SNP molecular marker combination for breeding high-puffed corn and application thereof

Abstract

This invention discloses a SNP molecular marker combination for breeding high-expansion-multiplicity popcorn and its application, belonging to the field of molecular breeding technology. The SNP molecular marker combination of this invention consists of 30 high-effect SNP loci located on the maize B73 reference genome v4 version. This combination was selected by performing genome-wide association analysis on 399 natural popcorn accessions and ranking the top 30 loci according to their SNP effect values. This invention allows for accurate prediction of the expansion-multiplicity trait by detecting the genotypes of these 30 SNP markers in the early stages of breeding, with a prediction accuracy of over 0.72. It has significant advantages such as high selection efficiency, short cycle, low cost, and insensitivity to environmental influences, and is suitable for molecular marker-assisted breeding and multi-trait aggregation breeding of popcorn.

Description

Technical Field

[0001] This invention belongs to the field of molecular breeding technology, specifically a combination of SNP molecular markers for selecting high-expansion-multiplication-rate popcorn and its application. Background Technology

[0002] Popcorn ( Zea mays L.var.everta Sturt. Popcorn is a special type of corn that expands rapidly when heated under suitable moisture conditions, forming a butterfly-shaped or spherical puffed structure. It is the sole raw material for producing popcorn. Therefore, the puffing characteristic is the core trait that defines popcorn and determines its market value, and it is also the primary goal of quality breeding for popcorn.

[0003] Popping characteristics are a comprehensive trait, with popping volume (usually referring to the volume of a fixed number of kernels after popping, measured in milliliters per kernel) considered a key indicator of quality. Because of its importance, efficient breeding of popping maize has become central to breeding efforts. While the conventional breeding process for popping maize is similar to that for other maize types, it must prioritize ensuring superior quality traits such as popping volume while considering yield and resistance. It is worth noting that the infiltration of non-popping maize germplasm often adversely affects popping characteristics, further increasing the complexity of quality breeding.

[0004] However, traditional breeding methods rely heavily on breeders' experience and post-harvest phenotyping. Field trial results are highly volatile and unstable due to interannual and locational environmental factors, directly leading to low accuracy and efficiency in selection. The root of this phenotypic identification dilemma lies in its complex genetic basis: the bursting trait is controlled by numerous minor and a few major QTLs, resulting in complex genetic effects that are significantly influenced by the environment and the interaction between genotype and environment.

[0005] Although molecular-assisted selection is considered an ideal approach to solving the above problems, existing technologies still have significant limitations: First, reliable molecular markers closely linked to key traits such as popcorn volume and popping maize varieties are still scarce; second, most reported markers are population-specific, lacking accuracy and universality in predictive materials with different genetic backgrounds; and finally, existing research mostly focuses on the association analysis between markers and traits, failing to effectively integrate information from multiple loci to construct practical predictive models that can be used for precise, high-throughput screening in the early stages of breeding.

[0006] Genome-wide association studies (GWAS) are an effective method for revealing the complex genetic basis of crop trait variation and have been widely applied to various crops. Maize possesses high levels of genetic diversity and contains rare alleles in its genome, making it highly suitable for GWAS studies of the genetic structure of kernel-related traits. Utilizing GWAS to elucidate the genetic basis of maize kernel-related traits is of great significance for crop improvement. Summary of the Invention

[0007] In view of this, the purpose of the present invention is to provide a combination of SNP molecular markers for breeding high-expansion-multiplicity popcorn, wherein the combination of SNP molecular markers can efficiently characterize genetic information related to expansion multiple.

[0008] To achieve the above-mentioned objectives, the present invention provides the following technical solution:

[0009] This invention provides a combination of SNP molecular markers for breeding high-expansion-fold popping maize. The combination consists of 30 SNP loci located in the maize B73 reference genome version Zm-B73-REFERENCE-GRAMENE-4.0, as shown in the table below:

[0010]

[0011]

[0012] Preferably, the 30 SNP loci are selected by genome-wide association analysis of a natural population of popcorn, based on the association effect values ​​between the SNP loci and the popping fold trait, and the top 30 loci with the highest effect values ​​are selected.

[0013] The present invention also provides a molecular probe assembly for specifically recognizing the SNP molecular marker assembly.

