individual wafers. When the credibility threshold is 0, the average test accuracy of the combined wafers is 89.78%, but for the separate Affy 5.0, Affy 6.0, and Illumina 550K, the average test accuracy is only 86.92%, 88.42%, and 88.06, respectively. % (as shown in Figure 2A) shows that higher single nucleotide polymorphisms increase the predictive accuracy of human leukocyte antigen dual genes.

Affy 6.0 produced the most accurate human leukocyte antigen dual gene prediction for comparison between three genetically shaped wafers. For example, at the HLA-DRB1 locus, Affy 6.0 is 3.52% more accurate than Affy 5.0; at the HLA-DPB1 locus, Affy 6.0 is 2.58% more accurate than Illumina 550K. Affy 6.0 may have the highest genotypic density in the major histocompatibility complex regions of humans. When the confidence threshold is 0, Affy 6.0 and HLA-DQB1 can obtain the highest test accuracy (95.79%), while Illumina 550K and HLA-B have the lowest accuracy (80.37%, as shown in Figure 2A). ). By using a confidence threshold of 0.9 to the maximum probability of all possible human leukocyte antigen pair pairs, the highest accuracy of the HLA-C locus is increased to 98.62% (obtained by Illumina 55K, the interpretation rate is 77.47%). The minimum accuracy of HLA-B increased to 87.67% (obtained by Affy 5.0, the interpretation rate was 64.94%). The accuracy range produced by the Illumina 550K-based predictive model is significant in the HLA locus compared to that observed in other genotyping wafers. When the confidence threshold is 0, the accuracy of the Illumina 550K predicted wafer in the HLB-B dual gene is only 80.37%, but the accuracy of the HLA-DQB1 dual gene is 95.29%. For HLA-B and HLA-DPB1 dual genes, the prediction of Affy 5.0 was 0.45% and 0.96% more accurate than Ilumina 550K, respectively. For the HLA-A and HLA-DRB1 dual genes, the Illumina 550K prediction was 1.56% and 0.27% more accurate than Affy 6.0, respectively (as shown in Figure 2). These results show that the slight advantage associated with Affy 5.0 and Illumina 550K may be due to the particular single nucleotide polymorphism on these wafers. In general, the accuracy of these predictive models is often comparable between genotyping wafers.

Conduct an effective flanking region (eg, to produce the shortest side gene sequence extension of the most accurate human leukocyte antigen dual gene prediction). The shortest effective flanking region was identified at the HLA-C locus using Illumina 550K (±10 kb). The HLA-C is 3,325 bp in length and this effective flanking region covers 22 single nucleotide polymorphisms, of which 13 single nucleotide polymorphisms are included in the HLA-C predictive model (when the threshold is credible) 0, the test accuracy rate is 92.01%). When using Affy 6.0 data, the longest effective flanking region of HLA-A is ±350 kb (as shown in Table 3), in this region 299 single nucleotide polymorphisms, of which 16 single nucleotide polymorphisms It is included in the HLA-A prediction model (when the credibility threshold is 0, the test accuracy is 85.29%)

For each human leukocyte antigen dual gene, the present invention further evaluates genotypes that overlap between wafers in a human leukocyte antigen genotype prediction model. When comparing Affy 6.0 with pooled wafers, the maximum overlap genotype ratio of HLA-DRB1 's unfilled single nucleotide polymorphism data was 21.36%, indicating that different wafers use unique single nucleotide polymorphisms. The wafer-specific genotypes were selected, and these genotypes were used to construct different human leukocyte antigen genotype prediction models.

Filling of different wafers

The combined wafer of the present invention (with a confidence threshold of 0 and an average test accuracy of 90.17%) is compared with three independent wafers (the reliability threshold is 0, the average test accuracy of Affy 6.0, Affy 5.0, and Illumina 550K, respectively). For 89.90%, 88.61%, and 89.75%), a more accurate prediction of the human leukocyte antigen dual gene was produced. Higher single nucleotide polymorphism density increases the accuracy of genotyping, and thus increases the accuracy of the final prediction.

