JP7158496B2 - Disease resistance breeding gene chip of flounder and its application - Google Patents

Description

TECHNICAL FIELD The present invention belongs to the technical field of aquatic genetic breeding, and specifically relates to a method and application of a gene chip used for selection of excellent disease-resistant flounder varieties.

Aquaculture is an important source of Chinese food, and fish farming is also the key industry of aquaculture. %. Farmed fish has already become an important source of Chinese protein.

However, with the rapid development of fish farming industry, there is a shortage of good breeds and the quality of farmed breeds is deteriorating. As the scale of aquaculture expands, the level of intensification increases, and the aquaculture environment deteriorates, aquaculture diseases will occur more frequently, and problems such as increased drug residues in aquaculture products will seriously constrain the sustainable development of the fish farming industry. do. For fish only, the immunosuppression formed by high-density aquaculture causes a decrease in disease resistance in farmed fish. Research on immune disease resistance mechanisms and molecular inheritance of disease resistance in fish is still incomplete, so it is difficult to propose preventive measures against fish diseases at the molecular level. Furthermore, disease resistance functional genes and disease resistance molecular markers are deficient, making it difficult to breed excellent disease-resistant varieties. The disease causes frequent outbreaks in fish farming. According to incomplete statistics, the direct economic loss from disease in China's fish farming industry amounts to RMB 10 billion every year. Diseases have already become a bottleneck constraining the sustainable development of fish farming industry in China.

Flounder is a global saltwater farmed fish, and is also one of the leading fish in saltwater farming in China. However, even in the flounder farming industry, there is a problem that disease damage occurs frequently and the mortality rate is high. Major diseases that threaten cultured fish such as flounder include bacterial and viral diseases. Among them, edwardieriasis, vibriopathy and lymphocystis disease are the most dangerous diseases, respectively. Disease prevention measures such as antibiotics and vaccines have certain effects, but they cannot fundamentally solve the problem of diseases in aquaculture. In addition, antibiotics tend to accumulate in fish bodies, degrade the product quality of farmed fish, have potential hazards to consumer health, cause drug resistance to pathogenic bacteria, and seriously pollute the farming environment. Its susceptibility to triggering increasingly limits its application in the aquaculture industry. At the same time, the use of antibiotics also cannot meet people's ever-growing demand for residue-free and pollution-free aquatic products. Therefore, the selection of fish varieties with excellent disease resistance is one of the serious problems that need to be solved urgently in China's aquaculture industry.

Until now, the selection of superior fish varieties has been based mainly on the selection of phenotypic characteristics, including group selection, family selection, cross selection, BLUP selection, etc., and is mainly calculated based on phenotypic values that are easy to measure, such as body length and weight. The selection is made based on the breeding value given. After the molecular markers are available, selection can be made economically by identifying the molecular markers associated with important economic properties. Although the selection effect of traits or quality traits was excellent, the selection effect of quantity traits determined by multiple genes was not good. Disease resistance traits are quantitative traits controlled by multiple genes, are difficult to measure directly, and have very low selection accuracy. However, there is an urgent need to solve this problem with new breeding techniques.

A gene chip, also known as a DNA chip or a DNA microarray, uses photoetching technology, uses a silicon wafer as a solid phase carrier, synthesizes a large amount of DNA sequences through selective optimization into oligonucleotides, attaches them to slides that have undergone special processing, A DNA microarray is formed after denaturation and fixation. Based on nucleic acid hybridization technology, the gene chip can perform parallel hybridization and analysis on tens of thousands and even hundreds of thousands of DNA fragments at the same time, and has the advantages of high throughput, parallelism, high efficiency and small sample quantity. At present, gene chips have been widely applied to the diagnosis of human diseases and tumors, and the genetic analysis and genetic breeding of animals and plants. In animal breeding, gene chips have already been successfully applied to the selection of excellent breeds, especially dairy cows and pigs. For example, for dairy cows, many gene chips such as Bovine3Kchip, Bovine25K, SNPchip, BovineHD700K, and BovinLD7K have been developed one after another. Until now, North America, Europe, Australia, etc. have been using the BovineSNP50Beadchip chip (54KSNP) as a general-purpose platform for genomic SNP marker classification and testing, and based on the SNP classification results obtained by a large-scale reference group, milk production, fertility, and Perform full genome association analysis of various economic properties such as disease resistance, and construct a selection system for dairy cow genomes. Through genome selection, the early selection of newborn bulls can be achieved, the generation interval of dairy cattle breeding can be shortened, genetic progress can be promoted, the selection efficiency of sires can be greatly improved, and breeding and breeding costs can be significantly reduced.

However, there have been no reports of breeding gene chips, especially disease-resistant breeding gene chips, for aquaculture animals.

The present invention provides a gene chip that is used to select flounder with excellent disease resistance, solves the problem of lack of gene chips in the breeding of excellent fish breeds, makes up for the shortcomings of conventional breeding techniques, and improves fish disease resistance. The aim is to provide a new molecular breeding method, realize a generational change in fish breeding technology, and promote the rapid development of the fish breeding industry in order to develop excellent breeds that combine both high productivity and high productivity.

The present invention first provides a SNP locus associated with flounder disease resistance, wherein the SNP locus is the 36th base of any one of the sequences SEQ NO:1 SEQ ID NO:48697.

The SNP locus of the present invention can be used for selection of flounder with excellent disease resistance.

The SNP locus to which the present invention is provided can also be used to produce inspection products for selection of excellent disease-resistant flounder varieties.

Said test product is preferably a gene chip.

A further aspect of the present invention provides a gene chip used for selection of flounder disease-resistant superior cultivars, which can detect SNP locus associated with flounder disease resistance.

A further aspect of the invention provides a method for selecting disease-resistant flounder individuals, which method can be performed using the gene chip described above.

The method includes the following steps:
1) Extract the genomic DNA of individuals in the candidate group, and use the above-mentioned gene chip to detect and obtain genotyping results for the SNP markers.
2) From the SNP group in the reference group, the results of genotyping of SNP loci similar to the gene chip are extracted, and the results of genotyping of the SNPs in the reference group and genotypes obtained from the candidate group using the chip Merge with the judgment result.
3) Utilizing the combined genotype and the type represented by the reference group to estimate the estimated breeding value (GEBV) of the candidate group using the weighted GBLUP method, and based on the GEBV value the disease resistance of the tested individuals. Determines sexual potential.

The genotype of the reference group is used to estimate prediction accuracy using the weighted best linear unbiased estimator (weighted GBLUP). Among them, the 5-fold cross-validation method is used as the judgment method of prediction accuracy, and the area under the feature curve (AUC) is used as the judgment index of prediction accuracy. The closer the AUC is to 1, the higher the prediction accuracy.

The gene chip of SNP locus associated with flounder disease resistance provided by the present invention can be used for selection of flounder disease-resistant individuals, and the actual selection accuracy is close to the theoretical value. The selection accuracy of varieties can be improved and the breeding period can be shortened. As a result, we have provided gene chip technology for the selection of flounder cultivars with excellent disease resistance, and have opened up a new way of breeding using gene chips to select cultivars with excellent disease resistance in fish.