[0014] The present invention also provides an application of the SNP molecular marker combination or the molecular probe combination described herein in any of the following:

[0015] (1) Construct a prediction model for the expansion ratio of popcorn kernels;

[0016] (2) Molecular marker-assisted breeding of high-expansion-explosion-ratio popcorn;

[0017] (3) Multi-trait aggregation breeding of popcorn.

[0018] This invention also provides a method for constructing a prediction model for the expansion fold of popcorn kernels, comprising the following steps: obtaining phenotypic data of kernel expansion fold from multiple popcorn breeding materials; detecting the genotypes of the 30 SNP molecular markers in the multiple popcorn breeding materials; and constructing a prediction model for the expansion fold using the phenotypic data and genotype data through a genomic selection statistical model.

[0019] Preferably, the plurality of popcorn breeding materials comprises no less than 300 popcorn inbred lines.

[0020] Preferably, the phenotypic data are the best linear unbiased predictions corrected through repeated trials in multiple environments.

[0021] The present invention also provides a prediction model for the expansion ratio of popcorn kernels, which is constructed by the method described above.

[0022] The present invention also provides an application of the aforementioned prediction model in predicting the expansion ratio of popcorn kernels or in assisting in the breeding of popcorn inbred lines with high expansion ratios.

[0023] This invention also provides a method for predicting the expansion ratio of popcorn kernels, comprising the following steps: detecting the genotype of the SNP molecular marker combination in the popcorn sample to be tested; inputting the genotype data into the prediction model to calculate the estimated genomic breeding value of each sample; sorting the samples according to the estimated genomic breeding value, and predicting that the sample with the higher breeding value has a higher kernel expansion ratio.

[0024] Compared with the prior art, the present invention has the following advantages:

[0025] (1) This invention uses genome-wide association analysis (GWAS) to precisely locate key genetic variations controlling fold expansion from 327,993 SNPs in the whole genome, and selects the top 30 SNPs with the highest genetic effect values ​​to construct a core marker set. The genome-wide selection model constructed using this marker set shows an average prediction accuracy of up to 0.72 in five-fold cross-validation. This demonstrates that the marker combination selected in this invention can efficiently and accurately capture the genetic essence of the target trait.

[0026] (2) The 30 SNP markers protected by this invention were not randomly selected, but were all derived from GWAS analysis of 399 broad-based popcorn natural populations and were selected through rigorous statistical screening, giving the marker combination a solid biological basis. This combination originates from a diverse population and, compared to markers developed for a single population, has stronger versatility and stability in popcorn breeding materials with different genetic backgrounds, overcoming the problem of poor universality of existing markers and facilitating large-scale application in breeding projects.

[0027] (3) Traditional methods require waiting for grains to mature before conducting physical expansion tests, resulting in a breeding cycle of over a year. Furthermore, these methods are highly susceptible to environmental variations and have low selection accuracy. This invention transforms phenotypic selection into genotypic selection, which is not limited by the growing season or location and can be conducted at any time and place. This allows for multiple rounds of selection within a single year's breeding cycle, significantly accelerating the accumulation of genetic gains. Simultaneously, genotypic testing offers high throughput and relatively fixed costs, avoiding the enormous costs associated with large-scale field trials and phenotypic determination, thus greatly reducing breeding costs and improving selection intensity and efficiency. Attached Figure Description

[0028] Figure 1 The figure shows the distribution of the PV trait in 399 natural maize accessions. Figure A is the original box plot of the kernel expansion ratio (PV) phenotypic value of the 399 natural maize accessions; Figure B is the box plot of the distribution after inverse normal transformation of the PV phenotypic BLUP value.

[0029] Figure 2 This is a distribution density map of 327,993 SNPs used for GWAS analysis on the ten chromosomes of maize in an embodiment of the present invention;

[0030] Figure 3 This is a QQ plot of a genome-wide association study (GWAS) of the popcorn kernel expansion fold in an embodiment of the present invention;

[0031] Figure 4 A comparison of the prediction accuracy of genome-wide selection models constructed using different numbers of SNP markers. Detailed Implementation

[0032] This invention provides a combination of SNP molecular markers for breeding high-expansion-fold popping maize. The combination consists of 30 SNP loci located in the maize B73 reference genome version Zm-B73-REFERENCE-GRAMENE-4.0, as shown in the table below:

[0033]

[0034]

[0035] In this invention, the download address for the maize B73 reference genome version is as follows: https: / / download.maizegdb.org / Zm-B73-REFERENCE-GRAMENE-4.0 / Zm-B73-REFERENCE-GRAMENE-4.0.fa.gz.