Affy 6.0 is generally more accurate in predicting the human leukocyte antigen dual gene (the accuracy of the HLA-DPB1 locus is as high as 4.23% compared to Affy 5.0 and as accurate as Illumina 550K up to 4.61%). Figure 2B shows). In these models (with a threshold of 0), the HLA-DQB1 locus has the highest test accuracy (96.75%, available from Illumina 550K). By using a confidence threshold of 0.9 to the maximum probability of all possible human leukocyte antigen pair pairs, the highest accuracy of the HLA-C locus was increased to 99.09% (from Illumina 55K, the interpretation rate was 77.47%). In addition to the HLA-B obtained by Affy 5.0, the improvement of the accuracy based on this credibility threshold is most significant in the HLA-DRB1 dual gene. When the credibility threshold is changed from 0 to 0.9, the accuracy rate increases from 86.67%. To 95.90%. However, at the HLA-A locus, Affy 5.0 (with a threshold of 0) produced a 0.45% accuracy prediction over Affy 6.0. Illumina 550K was 4.61% more accurate than Affy 6.0 in HLA-DPB1 , and Illumina 550K was predicted to be 0.27%, 1.10%, 1.11 better than Affy 6.0 at HLA-A, -B, -C, -DQB1, and -DRB1 loci, respectively. %, 0.05%, and 1.19% (as shown in Figure 2), the results show that the predicted advantages of Affy 5.0 and Illumina 550K may be derived from the unique single nucleotide polymorphism on these wafers.

The present invention also assesses the effective flanking region of each human leukocyte antigen locus padding. One of the shortest flanking regions is at the HLA-DPB1 locus (±20 kb) and is recognized by Affy 5.0 (as shown in Table 3). This region covers 133 single nucleotide polymorphisms, 34 of which were selected for the HLA-DPB1 prediction model (with a confidence threshold of 0 and a test accuracy of 88.28%). The other shortest flanking region is at the HLA-C locus (±20 kb) and is recognized by Affy 6.0, Illumina 550K, and pooled wafers. The longest effective flanking region is at HLA-A (±200 kb) and is obtained from Illumina 550K (as shown in Table 3). There were 515 single nucleotide polymorphisms in these regions, 17 of which were used in the HLA-A predictive model (the confidence threshold was 0 and the test accuracy was 86.93%).

For each human leukocyte antigen dual gene, the overlapping genotypes used between different predictive models are up to 60.08%. Therefore, the filling seems to reduce the difference between different wafers.

Filling and unfilling of different wafers

Compared with the test accuracy of filling and non-filling of different wafers between the prediction models of the present invention, the prediction model constructed by filling the single nucleotide polymorphism is more accurate than the prediction model constructed by the non-filled single nucleotide polymorphism. (The threshold of credibility is 0, and the average accuracy is 89.61% and 88.30%, respectively).

For the credibility threshold of 0, the filled merged wafer has the highest HLA-DQB1 dual gene test accuracy (97.18%), while the unfilled Illumina 550K has the lowest HLA-B dual gene test accuracy (80.37). %). By using a confidence threshold of 0.9 to the maximum probability of all possible human leukocyte antigen pair pairs, the highest accuracy of the Illumina 550K HLA-C locus was increased to 99.09% (filled and the interpretation rate was 82.40%).

Comparing the predicted and unfilled test prediction accuracy between different wafers, when using the filled single nucleotide polymorphism to construct the human leukocyte antigen genotype prediction model, the specific genotype variation is usually reduced. For each human leukocyte antigen dual gene between different wafers, the average ratio of overlap between the genotypes selected to construct the model was increased by 25.02%. These results minimize the differences between different genetically shaped wafers.

Example 1

Using the above results, a method for predicting a human leukocyte antigen dual gene can be developed according to which the steps include: (a) providing a human nucleic acid sample; and (b) identifying a single nucleotide polymorphic collection of the human nucleic acid sample. Genotype, the single nucleotide polymorphism set comprising various different single nucleotide polymorphisms on the human leukocyte antigen gene; (c) using a predictive model to analyze each single nucleotide in step (b) a type of genotype to obtain a calculated value, wherein the predictive model uses a single nucleotide polytype genotype to predict a human leukocyte antigen dual gene; and (d) predicts the calculated value obtained according to step (c) Human leukocyte antigen dual genotype in human samples.