The purpose of the present invention is to establish a method for producing and applying a gene chip for breeding flounder with excellent disease resistance, and to provide new molecular breeding technical means for breeding flounder and other fish with excellent disease resistance.

Next, explanations for the terminology with which the present invention relates are as follows:
SNP: An abbreviation for Single Nucleotide Polymorphism, that is, a single nucleotide polymorphism, which is a DNA sequence polymorphism caused by a single nucleotide mutation at the genome level.

Gene chip: A two-dimensional DNA probe array consisting of tens of thousands to millions of specific DNA sequence fragments that are regularly arranged and fixed on a support such as a silicon wafer or slide through microfabrication technology. (DNA, etc.) can be subjected to genetic typing and molecular testing.

Degenerate bases: Based on the degeneracy of codons, one symbol is often used to replace two or more unspecified bases.

For example, R indicates A/G, Y indicates C/T, M indicates A/C, K indicates G/T, S indicates G/C, W indicates A/T, and so on.

Reference group: In genome selection, the group with phenotypic data acquired through testing such as artificial infection is selected from a large group with normal phenotypic characteristics. It is a set of individuals who perform the genotype data, and perform the actual genome selection calculation.

Candidate group: in genome selection, the candidate group is the group that has obtained genotype data through genome rearrangement but has no phenotypic data, and has the potential for breeding, and the subsequent actual superior varieties. This is a set of individuals scheduled to be used for breeding work.

GBLUP: An abbreviation for GenomicBestLinearUnbiasedPrediction, that is, a genome best linear unbiased estimator, which is a method for estimating genome breeding value by using high-density molecular markers in the genome and utilizing genetic relationships (G matrix) between individuals.

GEBV: Abbreviation for Genomic Estimated Breeding Values, ie Genomic Estimated Breeding Values, obtained by adding up the effect estimates of all markers or haplotypes in the full genome.

Next, the present invention will be described in detail by combining examples.

Example 1: Selection of "Fish Chip No. 1" Gene Chip SNP Locus and Chip Production 1. Creation of flounder-resistant Edwardsella tarda reference group and measurement of phenotypic characteristics Individuals of reference group and candidate group for selection of flounder genome All of them are derived from the flounder family created by this project team since 2003, and in the process of breeding for many years, the rapid growth of the flounder group in Korea, Japan and China, and excellent properties such as disease resistance and reversal resistance. be done.

In particular, from 2013, in response to the situation where Edwardsiella tarda in the flounder farming industry is becoming more serious day by day, we will conduct research on flounder-resistant Edwardsiella tarda family selection.

From 2013 to 2015, peritoneal inoculation artificial infection Edwardsiella tarda inspection was continuously carried out in flounder families created in the current year, fin spines were collected from the infected inspection fish seedlings, and growth and disease resistance phenotypes were measured. , 4577, 5942 and 6919 samples taken in 2013, 2014 and 2015 and used to select and generate a reference group from which the flounder Edwardian la tarda genome was selected.

From the infection test samples, 96 families (32 in 2013, 10 in 2014, 48 in 2015) were selected, and each family selected 10-15 individuals who died and survived in an equal proportion according to mortality. , a reference group from which the genome is selected is formed, and the results of infection detection (death or survival) of the selected individuals are used as the phenotypic characteristics of the reference group (Table 2).

2. Flounder full genome rearrangement and SNP locus evaluation The flounder reference group has a total of 931 individuals available after undergoing DNA extraction and testing (Table 3).

The genomic DNA of 931 reference group individuals was extracted, and after passing the DNA test, the next generation sequence library was constructed. and the average amount of data acquired in quality control is 2G/individual. Using the flounder genome sequence (GenBankID: PRJNA73673) provided by this project team as a reference genome, sequence comparison was performed using BWA (http://bio-bwa.sourceforge.net/) software, and then Samtools (http:/ /www.htslib.org/) software is used for SNP prediction and evaluation to obtain a SNP set of 42.2M.

3. SNP locus evaluation and selection (1) Select the 42.2M flounder SNP markers obtained in step 2 according to the following standards:
Deletion of sites with a miss rate >0.1, minimum allele frequency (MAF) <0.05, deletion of sites at or in repeats, deletion of sites that do not fit the Hadwin balance, deletion of 3.4M Acquire flounder SNP markers.

モデルにおいて、ｙは表現型値で、ｕはグループ平均値で、グループｑｉはマーカー効果が正規分布ｑｉ～Ｎ（０、

）で、ｍはマーカーの総数で、Ｘはｑｉに対応する関連マトリックスで、ｅは残差である。 (2) Perform SNP locus effect value analysis and estimated breeding value calculation for the 3.4M molecular markers selected in step (1),
Flounder genome selection calculation uses the Bayes C π algorithm, analysis model equation:

In the model, y is the phenotypic value, u is the group mean value, and group q is the marker effect normally distributed q ~ N(0,

), m is the total number of markers, X is the association matrix corresponding to qi, and e is the residual.

Using the BayesCπ algorithm provided with the R language pack, the combined genotype data genotype. csv and phenotype data phonetype. csv and perform genome selection calculations on a total of 931 flounder individuals in the reference group for full genome rearrangement. After that, the obtained SNP locus effect values are rearranged from the largest to the smallest, sites with estimated breeding values <10-5 are deleted, and a total of 864,229 SNP loci are obtained and used for selection of gene chip SNP locus.

(3) Further probe design and evaluation are performed for each of the SNPs selected in step (2) using the Affymetrix Axiom genetic classification probe design bioanalysis process, and sites with probe conversion feasibility evaluation scores <0.6 are deleted. In addition, the SNPs covered and evenly distributed throughout the genome, there were no other SNPs within the 35 bp flanking sequence of the SNP, the GC content of the 35 bp flanking sequence of the SNP was 30-70%, and the final 48,697 flounder SNP markers were screened to ensure their use in chip manufacturing, and the sequence records of said 48,697 SNP molecular markers are in the sequence listing.
A flounder SNP chip (gene chip) was manufactured using Affymetrix Axiom chip production technology manufactured by ThermoFisher, USA, containing a total of 48,697 flounder SNP loci, and each chip can detect 24 samples simultaneously.