[0036] In this invention, the 30 SNP loci are selected by genome-wide association analysis of a natural population of popcorn, based on the association effect values ​​between the SNP loci and the popping fold trait, and the top 30 loci with the highest effect values ​​are selected.

[0037] In a specific embodiment of the present invention, the screening process for the 30 SNP molecular marker combinations used for genome-wide selection is as follows:

[0038] (1) The optimal linear unbiased prediction value for each material was calculated using mixed linear models based on grain expansion ratio (PV) data from at least three independent environments and at least three biological replicates for each environment, and an inverse normal transformation was performed to satisfy the analytical assumptions. The 399 maize inbred lines of this invention were planted at the Zhuangxing Base of the Shanghai Academy of Agricultural Sciences in the spring of 2020 and 2021, and at the Lingshui Base of the Shanghai Academy of Agricultural Sciences in Hainan in the winter of 2021.

[0039] (2) The natural population was subjected to simplified genome sequencing, and after standard bioinformatics procedures, a genotype matrix of 327,993 high-quality, biallelic SNP markers covering the entire maize genome was obtained.

[0040] (3) Using the FarmCPU model in the GAPIT software package (version 3.0), the PV-BLUP values ​​after the inverse normal transformation were used as the target trait and subjected to genome-wide association analysis with the genotype data of 327,993 SNPs. To control for population structure, the kinship matrix calculated based on genome-wide SNPs was included as a covariate in the model. The analysis output an association effect value and its corresponding statistical significance P-value for each SNP locus. The effect value is usually expressed as a Beta value or equivalent effect size, reflecting the average direction and magnitude of the difference between different alleles of the SNP locus and the phenotypic value of the target trait.

[0041] (4) Screening Criteria and Procedures for Core Markers After completing the GWAS analysis, core markers were screened according to the following steps: a) The effect values ​​(Beta values) of all 327,993 SNP loci in the GWAS analysis results were extracted. The effect values ​​were then sorted in descending order of absolute value. The larger the absolute value of the effect value, the greater the genetic contribution of the locus to the ploidy trait. b) From the above ranking list, the top 30 SNP loci were selected. These 30 loci are the core SNP molecular marker combinations with the highest effect values ​​described in this invention. The above screening criteria do not rely on artificially set significance thresholds (P-values), but are based entirely on the objective ranking and selection of the SNP's predictive contribution to the trait (effect value size). This method aims to maximize the explanatory power and predictive efficiency of the selected marker combinations for trait genetic variation, thereby constructing an efficient genome-wide selection model.

[0042] The present invention also provides a molecular probe assembly for specifically recognizing the SNP molecular marker assembly.

[0043] The present invention also provides an application of the SNP molecular marker combination or the molecular probe combination described herein in any of the following:

[0044] (1) Construct a prediction model for the expansion ratio of popcorn kernels;

[0045] (2) Molecular marker-assisted breeding of high-expansion-explosion-ratio popcorn;

[0046] (3) Multi-trait aggregation breeding of popcorn.

[0047] This invention also provides a method for constructing a prediction model for the expansion fold of popcorn kernels, comprising the following steps: obtaining phenotypic data of kernel expansion fold from multiple popcorn breeding materials; detecting the genotypes of the 30 SNP molecular markers in the multiple popcorn breeding materials; and constructing a prediction model for the expansion fold using the phenotypic data and genotype data through a genomic selection statistical model.

[0048] In a specific embodiment of the present invention, the steps of constructing the prediction model are as follows:

[0049] Step 1: Obtain high-quality, stable phenotypic data

[0050] 1. Materials and Experimental Design: Select no fewer than 300 popcorn breeding materials with diverse genetic backgrounds as the training population for model construction. This population should be field-tested in at least two independent planting seasons or geographical locations, using a completely randomized block design with at least two biological replicates.