The present invention uses the single nucleotide polymorphism of the Han nationality specificity and predicts the genotype of the dual gene of each of the six human leukocyte antigen loci by an algorithm. The optimal number of single nucleotide polymorphisms for each wafer is different, as shown in Table 4.

The content of the single nucleotide polymorphic set of the above three wafers and their combined wafers As shown in Table 5. The present invention predicts the human human leukocyte antigen locus dual gene by using the group-specific single nucleotide polymorphism set.

discuss

Because the direct typing technique of the known human leukocyte antigen genotype is not economical, the present invention recognizes a specific human leukocyte antigen genotype based on the confounding genotype corresponding to the human leukocyte antigen dual gene, and constructs it. Predictive models of human type 1 human leukocyte antigens ( HLA-A, HLA-B , and HLA-C ) and type II human leukocyte antigens ( HLA-DRB1, HLA-DQB1 , and HLA-DPB1 ). The present invention compares the frequency distribution of the dual genes of Asians (Taiwanese Han Chinese) and Caucasians (International Human Genome HapMap Scheme) and identifies significant differences between different ethnic groups. The present invention constructs several predictive models of human leukocyte antigen genotypes with high predictive accuracy and validates important parameters related to the models (eg, effective flanking regions, wafer accuracy, and effects of filling). Therefore, the model provided by the present invention can accurately predict these genotypes in Asian races, and thus can be applied to the detailed analysis of the direct effects of diseases associated with human leukocyte antigens.

The present invention discriminates whether a denser single nucleotide polymorphism set can produce a more accurate human leukocyte antigen dual gene prediction. In the present invention, the MaCH soft system used to fill wafer data and construct denser single nucleotide polymorphisms is based on the International Human Genome HapMap Project and/or the 1000 Genome Project, http: //www.1000genomes.org) data for reference. The present inventors have found that a constructed human leukocyte antigen genotype prediction model is typically provided to provide a higher predictive accuracy, emphasizing the positive effect of using a higher density single nucleotide polymorphism. Therefore, a novel customized single nucleotide polymorphic array can be constructed, which includes all the single nucleotide polymorphisms of the human leukocyte antigen genotype prediction model that constitute the genotyping or filling of the genes, so as to improve the prediction accuracy. Confirmation rate.

By increasing the threshold of credibility to 0.5 or 0.9, the prediction accuracy of the Asian-specific human leukocyte antigen genotype prediction model of the present invention is close to 100%.

At the same time, in order to produce a more accurate prediction model, the present invention changes the sample size of the validation set and the test set. The Affy 6.0 genotype data of the present invention was used to construct a human leukocyte antigen genotype prediction model and cross-validation was performed using two, four, and ten folds. When the confidence threshold is 0 (ie, the interpretation rate is 100%), different cross-validation results in a consistent sample size. In order to rule out the effect of sample size, the present invention compares the effects of cross-validation with a confidence threshold of zero. An estimate of the test accuracy obtained from the evaluation of the predictive model constructed using the second, fourth, and ten-fold cross-validation (the threshold of confidence is zero). Ten-fold cross-validation of the HLA-DQB1 locus has the best accuracy of the test (95.55%), while the second fold cross-validation of the HLA-B locus of the lowest accuracy of test (76.64%). For the two-fold cross-validation, the test accuracy ranged from 76.64% ( HLA-B ) to 94.39% ( HLA-DQB1 ). As cross-validation increases to 10x, the test accuracy is close to 95.55% ( HLA-DQB1 ). This improvement is most pronounced for HLA-A because the accuracy of the two-fold cross-validation prediction model is 80.84%, while the 10-fold cross-validation results in an accuracy of 85.29%. A similar trend was observed in the other five human leukocyte antigen loci. These trends may reflect a large sample size, including an adequate number of human leukocyte antigen dual genes in the validation set, thereby increasing the accuracy of the prediction. Although the degree of cross-validation affects the accuracy of the test, its variation is extremely small. Therefore, the population-specific human leukocyte antigen genotype prediction model of the present invention is not affected by different cross-validation.