Example 2, How to use "Fish Chip No. 1" gene chip 1. Manufacture and crossbreeding of samples for chip detection The fin spine genomic DNA was extracted using 1% agarose gel electrophoresis and nucleic acid spectrophotometer, and the final acceptable sample specifications were as follows: As follows: electrophoresis sought to produce a single strip of DNA, fragment length greater than 10 kb, excellent integrity, no biodegradation, sample quality DNA detection results. A260/280 is measured using UV spectrophotometer: 1.8-2.0, A260/230>1.5, concentration not lower than 20 ng/μl, total amount not lower than 4 μg. After that, according to the manufacturing standard operating process for SNP chip test samples manufactured by ThermoFisher, USA (https://www.thermofisher.com/), samples for chip test are manufactured. 1.4 ug or more of high-quality DNA template is added to a 2 ml*96 deep-well plate, a denaturant is added, denaturation is performed at room temperature, denaturation is stopped after denaturation for 10 minutes, and single-stranded DNA is obtained. 48697 pair primers for chip site amplification, isothermal amplification enzymes, dNTPs, etc. are added to the deep hole plate, the deep hole plate is sealed, and isothermal amplification is performed at 37°C. After 24 h of amplification, the amplified products are fragmented, added with an equal volume of isopropanol and precipitated in a refrigerator at -20°C. After precipitation for 24 h, centrifuge at 3000 g at 4° C. to obtain a precipitate of DNA product, remove the remaining isopropanol at 37° C., dissolve the precipitate, obtain a hybrid fluid, and the hybrid fluid is 5%. Using the mass of the gel electrophoresis detection results, the amplification product quality detection results show that the strips are clear and bright. The hybrid was hybridized using a temperature-controlled amplifier under the conditions of 95°C for 10 min and 48°C for 3 min, after which the hybrid was maintained at 48°C continuously. The chip block was immersed in a hybridization solution, hybridized for 24 hours in a hybridization oven at 48°C, then eluted, fluorescent proteins were connected, the fluorescent proteins were immobilized, hybridization probes were scanned, and each signal point was scanned. After obtaining the hybridization results for each site from the results of one probe hybridization, the chip scan results are analyzed using Axiom Analysis Suite (AxAS) software (manufactured by ThermoFisher, USA).

2. Chip inspection and gene classification data analysis (1) Collection of samples of the group to be tested and DNA extraction A part of rearranged flounder individual DNA is selected, the chip is used for inspection, and chip gene classification is performed. The accuracy and reproducibility of performing genome selection calculations on genotypic data obtained using rearrangement and gene chip classification are tested. Then, in the selection process of flounder breeding, establish candidate adult individuals whose pedigrees are used, perform genomic DNA extraction and chip detection, and verify the application effect of gene chips in flounder genome selection breeding. Information on the adopted individuals is shown in Table 4.

(2) Chip detection Following the standard process of gene chip detection, Affymetrix GeneTitan gene chip processing system is used to complete probe hybridization/staining and chip scanning. The specific operating method is as follows: add 4 μg of high-quality DNA template to a 2 ml*96 deep-well plate, add a denaturant to denature (28° C.), denature for 10 min, then denature stop solution ( (Reaction time not longer than 10 min) Stop denaturation and obtain single-stranded DNA. 48697 pair primers for chip site amplification, isothermal amplification enzyme/dNTP, reaction solution, etc. are added to the deep hole plate, the deep hole plate is sealed, and isothermal amplification is performed at 37° C. for 22 to 36 hours. Preferably, after amplification for 24 hours, the reaction mixture is inactivated at a high temperature of 65°C for 20-30 minutes, then transferred to a culture box at 37°C and cultured for 40 minutes. Fragment the amplified product, add an equal volume of isopropanol to the existing reaction mixture, mix the reaction mixture evenly until the reaction mixture is clear, and then place the product in a refrigerator at -20°C to precipitate. After 24 h of precipitation, the DNA product was precipitated by centrifugation at 3,000 g for 40-60 min at 4°C, the supernatant was removed, the precipitate was retained, and the remaining isopropanol was completely removed at 37°C. , to dissolve the precipitate and obtain the hybridization fluid. The hybrid was hybridized using a temperature-controlled amplifier under the conditions of 95°C for 10 min and 48°C for 3 min, after which the hybrid was maintained at 48°C continuously. The chip blocks are immersed in the hybridization solution and hybridized for 24 h in a hybridization oven at 48°C. After that, elution, fluorescent protein is attached, fluorescent protein is immobilized, hybridization probe and fluorescence signal are scanned to obtain the hybridization result of each site, and the chip scan result is obtained with Axiom Analysis Suite (AxAS) software (manufactured by ThermoFisher, USA). Analysis is performed using

(3) Data Analysis The chip scan results are analyzed using AxAS software (manufactured by ThermoFisher, USA) to obtain gene classification results for each sample. According to the analysis results, the average classification rate of the chips is 98.77%, which has excellent classification effect. Among them, the high-quality SNP ratio is 74.61%, and each sample can generate high-quality classification information.

Example 3 Application of “Fish Chip No. 1” Gene Chip to Disease Resistance Breeding of Flounder Used for validation, these selected individuals were both genotyped for rearrangement and genotyped using the "Fish Chip No. 1" gene chip classification. Classification and statistics are performed by selecting some individual gene chips from the inventor's existing flounder reference group. Consistency between genotypes obtained using a chip (0/1/2 indicates AA/Aa/aa) and genotypes obtained by rearrangement, and association between rearrangement and GEBV estimated using chip data Evaluate the effect of chip classification by statistical coefficients. A chip is considered to have a good classification result if the consistency of the classification result reaches 88% or more and the correlation coefficient between GEBV reaches 0.9 or more.

The analysis results show that 90.08% of the site information obtained using the “fish chip No. 1” gene chip classification is similar to the rearrangement, and the association coefficient between the two sets of GEBV is 0.958. Therefore, the classification result of the flounder gene chip developed by the present invention is basically consistent with the rearrangement, and accurate gene classification can be performed for flounder.

The specific operation method is as follows:
Use PLINK software to read the chip data, and enter the following command on the server to process the above data:
plink --vcf op2-1. vcf --make-bed --out op_Val_1
plink --vcf cs2-2. vcf --make-bed --out op_Val_2
plink --vcf op2-3. vcf --make-bed --out op_Val_3
plink --vcf op2-4. vcf --make-bed --out op_Val_4

After reading, the information in the 4 vcf is according to Table 5:

a) Rename the SNPs in R and extract the common marker information in the four files, the commands are:
# library (data.table) to unload the required R packages
# Read site info for cs_Val_1 and cs_Val_2 val_1 <- fread("op_Val_1.bim", header = F)
val_2 <- fread("op_Val_2.bim", header = F)
val_3 <- fread("op_Val_3.bim", header = F)
val_4 <-fread("op_Val_4.bim", header = F)
# Unify the SNP naming scheme and output a renamed file val_1$V2 <- paste (paste(rep(“rs”, nrow(val_1)), val_1$V1, sep = “ ”), val_1$V4, sep = “:”)
val_2$V2 <- paste(paste(rep("rs", nrow(val_2)), val_2$V1, sep = ""), val_2$V4, sep = ":")
val_3$V2 <- paste(paste(rep("rs", nrow(val_3)), val_3$V1, sep=""), val_3$V4, sep=":")
val_4$V2 <- paste(paste(rep("rs", nrow(val_4)), val_4$V1, sep=""), val_4$V4, sep=":")
write. table(val_1, "op_Val_1.bim", sep = "\t", col.names = F, row.names = F, quote = F)
write. table(val_2, "op_Val_2.bim", sep = "\t", col.names = F, row.names = F, quote = F)
write. table(val_3, "op_Val_3.bim", sep = "\t", col.names = F, row.names = F, quote = F)
write. table(val_4, "op_Val_4.bim", sep = "\t", col.names = F, row.names = F, quote = F)
# Extract and output common site information comm <- Reduce(intersect, list (a = val_1$V2, b = val_2$V2, c = val_3$V2, d = val_4$V2))
write. table(comm, "common_snps.txt", sep = "\t", col.names = F, row.names = F, quote = F)

b) Combine the 4 files using PLINK software and retain the common markers, the command is:
plink --bfile op_Val_1 --merge-list merge_op. txt--extract common_snps. txt --recode A --out op_chip
The following information is stored in the file "merge_op.txt":
op_Val_2. bed op_Val_2. bim op_Val_2. fam
op_Val_3. bed op_Val_3. bim op_Val_3. fam
op_Val_4. bed op_Val_4. bim op_Val_4. fam

c) Using PLINK software to extract similar individuals and sites from the reference group, . Organize the fam information into a file and name it "op_chip_indi.txt", where "..." represents the file catalog, the command is:
plink --bfile .../Val_ref --keep op_chip_indi. txt--extract common_snps. txt --recode A --out op_rseq
After processing, the number of markers that are common to the above four files is 11,719, and the number of individuals that can be retrieved in the reference group is 95.