[0051] 2. Phenotypic determination and treatment: At the physiological maturity of the grains, the grain expansion ratio (PV) of each material in each environment and each replicate was determined according to standard methods.

[0052] 3. Calculate the genetic estimate: Analyze the obtained raw phenotypic data from multiple environments and replicates using a mixed linear model. The lme4 package in R or specialized software such as Meta-R is recommended. Using genotype as a fixed effect and environment and within-environment replicates as random effects, calculate the best linear unbiased predictor value (BLUP) for each material. This BLUP value serves as the target variable (y) for subsequent modeling, minimizing environmental errors and representing the inherent genetic potential of the material.

[0053] Step 2: Detect the genotype of the training population at the core SNP loci.

[0054] 1. Genotyping Technique Selection: Genotyping is performed on all materials in the training population targeting the 30 core SNP loci. Targeted genotyping techniques, including but not limited to KASP, can be used.

[0055] 2. Liquid-phase chip: Design specific probes to simultaneously detect all 30 SNPs in a single reaction.

[0056] 3. Targeted sequencing: Amplicon sequencing of genomic regions containing these sites.

[0057] 4. Data Generation and Formatting: The test results are organized into a genotype data matrix (X). The rows of the matrix correspond to each material, and the columns correspond to the 30 SNP loci. Genotypes are usually encoded as allele counts or represented as two alleles.

[0058] Step 3: Construct a predictive model using a genomic selection statistical model

[0059] 1. Model Selection and Software Implementation: A statistical model of genome-wide selection is used to establish the mathematical relationship between genotype (X) and phenotype (y, i.e., BLUP value). In this embodiment of the invention, the ridge regression optimal linear unbiased prediction model is preferably used, which can be easily implemented using the rrBLUP package in R language.

[0060] 2. Specific Modeling Operations: Use the BLUP value vector (y) obtained in Step 1 and the genotype matrix (X) obtained in Step 2 as input data. Call the kin.blup function in the rrBLUP package or a similar core function. Typically, use the default parameters or set the following key parameters: K=NULL (to allow the function to calculate the kinship matrix based on the genotype data), geno=X, pheno=y. After running the model, the software will output a vector containing the effect value of each SNP locus, as well as the possible model intercept. This set of effect value vectors (or equivalent prediction functions) constitutes the final prediction model.

[0061] To evaluate model performance, cross-validation can be used. Five-fold cross-validation is preferred, where the training population is randomly divided into five groups, and four groups are used in turn to model and predict the remaining group. The correlation coefficient (prediction accuracy) between the predicted and actual BLUP values ​​is then calculated. This step does not change the final released model but is used to demonstrate its effectiveness.

[0062] The present invention also provides a prediction model for the expansion ratio of popcorn kernels, which is constructed by the method described above.

[0063] This invention also provides an application of the aforementioned prediction model in predicting the expansion ratio of popcorn kernels or in assisting in the breeding of high-expansion-ratio popcorn inbred lines. In actual breeding, the process of applying the prediction model of this invention is as follows:

[0064] (1) Genotyping: DNA was extracted from the leaves of seedlings of candidate materials. KASP genotyping technology was used to perform high-throughput, low-cost genotyping of the 30 core SNP loci described in this invention.

[0065] (2) Phenotypic prediction: The genotype data matrix of the candidate materials is obtained and directly input into the whole genome selection prediction model determined in this invention.

[0066] (3) Calculation and ranking: Run the model and the software will automatically output the estimated breeding value of the genome for each material. Then, rank all candidate materials from highest to lowest according to the breeding value.

[0067] (4) Assisted selection decision: Based on the selection intensity in conventional breeding practices (such as the top 20%), the top 20 materials are selected as the preferred strains predicted to have the highest genetic potential for expansion multiples in this round of breeding.

[0068] (5) Result Verification: To objectively evaluate the effectiveness of this model-assisted selection, traditional grain expansion fold phenotypic determination was subsequently performed on all candidate materials. The expected verification results are as follows: The average expansion fold of the top 20% of the preferred materials selected by the model was significantly higher than that of the remaining materials. In addition, the ranking of the actual expansion fold within the top 20% of materials showed a highly consistent order with the ranking of the breeding values ​​predicted by the model. The method provided by this invention can be directly and effectively applied to actual breeding, achieving early and accurate selection without relying on phenotypic determination, and greatly improving breeding efficiency.