The present invention focuses on the use of 214 samples of Han Chinese in Taiwan to produce a human leukocyte antigen genotype prediction model, and the model uses Affy 5.0, Affy 6.0, and Illumina 550K wafers for genotyping of nucleotide polymorphisms (SNPs) (Genotyping). ). At the same time, in order to evaluate the influence of the number of samples, the present invention also uses 437 Han Chinese samples to analyze the genotyping of the Illumina 550K wafer (including the original 214 samples) to generate a human leukocyte antigen genotype prediction model. The human leukocyte antigen genotype prediction model constructed with 437 samples (average test accuracy rate was 90.36%) was better than the 214 sample construction prediction model (average test accuracy rate was 86.84%). Therefore, a larger sample size can increase the accuracy of human leukocyte antigen prediction.

in conclusion

After a decade of research, many human leukocyte antigen dual genes are known to have specific immune functions. The experimental method of linking single nucleotide polymorphism with human leukocyte antigen dual gene saves considerable time and cost compared with human leukocyte antigen direct typing technology, and makes the study of large-scale human leukocyte antigen mutation feasible. Although human leukocyte antigen distribution varies among human populations, most existing human leukocyte antigen genotype prediction models are based on Caucasian samples. By genetically shaping a large number of Han Chinese samples, the present invention finds many Han human-specific human leukocyte antigen dual genes and constructs a population-specific human leukocyte antigen genotype prediction model. The validation set of the invention covers many uncommon and ethnic groups in the human leukocyte antigen locus A dual gene substantially increases the accuracy of the prediction.

The specific method parameters (e.g., sample size, single nucleotide polymorphism, and padding) used in the present invention are factors that produce a human leukocyte antigen genotype prediction model for Asian races. The present invention is to obtain a good HLA-A Han Chinese samples, -B, -C, -DRB1, -DQB1 , and - DPB1 alleles prediction accuracy. The present invention produces effective human leukocytes using single nucleotide polymorphism data from Affymetrix Genome-Wide Human SNP Array 5.0, Affymetrix Genome-Wide Human SNP Array 6.0, Illumina HumanHap550 BeadChip, or a combined wafer of these three wafers Antigen genotype prediction model to identify human leukocyte antigen genotypes in Asia. The novel predictive tools of the present invention can help identify genetic risk factors for immune related diseases, such as Grave's disease. In addition, it is also possible for a person of ordinary skill in the art to study the human leukocyte antigen genotype in a large population of people.

The prediction method and the application device provided by the present invention have industrial use value, but the above description is only for the preferred embodiment of the present invention, and those skilled in the art can make other according to the above description. Various modifications, but such changes are still within the spirit of the invention and the scope of the patents defined below.

Claims (5)