d) Using the agreement of R statistic tip and rearrangement classification, the method is as follows:
# library (data.table) to unload the required R packages
# chip <- fread("op_chip.raw") to read the classification information obtained by chip and rearrangement respectively
rseq <-fread("op_rseq.raw")
# Unify the individual arrays in the file fid <- data. frame(rseq$FID)
colnames(fid) <- “FID”
chip <- data. table(merge(fid, chip, sort=F))
# chip[, c(1:6) := NULL] to delete the first 6 columns in the file and output the genotype
rseq[,c(1:6):= NULL]
fwrite(chip, "geno_op_chip.csv", sep = ",", row.names = F, quote = F)
fwrite(rseq, "geno_op_rseq.csv", sep = ",", row.names = F, quote = F)
# Sum(chip == rseq) / (nrow(chip) * ncol(chip)) * 100 to stat match
Statistics showed that the 95 individuals had a total of 1,113,305 markers and 1,002,829 markers that were completely similar. Therefore, the results of chip and rearrangement classification are 90.08% perfect match.

e) To ensure the accuracy of the GEBV estimation, genotype the remaining individuals in the reference group using PLINK software, the command is:
plink --bfile .../Val_ref --remove op_chip_indi. txt--extract common_snps. txt --recode A --out ref

f) Utilizing R to merge reference groups and verify individual genotypes, the method is as follows:
# library (data.table) to unload the required R packages
# read the classification information obtained by chip and rearrange respectively chip <- as. matrix(fread(“geno_op_chip.csv”))
rseq <- as. matrix(fread(“geno_op_rseq.csv”))
ref <-fread("ref.raw")
# Remove the first 6 columns in the file ref[, c(1:6) := NULL]
ref <- as. matrix (ref)
# Combine genotype files and output genotype file after combination geno_chip <- rbind(chip, ref)
geno_rseq <-rbind(rseq, ref)
write. table(geno_chip, "geno_Val_Chip.csv", sep = ",", row.names = F, quote = F)
write. table(geno_rseq, "geno_Val_Rseq.csv", sep = ",", row.names = F, quote = F)

g) the two xxx. A csv file estimates the GEBV using the weighted GBLUP method in R. The specific operation method is as follows (Linux environment):
# library (parallel) to unload the required R packages and functions
library(data.table)
library (asreml)
library (pROC)
source("ginv.R")
The definition of the function ginv is:
ginv <- function(invG) {
Ginv <- data. frame(row = rep(1: nrow(invG), nrow(invG)), column = rep(1: nrow(invG), each = nrow(invG)), value = as.numeric(invG), lower.mat = as.logical(lower.tri(invG,diag=T)))
Ginv <- Ginv [Ginv$lower. mat == T, c(“row”, “column”, “value”)]
Ginv <- Ginv [order(Ginv$row, Ginv$column), ]
return (Ginv)
}
# geno_chip to read genotype and phenotype information <- as. matrix(fread("geno_Val_Chip.csv", nThread=10))
geno_rseq <- as. matrix(fread("geno_Val_Rseq.csv", nThread=10))
pheno <- asreml. read. table("pheno_op_Val.csv", header=T, sep=",")
rnames <- as. matrix(fread(“pheno_op_Val.csv”)) [, 1]
M_1 <- geno_chip
M_2 <- geno_rseq
# Calculate the secondary coordinate gene frequency of each site pi_1 <- round(colSums(M_1)/(2*nrow(M_1)), 3)
pi_2 <- round(colSums(M_2)/(2*nrow(M_2)), 3)
# Construct a P matrix P_1 <- matrix(2*pi_1, byrow = T, nrow = nrow(M_1), ncol = ncol(M_1))
P_2<−matrix(2*pi_2, byrow=T, nrow=nrow(M_2), ncol=ncol(M_2))
# Construct Z matrix Z_1 <- as. matrix(M_1 - P_1)
Z_2 <- as. matrix(M_2-P_2)
# Build the terms of the equation numerator ZZt_1 <- do. call('rbind', mclapply(1: nrow(Z_1), FUN = function(x) {tcrossprod(Z_1[x, ], Zt_1)}, mc.cores = 20))
ZZt_2 <- do. call('rbind', mclapply(1: nrow(Z_2), FUN = function(x) {tcrossprod(Z_2[x, ], Zt_2)}, mc. cores = 20))
# Build the terms of the equation denominator denom_1 <- 2 * (sum(pi_1 * (1 - pi_1)))
denom_2<−2*(sum(pi_2*(1−pi_2)))
# Build the G matrix G_chip <- ZZt_1/denom_1
G_ rseq <- ZZt_2 / denom_2
diag(G_chip) <− diag(G_chip) + 0.01
diag(G_rseq) <− diag(G_rseq) + 0.01
# G matrix inversion invG_chip0 <- solve(G_chip)
invG_rseq0 <- solve(G_rseq)
# Ginv_chip <- ginv(invG_chip) which converts and outputs the G matrix inversion to a three-column format that ASReml can use
Ginv_rseq <- ginv(invG_rseq)
write. table(Ginv_chip, "Ginv_op_Val_Chip_at_iter_0.csv", sep = ",", row.names = F, quote = F)
write. table(Ginv_rseq, "Ginv_op_Val_Rseq_at_iter_0.csv", sep = ",", row.names = F, quote = F)
# Calculate GEBV attr(Ginv_chip, "rowNames") <- paste(rnames)
attr(Ginv_rseq, "rowNames") <- paste(rnames)
WGBLUP1 <-asreml(status ~ 1,
random = ~ giv(IID),
ginverse = list(IID = Ginv_chip),
rcov = ~units,
family = asreml. binomial (link = “logit”)
data = pheno,
maxiter = 100,
na. method. X = 'omit')
gebv01 <- coef(WGBLUP1)$random
write. table(gebv01, "GEBVs_at_iter_0_chip.csv", sep = ",", row.names = F, quote = F)