A method for predicting a genotype of a human leukocyte antigen, the method comprising the steps of: (a) providing a human nucleic acid sample; and (b) identifying a genotype of a single nucleotide polymorphic collection of the human nucleic acid sample, the single nucleotide The polymorphic collection comprises various single nucleotide polymorphisms located on a human leukocyte antigen gene; and (c) predicting the human leukocyte antigen dual genotype of the human nucleic acid sample based on the results discerned in step (b) Among them, the human globular antigen dual genotype can be predicted to include: (1) HLA-A gene, (2) HLA-B gene, (3) HLA-C gene, (4) HLA-DPB1 gene, (5) a genotype of the HLA-DQB1 gene or (6) HLA-DRB1 gene; wherein, the single nucleotide polymorphism of the (1) HLA-A genotype is predicted to be selected from a first single nucleotide polymorphism a group consisting of a collection, a second single nucleotide polymorphism set, a third single nucleotide polytype set, and a fourth single nucleotide polytype set, wherein: (i) the The first mononucleotide polymorphic collection includes: rs1633085, rs2254071, rs407238, rs9258881, rs2975046, rs2735096, rs417162, rs9260954, rs 6917477, rs6457144, rs9261394, and rs2523990; (ii) the second mononucleotide polytype collection includes: rs4122198, rs16895757, rs1632973, rs9357086, rs11759549, rs3115628, rs3094165, rs2734925, rs2517755, rs2256919, rs11756025, rs7382061 Rs6457144, rs2517646, and rs7744914; (iii) the third single nucleotide polymorphic collection includes: rs3094165, rs9258883, rs3132714, rs1611493, rs2524005, rs2860580, rs12665039, rs6457109, rs3869062, rs3893464, rs5009448, rs2571375, rs7758512 And rs9261394; (iv) the fourth single nucleotide polymorphic collection includes: rs2523409, rs1611133, rs3115628, rs2517859, rs1611732, rs2523998, rs2860580, rs12202296, rs2248153, rs2975046, rs6457109, rs5009448, rs9260932, and rs6457144; The single nucleotide polymorphism of the (2) HLA-B genotype is predicted to be selected from a fifth single core. a group consisting of a polymorphic glycoside collection, a 6th single nucleotide polymorphic collection, a 7th single nucleotide polymorphic collection, and an 8th single nucleotide polymorphic collection, wherein (i) the fifth mononucleotide polymorphic collection includes: rs3130944, rs3130532, rs3130534, rs3134762, rs16899207, rs2524089, rs9366778, rs2524166, rs9295984, rs4394275, rs9378249, rs2523534, rs9266406, rs2844558, rs5022119 rs3099848, rs4081552. Rs2848716, rs2596454, and rs2248462; (ii) the sixth mononucleotide polymorphic collection includes: rs11966319, rs2853948, rs6906846, rs9378228, rs2524051, rs9366778, rs16867947, rs4394274, rs4394275, rs2523591, rs9501572, rs7761068, rs2523535, Rs9266406, rs5006724, rs13198903, rs9266669, rs9266689, rs3099849, rs2442749, rs1051796, rs2596464, rs3099836, and rs3131622; (iii) the seventh single nucleotide polymorphism collection includes: rs9264868, rs9264942, rs3094691, rs2156875, rs2523619, Rs2442719, rs2596501, rs2523589, rs2523554, rs2844573, rs9266395, rs9266440, rs929598 6. rs2442749, rs2596560, rs3128982, rs2284178, and rs7758090; (iv) the eighth mononucleotide polymorphic collection includes: rs3094691, rs7453967, rs4394274, rs4394275, rs2596509, rs2596501, rs1058026, rs2523591, rs2523589, rs2523554, Rs2523545, rs9501572, rs2844575, rs9266395, rs9266406, rs5006725, rs9295986, rs6933050, rs4959068, rs5022119, rs13198903, rs9266689, rs2251396, rs1051796, rs3094584, rs9765960, and rs3128982; wherein, the (3) single-core of HLA-C genotype is predicted The polymorphism is selected from a ninth single nucleotide polymorphic set, a tenth single nucleotide polymorphic set, an eleventh single nucleotide polymorphic set, and a tenth single a group consisting of a polymorphic collection of nucleotides, wherein (i) the ninth single nucleotide polymorphic collection comprises: rs2073724, rs3130713, rs3130531, rs3095250, rs3130532, rs3130534, rs2844615, rs6906846, rs2524067, rs7382297 , rs2394963, rs2524095, rs16899203, rs9366778, rs9295970, and rs2523534; (ii) the 10th single nucleotide polymorphic collection includes: rs3130712 rs28480108, Rs3134762, rs19966319, rs9264523, rs3132488, rs3134745, rs3130693, rs3132486, rs2853948, rs6906846, rs9378228, rs6457372, rs2394963, rs2524057, rs12191877, and rs9366776; (iii) the 11th single nucleotide polymorphic collection includes: rs2516049, Rs2858870, rs660895, rs532098, rs3129763, rs1063355, rs9275141, rs9275184, rs7774434, rs7775228, and rs9275224; (iv) the 12th single nucleotide polymorphism collection includes: rs9263957, rs9263969, rs3134762, rs11966319, rs2248880, rs9264532 Rs2524099, rs2074488, rs2395471, rs5010528, rs13207315, rs3132488, rs3130693, rs9391714, rs4386816, rs2524057, rs16899205, and rs9295970; wherein the single nucleotide polymorphism of the (4) HLA-DPB1 genotype is predicted to be selected from a 13th mononucleotide polytype set, a 14th single nucleotide polymorphic set, a 15th single nucleotide polymorphic set, and a 16th single nucleotide polymorphic set a group, wherein (i) the 13th single nucleotide polymorphic collection comprises: rs3128955, rs3130588, rs9277194, rs9348904, rs9296 073, rs2856816, rs3135021, rs1431403, rs3128963, rs3117229, rs7763822, rs2295120, rs3117242, rs6937034, and rs1003979; (ii) the 14th single nucleotide polymorphic collection includes: rs9296068, rs9277183, rs3135402, rs9348904, rs2856830, Rs9296073, rs2071350, rs1431402, rs1431403, rs9277550, rs3128963, rs3117229, rs9277567, rs3128918, and rs6937034; (iii) the 15th single nucleotide polymorphic collection includes: rs206769, rs6920606, rs375912, rs1431399, rs987870, rs3135021 Rs9277535, rs9277554, rs10484569, rs2281390, rs3128917, rs2281388, rs3130215, and rs2269346; (iv) the 16th single nucleotide polymorphic collection includes: rs2216264, rs423639, rs3097669, rs987870, rs1431402, rs1431403, rs9277378, rs9277535, Rs9277550, rs9277554, rs9277565, rs2281390, rs2281388, rs3130215, rs6937034, rs6937061, and rs2395357; wherein, the single nucleotide polymorphism of the (5) HLA-DQB1 genotype is predicted to be selected from a 17th a group consisting of a single nucleotide polymorphic collection, an 18th single nucleotide polymorphic collection, a 19th single nucleotide polymorphic collection, and a 20th single nucleotide polymorphic collection Wherein (i) the 17th single nucleotide polymorphic collection comprises: rs9269186, rs9270986, rs615672, rs3129768, rs9272219, rs9272346, rs6908943, rs9275134, rs9469220, rs6457617, rs2647046, rs2858308, and rs9275418; (ii) The 18th mononucleotide polytype collection includes: rs2647073, rs502055, rs3129768, rs9272535, rs9272723, rs34485459, rs3129716, rs7775228, rs6469219, rs5000634, rs6457617, and rs9275418; (iii) the 19th single nucleotide polymorphism The sexual collections include: rs2516049, rs2858870, rs660895, rs532098, rs3129763, rs1063355, rs9275141, rs9275184, rs7774434, rs7775228, and rs9275224; (iv) the 20th single nucleotide polymorphism collection includes: rs17533090, rs9272219, rs17211510 , rs41269947, rs34485459, rs1063355, rs9275141, rs3129716, rs7774434, rs9405119, rs9469219, rs9469220, and rs9275224; wherein, the (6) HLA is predicted The single nucleotide polymorphism of the -DRB1 gene genotype is selected from a 21st single nucleotide polymorphism set, a 22nd mononucleotide polytype set, and a 23rd single nucleotide polytype. a group consisting of a collection of 24th single nucleotide polymorphisms, wherein (i) the 21st single nucleotide polymorphism collection comprises: rs9268831, rs9268861, rs7747521, rs9268877, rs9269186, rs2027852 , rs615672, rs3129768, rs9272219, rs9272346, rs9275134, rs7775228, rs9469220, rs6457617, rs2647046, and rs2858308; (ii) the 22nd mononucleotide polymorphic collection includes: rs9268877, rs4410767, rs7749092, rs17210980, rs2647073, rs615672 , rs674343, rs502771, rs3997872, rs9271367, rs9271720, rs2187668, rs34485459, rs3129716, and rs9405119; (iii) the 23rd single nucleotide polymorphic collection includes: rs9405098, rs3129871, rs13209234, rs9268832, rs6903608, rs602875, rs660895 , rs9271366, rs3129769, rs17211510, rs2187668, rs9275141, rs9275184, rs9275383, Rs2856717, rs2858305, rs13192471, and rs3104405; (iv) the 24th single nucleotide polymorphic collection includes: rs2395175, rs9405035, rs9268831, rs6903608, rs9268877, rs9269186, rs7749092, rs2027852, rs17210980, rs2516049, rs615672, rs660895, Rs674313, rs502771, rs3997872, rs9271366, rs2187668, rs34485459, rs9275141, rs7755224, rs3129716, and rs3104404. The method of claim 1, wherein the human nucleic acid sample is an Asian ethnic group. The method of claim 1, wherein the human nucleic acid sample is a Han nationality group. A device for predicting a human leukocyte antigen dual gene comprising no more than 200 nucleotide probes, wherein the probe detects a single nucleotide polymorphism as described in claim 1 of the patent application. The device of claim 4, wherein the probe is attached to the device.