WGBLUP2 <-asreml(status ~ 1,
random = ~ giv(IID),
ginverse = list(IID = Ginv_rseq),
rcov = ~units,
family = asreml. binomial(link = “logit”),
data = pheno,
maxiter = 100,
na. method. X = 'omit')
gebv02 <- coef(WGBLUP2)$random
write. table(gebv02, "GEBVs_at_iter_0_rseq.csv", sep = ",", row.names = F, quote = F)

# Weight the GBLUP process chip part and use the for cycle and run 6 weighted iterations # Estimate the effect value of the marker Zt1 <- t(Z_1)
p1 <- do. call('rbind', mclapply(1: nrow(Zt1), FUN = function(x) {Zt1[x, ] %*% invG01}, mc. cores = 5))
p2 <- p1 %*% gebv01
u01 <- p2/pp_1
Varu01<-u01*u01*2*pi_1*(1-pi_1)
write. table(u01, "markerEff_chip_at_iter_0.csv", sep = ",", row.names = F, quote = F)
write. table(Varu01, "markerVar_chip_at_iter_0.csv", sep = ",", row.names = F, quote = F)
# Estimate the weight of each marker D <- rep(1, ncol(M_1))
D0 <- rep(1, ncol(M_1))
pre_D <- D0
u <-u01
cal <-6
inter <- 1
for(t in 1:cal) {
d <-as. vector(u*u*2*pi_1*(1-pi_1))
for(i in 1: (length(d)/inter)) {
di <−mean(d[(inter*(i−1)+1):(inter*i)])
pre_D[(inter * (i - 1) + 1): (inter * i)] <- di
}
D<−sum(D0)/sum(pre_D)*pre_D
write. table(D, file = paste("weights_chip_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
# Estimate the weighted G matrix p1 <- do. call('cbind', mclapply(1: ncol(Z_1), FUN = function(x) {Z_1[, x]*D[x]}, mc. cores = 5))
p2 <- do. call('rbind', mclapply(1: nrow(p1), FUN = function(x) {p1[x, ] %*% Zt1}, mc. cores = 5))
G<-p2/pp_1
write. table(G, file = paste("G_chip_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
diag(G) <− diag(G) + 0.01
invG <- solve(G)
write. table(invG, file = paste("invG_chip_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
Ginv<-ginv(invG)
write. table(Ginv, file = paste("Ginv_chip_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
attr(Ginv, "rowNames") <- paste(rnames)
WGBLUP <-asreml(status ~ 1,
random = ~ giv(IID),
ginverse = list(IID = Ginv),
rcov = ~units,
family = asreml. binomial(link = “logit”),
data = pheno,
maxiter = 100,
na. method. X = 'omit')
gebv <- coef(WGBLUP)$random
write. table(gebv, file = paste("GEBVs_chip_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
# Estimate the effect value of the weighted marker p1 <- do. call('rbind', mclapply(1: nrow(Zt1), FUN = function(x) {D[x] * Zt1[x, ]}, mc. cores = 5))
p2 <- do. call('rbind', mclapply(1: nrow(p1), FUN = function(x) {p1[x, ] %*% invG}, mc. cores = 5))
p3 <- do. call('rbind', mclapply(1: nrow(p2), FUN = function(x) {p2[x, ] %*% gebv01}, mc. cores = 5))
u<−p3/pp_1
Varu<−u*u*2*pi_1*(1−pi_1)
write. table(u, file = paste("markerEff_chip_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
write. table(Varu, file = paste("markerVar_chip_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
}
# GBLUP process - weight the rearrangement part, use the for cycle and perform 6 weighted iterations Zt2 <- t(Z_2)
p1 <- do. call('rbind', mclapply(1: nrow(Zt2), FUN = function(x) {Zt2[x, ] %*% invG02}, mc. cores = 5))
p2 <- p1 %*% gebv02
u02 <- p2/pp_2
Varu02<-u02*u02*2*pi_2*(1-pi_2)
write. table(u02, "markerEff_rseq_at_iter_0.csv", sep = ",", row.names = F, quote = F)
write. table(Varu02, "markerVar_rseq_at_iter_0.csv", sep = ",", row.names = F, quote = F)
# Estimate the weight of each marker D <- rep(1, ncol(M_2))
D0 <- rep(1, ncol(M_2))
pre_D <- D0
u <-u02
cal <-6
inter <- 1
for(t in 1:cal) {
d <-as. vector(u*u*2*pi_2*(1-pi_2))
for(i in 1: (length(d)/inter)) {
di <−mean(d[(inter*(i−1)+1):(inter*i)])
pre_D[(inter * (i - 1) + 1): (inter * i)] <- di
}
D<−sum(D0)/sum(pre_D)*pre_D
write. table(D, file = paste("weights_rseq_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
# Estimate the weighted G matrix p1 <- do. call('cbind', mclapply(1: ncol(Z_2), FUN = function(x) {Z_2[, x]*D[x]}, mc.cores = 5))
p2 <- do. call('rbind', mclapply(1: nrow(p1), FUN = function(x) {p1[x, ] %*% Zt2}, mc. cores = 5))
G<-p2/pp_2
write. table(G, file = paste("G_rseq_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
diag(G) <− diag(G) + 0.01
invG <- solve(G)
write. table(invG, file = paste("invG_rseq_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
Ginv<-ginv(invG)
write. table(Ginv, file = paste("Ginv_rseq_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
attr(Ginv, "rowNames") <- paste(rnames)
WGBLUP <-asreml(status ~ 1,
random = ~ giv(IID),
ginverse = list(IID = Ginv),
rcov = ~units,
family = asreml. binomial (link = "logit"),
data = pheno,
maxiter = 100,
na. method. X = 'omit')
gebv <- coef(WGBLUP)$random
write. table(gebv, file = paste("GEBVs_rseq_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
# Estimate the effect value of the weighted marker p1 <- do. call('rbind', mclapply(1: nrow(Zt2), FUN = function(x) {D[x]*Zt2[x, ]}, mc. cores = 5))
p2 <- do. call('rbind', mclapply(1: nrow(p1), FUN = function(x) {p1[x, ] %*% invG}, mc. cores = 5))
p3 <- do. call('rbind', mclapply(1: nrow(p2), FUN = function(x) {p2[x, ] %*% gebv02}, mc. cores = 5))
u<−p3/pp_2
Varu<−u*u*2*pi_2*(1−pi_2)
write. table(u, file = paste("markerEff_rseq_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
write. table(Varu, file = paste("markerVar_rseq_at_iter_", t, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
}
Through computation, the weighted GBLUP method approaches stability at the fourth iteration, so subsequent studies are performed on the iteration results at this time. The association coefficient among the 95 individuals GEBV used for validation was 0.958, and these individuals had GEBV according to Table 6:

2. "Fish chip No. 1" gene chip site is verified in the reference group The design site of "fish chip No. 1" gene chip is extracted from the existing reference group of the inventor, and weighted GBLUP is performed using these site information. and using a random grouping of a 5-fold cross-validation method as an assessment method for weighted GBLUP predictive accuracy, the subject's area under the operating characteristic curve (AUC) was measured as a weighted GBLUP assessment method for accuracy. as an indicator of The analytical model uses a generalized linear mixed model. To reduce the random error of grouping, the data set is grouped 10 times and each group is calculated 5 times. Therefore, a total of 50 calculations are performed, and the average value of the 50 AUCs is used as the final evaluation result.

According to the analysis results, the genome selection was performed using the same markers as the SNP chip in the flounder reference group, and the AUC (accuracy) value was 0.885, which was higher than the AUC (0.579) value by the conventional BLUP method. high. Therefore, using the chip site designed by the inventor, genome selection can be carried out smoothly and highly efficiently.

The specific operation method is as follows:
Chip design sites extracted from the flounder reference group using fcGENE, BEAGLE and PLINK software: fill in the missing sites and output the genotype file, the commands are as follows:
fcgene --ped geno_op_Rseq_chip. ped --map geno_op_Rseq_chip. map --format beagle --out plink2beagle

java -Xmx5120m -jar beagle. jar unphased=plink2beagle. bgl missing=0 niterations=20 gprobs=true out=imputed_geno

gunzip imputed_geno. plink2beagle. bgl. phased. gz

fcgene--bgl imputed_geno. plink2beagle. bgl. phased --pedinfo plink2beagle_pedinfo. txt --snpinfo plink2beagle_snpinfo. txt --format plink --out beagle2plink

plink --file beagle2plink --record A --out genotype_op_chip_Rseq
Then enter the following command in R:
library(data.table)
geno <-fread("genotype_op_chip_Rseq.raw")
geno[, c(1:6) := NULL]
fwrite(geno, "genotype_op_chip_Rseq.csv", sep = ",", row.names = F, quote = F)

a) Perform a weighted GBLUP in R using the genotype file obtained in a). A specific method of weighted GBLUP is performed with reference to part h) in 1).

b) Substitute the pre-constructed weighted G matrix into ASReml-R to perform the cross-validation method. Before the cross-validation method, it is necessary to perform grouping: in R, use the function sample(1:931, 931) to perform random sorting on all individuals, and further divide the digital after sorting into 5 columns, each column are 186, 186, 186, 186 and 187, respectively. The above process is repeated 10 times to obtain a total of 10 files. Place these 10 files in the same folder and prepare for use. The analysis uses a generalized linear mixed model, with different test lots and individual ages as fixed effects and each individual as a random effect for fitting. A specific implementation of the 5-fold cross-validation method is as follows:
# library (parallel) to unload the required R packages and functions
library (asreml)
library (pROC)
# Phenotype, pheno <- asreml. read. table("phenotype_931.csv", header=T, sep=",", na.string=NA)
ped <- asreml. read. table("pedigree_931.csv", header=T, sep=",", na.string=NA)
ainv <- asreml. Ainverse (ped) $ginv
Ginv<-fread(".../Ginv_at_iter_4.csv")
attr(Ginv, "rowNames") <- paste(pheno[, 1])
# Set the number of external cycles N <- 10
# Set result variables res <- matrix(NA, nrow = 5*N, ncol = 1)
colnames(res) <- c("auc")
# Perform cross-validation method and output validation result for (i in 1: N) {
coor <- read. table(file = paste("./coor/coor_", i, ".csv", sep = ""), sep = ",", header = F)
cyc <- ncol(coor)
gebv <- matrix(NA, nrow = nrow(pheno), ncol = cyc)
for (j in 1: cyc) {
y <- pheno
y$status[coor[(1: sum(coor[, j] > 0, na.rm = T)), j]] <- NA
CV<−asreml(status ~ Batch + Age,
random = ~ giv(AnimalID),
ginverse = list(AnimalID = Ginv),
rcov = ~units,
family = asreml. binomial(link = “logit”),
data = y,
maxiter = 50)

gebv[,j] <- coef(CV)$random
write. table(gebv, file = paste("GEBVs_chip_ref_coor_", i, ".csv", sep = ""), sep = ",", row.names = F, quote = F)
res[(j+(i-1)*cyc), ] <- roc(as.vector(pheno$status[coor[(1:sum(coor[, j]>0, na.rm=T)), j]]), as.vector(gebv[coor[(1: sum(coor[, j] > 0, na.rm = T)), j], j])) $auc
write. table(res, "results_of_roc_op_chip_ref.csv", sep = ",", row.names = F, quote = F)
}
}
# The conventional BLUP estimation method is similar to the weighted GBLUP, and we can use ainv to replace Ginv in the model The mean AUC by method is 0.579. The results of the 50-fold cross-validation method are according to Table 7:

で変換を実施する。変換後、平均値より高い家系の生存率を１と記載し、平均値より低いものを０と記載する。遺伝子チップ方法がマーカーに対する要件を満たし、すべての候補個体の分類結果に組み合わせを行い、各チップの分類結果を単独で参考グループ遺伝子型と合併する必要がある。 3. Application of "fish chip No. 1" gene chip in flounder disease resistance breeding ) is combined with the inventor's existing reference group genotypes, and R is used to construct a weighted G matrix, and finally the prepared weighted G matrix and phenotypic data are substituted into ASReml-R, weighted GBLUP The method is used to estimate the GEBV of each pedigree adult, and the mean of the adult GEBV is taken as the GEBV of the corresponding pedigree. Each family was classified according to infection survival rate, high survival rate families (>55% survival rate) and low survival rate families (less than 55% survival rate), and AUC values between each family's GEBV and infection survival rate. and further compare the AUC value with the AUC value obtained in 2, and if it is close or even higher than the AUC obtained in 2, the gene chip designed by the inventor meets the requirements of genome selection technology and flounder disease resistance It shows that it has excellent application effect in sorting. Calculate the AUC value between each pedigree GEBV and the infection survival rate and compare the AUC value with the AUC value obtained in 2, whereby the gene chip and genome selection technology designed by the inventors is used for flounder disease resistance screening. It is used to verify the actual application effect in Before estimating the AUC value, the infection survival rate of offspring was

to perform the conversion. After transformation, family survival rates above the mean are described as 1 and those below the mean as 0. The gene chip method should meet the requirements for markers, combine the classification results of all candidate individuals, and merge the classification results of each chip alone with the reference group genotype.

The results of the analysis showed that in the 16 candidate groups of progeny of offspring, 6 high survival families had a mean survival rate of 62.4% and 10 low survival families had a mean survival rate of 33.47% ( Table 8). Of which, the average GEBV for adults in high survival families is 2.10, and the average GEBV for adults in low survival families is 1.34. Calculations show that using the GEBV values of these flounder families, the accuracy of predicting their infection survival rate can reach 0.794, which is close to the theoretical value. Therefore, the gene chip designed by the inventors can be effectively applied to the screening of flounder for disease-resistant traits.

The specific operation method is as follows:
PLINK and R are used to use the candidate individual classification file, select the individual to be used for subsequent verification from the candidate individual, store this individual information in a single text file, and store the family number, individual number, and father book. A text file is prepared according to the sequence of number, parental book number, sex and phenotype value, one individual per line, and each item information of each individual is separated using a table delimiter. MODES FOR CARRYING OUT THE INVENTION:
# Read the classification result of each SNP chip, the result is xxx. bed, xxx. bim and xxx. Stored in fam.
plink --vcf op1. vcf --make-bed --out op1
plink --vcf op2. vcf --make-bed --out op2
plink --vcf op3. vcf --make-bed --out op3

Convert the SNP naming method in each file in R, the specific operation method is as follows:
# unload R package library(data.table)
# read data op1 <- fread("op1.bim", header = F)
op2 <-fread("op2.bim", header = F)
op3 <-fread("op3.bim", header = F)
# Convert SNP names op1$V2 <- paste (paste(rep(“rs”, nrow(op1)), op1$V1, sep = “ ”), op1$V4, sep = “:”)
op2$V2<-paste(paste(rep("rs", nrow(op2)), op2$V1, sep=""), op2$V4, sep=":")
op3$V2<-paste(paste(rep("rs", nrow(op3)), op3$V1, sep=""), op3$V4, sep=":")
# output the classification information acquired by each chip write. table (op1$V2, "snps_op1.txt", sep = "\t", col.names = F, row.names = F, quote = F)
write. table (op2$V2, "snps_op2.txt", sep = "\t", col.names = F, row.names = F, quote = F)
write. table (op3$V2, "snps_op3.txt", sep = "\t", col.names = F, row.names = F, quote = F)
# fwrite (op1, “op1.bim”, sep = “\t”, col.names = F, row.names = F, quote = F) to output the bim file after organization
fwrite(op2, "op2.bim", sep = "\t", col.names = F, row.names = F, quote = F)
fwrite(op3, "op3.bim", sep = "\t", col.names = F, row.names = F, quote = F)
# Exit R

Prepare verification individual information and put it in a file with the title "selected_indi.txt", and organize the file according to the following method:
1701. CEL 1701. CEL 0 0 0 -9
1702. CEL 1702. CEL 0 0 0 -9
# Using PLINK software to extract the individuals contained in each SNP chip, the method is as follows:
plink --bfile op1 --keep selected_indi. txt --make-bed --out op_can_1
plink --bfile op2 --keep selected_indi. txt --make-bed --out op_can_2
plink --bfile op3 --keep selected_indi. txt --make-bed --out op_can_3

# Output the individual genotype file used for verification plink --bfile op_can_1 --recode A --out op1
plink --bfile op_can_2 --recode A --out op2
plink --bfile op_can_3 --recode A --out op3
In addition, output the genotype file using the following command in R:
library(data.table)
op1 <-fread("op1.raw")
op2 <-fread("op2.raw")
op3 <-fread("op3.raw")
op1[, c(1:6) := NULL]
op2[, c(1:6) := NULL]
op3[, c(1:6) := NULL]
fwrite(op1, "geno_op1_can.csv", sep = ",", row.names = F, quote = F)
fwrite(op2, "geno_op2_can.csv", sep = ",", row.names = F, quote = F)
fwrite(op3, "geno_op3_can.csv", sep = ",", row.names = F, quote = F)
a) Use PLINK and R to combine the genotypes of the reference group with the genotypes of each chip candidate individual, and the specific operation method is as follows:
plink --bfile .../op_Ref --extract snps_op1. txt --recode A --out op_Ref_op1
plink --bfile .../op_Ref --extract snps_op2. txt --recode A --out op_Ref_op2
plink --bfile .../op_Ref --extract snps_op3. txt --recode A --out op_Ref_op3
Further process and output the genotype file in R using the following command:
library(data.table)
# Read the reference group genotype and perform processing on it op1_ref <- fread("op_Ref_op1.raw")
op2_ref <-fread(“op_Ref_op2.raw”)
op3_ref <-fread(“op_Ref_op3.raw”)
op1_ref[, c(1:6) := NULL]
op2_ref[, c(1:6) := NULL]
op3_ref[, c(1:6) := NULL]
op1_ref <- as. matrix (op1_ref)
op2_ref <- as. matrix(op2_ref)
op3_ref <- as. matrix(op3_ref)
# Process candidate individual genotypes: read, combine op1_can <- as. matrix(fread(“geno_op1_can.csv”))
op2_can <- as. matrix(fread(“geno_op2_can.csv”))
op3_can <- as. matrix(fread(“geno_op3_can.csv”))
geno_op1 <-rbind(op1_can, op1_ref)
geno_op2 <- rbind(op2_can, op2_ref)
geno_op3 <-rbind(op3_can, op3_ref)
write. table(geno_op1, "geno_op1.csv", sep = ",", row.names = F, quote = F)
write. table(geno_op2, "geno_op2.csv", sep = ",", row.names = F, quote = F)
write. table(geno_op3, "geno_op3.csv", sep = ",", row.names = F, quote = F)

b) Use the four genotype files processed in b) to construct a weighted G matrix respectively in R, and the construction method of the weighted G matrix is the same as described in 1).

c) Estimate the candidate individual's GEBV using ASReml-R, the code is:
# library (data.table) to unload the required R packages and functions
library (asreml)
###op1###
pheno <- asreml. read. table("phenotype_op1.csv", sep = ",", header = T)
Ginv<-fread(".../Ginv_op1.csv")
attr(Ginv, "rowNames") <- paste(pheno[, 1])
op1 <- asreml(status ~ Batch + Age, random = ~ giv (AnimalID), ginverse = list (AnimalID = Ginv), rcov = ~ units, family = asreml.binomial (link = "logit"), na.method. X = “omit”, data = pheno, maxiter = 50)
write. table(coef(op1)$random, "gebv_op1.csv", sep = ",", col.names = F, quote = F)

###op2###
pheno <- asreml. read. table("phenotype_op2.csv", sep = ",", header = T)
Ginv<-fread(".../Ginv_op2.csv")
attr(Ginv, "rowNames") <- paste(pheno[, 1])
op2 <- asreml(status ~ Batch + Age, random = ~ giv (AnimalID), ginverse = list (AnimalID = Ginv), rcov = ~ units, family = asreml.binomial (link = "logit"), na.method. X = “omit”, data = pheno, maxiter = 50)
write. table(coef(op2)$random, "gebv_op2.csv", sep = ",", col.names = F, quote = F)
###op3###
pheno <- asreml. read. table("phenotype_op3.csv", sep = ",", header = T)
Ginv<-fread(".../Ginv_op3.csv")
attr(Ginv, "rowNames") <- paste(pheno[, 1])
op3 <- asreml(status ~ Batch + Age, random = ~ giv (AnimalID), ginverse = list (AnimalID = Ginv), rcov = ~ units, family = asreml.binomial (link = "logit"), na.method. X = “omit”, data = pheno, maxiter = 50)
write. table(coef(op3)$random, "gebv_op3.csv", sep = ",", col.names = F, quote = F)

を用いて各家系の感染生存率に変換を行い、変換後の平均値より高い家系の生存率を１に設定し、平均値より低いものを０に設定する。最後に、各家系ＧＥＢＶと変換後の生存率の間のＡＵＣ値を計算し、ＡＵＣ値の計算方法は以下のとおりである：
＃必要なＲパッケージをアンロードする
ｌｉｂｒａｒｙ（ｄａｔａ．ｔａｂｌｅ）
ｌｉｂｒａｒｙ（ｐＲＯＣ）
＃各家系ＧＥＢＶおよび相応の感染生存率を一つのファイルに整理し、家系番号・ＧＥＢＶと変換後の感染生存率の配列に従って配列し、各行は一つの家系情報のみを含み、整理後のファイルを読み取る
ｒｅｓ ＜－ ｆｒｅａｄ（“…／ｇｅｂｖ＿ａｎｄ＿ｓｒ＿ｏｐ＿ｃａｎ．ｃｓｖ”）
＃子の代の家系生存率を変換する
ｒｅｓ＄ＳＲ＿ｔｒａｎｓ ＜－ ｅｘｐ（ｒｅｓ＄ＳＲ） ／ （１ ＋ ｅｘｐ（ｒｅｓ＄ＳＲ））
＃変換後の生存率を平均値に従ってさらに０と１に変換する
ＳＲ＿ｂｉｎａｒｙ ＜－ ｍａｔｒｉｘ（ＮＡ， ｎｒｏｗ ＝ ｎｒｏｗ（ｒｅｓ）， ｎｃｏｌ ＝ １）
ＳＲ＿ｂｉｎａｒｙ［ｗｈｉｃｈ（ｒｅｓ＄ＳＲ＿ｔｒａｎｓ ＞ ｍｅａｎ（ｒｅｓ＄ＳＲ＿ｔｒａｎｓ））， ］ ＜－ １
ＳＲ＿ｂｉｎａｒｙ［ｗｈｉｃｈ（ｒｅｓ＄ＳＲ＿ｔｒａｎｓ ＜ ｍｅａｎ（ｒｅｓ＄ＳＲ＿ｔｒａｎｓ））， ］ ＜－ ０
＃ＡＵＣ値を計算する
ｒｏｃ（ＳＲ＿ｂｉｎａｒｙ［， １］， ｒｅｓ＄ＧＥＢＶ）
計算を経て、１６のヒラメ家系のＧＥＢＶを取得し、計算は各家系ＧＥＢＶと相応の感染生存率の間のＡＵＣ（正確性）が０．７９４であることを示す。各家系ＧＥＢＶおよび感染生存率は表８による。 d) Calculate the corresponding pedigree GEBV according to the estimated GEBV of all candidate individuals, and formula

is used to convert the infection survival rate of each family, and the survival rate of families higher than the mean after conversion is set to 1, and the survival rate lower than the mean is set to 0. Finally, the AUC value between each pedigree GEBV and post-conversion survival rate was calculated, and the method for calculating the AUC value is as follows:
# library (data.table) to unload the required R packages
library (pROC)
# Organize each family GEBV and corresponding infection survival rate into one file, arranged according to the order of family number/GEBV and infection survival rate after conversion, each line contains only one family information, file after sorting read res <- fread(".../gebv_and_sr_op_can.csv")
# Translate pedigree viability res$SR_trans <- exp(res$SR) / (1 + exp(res$SR))
# Further transform the post-transformed viability to 0 and 1 according to the mean SR_binary <- matrix(NA, nrow = nrow(res), ncol = 1)
SR_binary[which (res$SR_trans > mean(res$SR_trans)), ] <- 1
SR_binary[which (res$SR_trans < mean(res$SR_trans)), ] <- 0
# Calculate AUC value roc(SR_binary[, 1], res$GEBV)
Through calculation, the GEBV of 16 flounder families are obtained, and the calculation shows that the AUC (accuracy) between each family's GEBV and the corresponding infection survival rate is 0.794. Each kindred GEBV and infection survival rate are according to Table 8.

候補群における１６の子の代家系を感染生存率に従って６の高生存率家系（平均的な生存率が６２．４％である）と１０の低生存率家系（平均的な生存率が３３．４７％である）の二大類（表８）に分け、高生存率と低生存率家系の成体のＧＥＢＶを比較し、高生存率家系成体の平均的なＧＥＢＶが２．１０で、低生存率家系成体の平均的なＧＥＢＶが１．３４であることがわかる。計算はよると、これらのヒラメ家系のＧＥＢＶ値を使用してはその感染生存率が予測される正確性が０．７９４に達することができ、理論値に近い。そのため、発明者が設計される遺伝子チップはヒラメ耐病性性状の選別に効果的に応用することができる。

The 16 offspring in the candidate group were divided into 6 high-survival families (with an average survival rate of 62.4%) and 10 low-survival families (with an average survival rate of 33.4%) according to infection survival rate. 47%) were divided into two categories (Table 8) and compared the GEBV of adults in high-survival and low-survival families. It can be seen that the average GEBV for an adult family is 1.34. Calculations show that using the GEBV values of these flounder families, the accuracy of predicting their infection survival rate can reach 0.794, which is close to the theoretical value. Therefore, the gene chip designed by the inventors can be effectively applied to the screening of flounder for disease-resistant traits.

According to the above results, the “Fish Chip No. 1” gene chip was used to perform genetic classification on the individuals of the flounder candidate group, the weighted GBLUP was used to calculate the genomic breeding value (GEBV), and the flounder according to the magnitude of the GEBV number Disease-resistant parent fish are selected, and the infection-resistant survival rate of progeny seedlings grown using these parent fish is remarkably increased. It turns out that it is possible to execute

Feasibility in Industry The gene chip of the SNP locus associated with flounder disease resistance provided by the present invention can be used for selection of flounder disease-resistant individuals, and the actual selection accuracy is close to the theoretical value. Increase the accuracy of selection of excellent disease-resistant varieties and shorten the breeding period, thereby providing gene chip technology for selecting excellent disease-resistant flounder varieties and gene-chip breeding for selecting excellent disease-resistant fish varieties. can open up new avenues for

Claims (2)

A method for selecting disease-resistant individuals of flounder, wherein the disease related to disease resistance is Edwardziellasis, the method is detected using a gene chip, and the gene chip is a set of SNP locus associated with disease resistance of flounder and the set of SNP loci includes 48697 SNP loci, and the 48697 SNP loci are in 48697 sequences whose sequences are SEQ ID NO: 1-SEQ ID NO: 48697, A method for selecting a disease-resistant individual of Japanese flounder, characterized in that it is the 36th base of each sequence . The method for selecting disease-resistant flounder individuals according to claim 1 ,
The method includes:
1) A step of extracting individual genomic DNA in the candidate group and performing detection using the gene chip to obtain results of genotyping of SNP markers;
2) From the SNP set of the reference group, the results of genotyping of SNP loci similar to those of the gene chip are extracted, and the results of genotyping of the SNPs of the reference group and genotypes of the candidate group obtained using the chip are obtained. merging the result of the determination;
3) Using the combined genotype and the type represented by the reference group, the weighted GBLUP method is used to estimate the estimated breeding value GEBV of the candidate group, and the disease resistance potential of the tested individuals based on the GEBV value. determining capabilities;
including
and using the genotype of the reference group to estimate the prediction accuracy using the weighted best linear unbiased estimation method,
Among them, a 5-fold cross-validation method is used in estimating prediction accuracy, and the area under the characteristic curve AUC is used as an index for determining prediction accuracy, and the closer the AUC is to 1, the higher the prediction accuracy. A method for selecting disease-resistant flounder individuals.