[0069] This invention also provides a method for predicting the expansion ratio of popcorn kernels, comprising the following steps: detecting the genotype of the SNP molecular marker combination in the popcorn sample to be tested; inputting the genotype data into the prediction model to calculate the estimated genomic breeding value of each sample; sorting the samples according to the estimated genomic breeding value, and predicting that the sample with the higher breeding value has a higher kernel expansion ratio.

[0070] The technical solutions provided by the present invention will be described in detail below with reference to the embodiments, but they should not be construed as limiting the scope of protection of the present invention.

[0071] Unless otherwise specified, the experimental methods used in the following embodiments are conventional methods. Unless otherwise specified, the experimental materials used in the following embodiments are commercially available products.

[0072] Example 1: Phenotypic Identification of Natural Population Grains

[0073] 1. Breeding population materials and field trial design

[0074] This invention constructed a natural population comprising 399 genetically diverse pop maize inbred lines, covering current mainstream breeding materials and representative germplasm resources. To obtain stable and reliable phenotypic data, the population was planted in the spring of 2020 and spring of 2021 at the Zhuangxing Comprehensive Experimental Station of the Shanghai Academy of Agricultural Sciences, and in the winter of 2021 at the Lingshui Nanfan Base in Hainan Province, encompassing three independent environments. Each environment employed a completely randomized block design, with each material planted in a single plot containing two rows, each 3 meters long. All materials underwent routine field management and were strictly self-pollinated using bagging to ensure genetic uniformity. Each environment was replicated three times.

[0075] 2. Grain expansion ratio phenotypic determination and data analysis

[0076] After the kernels reached physiological maturity, three uniformly grown ears of kernels were randomly selected from each material in each plot and air-dried indoors to a uniform moisture content (approximately 14%). Each ear of kernels was popped using a commercial popcorn machine under standard conditions (e.g., 220°C, constant time), and the total volume of the popped product was accurately measured. The popped volume (PV) was calculated using the following formula: PV = Total volume after popping (mL) / Total number of kernels used. To eliminate environmental errors and obtain stable genetic estimates for individual materials, a mixed linear model was used to analyze phenotypic data from multiple environments and replicates. Using the lme4 package in R, with genotype as a fixed effect and environment and intra-environment replicates as random effects, the best linear unbiased prediction value of PV for each material was calculated. This BLUP value represents the genetic potential after excluding environmental interference.

[0077] like Figure 1 As shown in Figure A, the BLUP values ​​of the population PV phenotype exhibit a skewed distribution. To meet the assumptions regarding data distribution for subsequent genome-wide association studies (GWAS), the BLUP values ​​of PV were subjected to an inverse normalization transformation using R language. The transformed data conforms to a normal distribution. Figure 1 B). The minimum PV in the population was 1.26, the maximum was 25.54, and the average was 4.71. The heritability of the PV trait was 0.55, indicating that the PV trait is a quantitative trait with high heritability.

[0078] Example 2 Genotyping of a Natural Population

[0079] 1. Simplified whole-genome sequencing and development of high-quality SNP marker sets

[0080] High-molecular-weight genomic DNA was extracted from the leaves of seedlings from each material using the CTAB method. Quality-tested DNA samples were sent to BGI Genomics Co., Ltd. (https: / / www.bgi.com / ) in Shenzhen for double-digestion and simplified genotyping using MseI and EcoRI enzymes. The obtained raw sequencing data were processed using the following standard bioinformatics workflow to ensure a high-quality, reproducible genotyping dataset:

[0081] (1) Data quality control: The original sequence was filtered using the fastp software (version 0.23.2) to remove low-quality (Q<20) bases, adapter contamination and too short reads.

[0082] (2) Sequence alignment: The mem algorithm of BWA software (version 0.7.17) was used to align the filtered sequences to the maize reference genome b73 (version: v4, download address: https: / / download.maizegdb.org / Zm-B73-REFERENCE-GRAMENE-4.0 / Zm-B73-REFERENCE-GRAMENE-4.0.fa.gz).

[0083] (3) Marking PCR repeats: Use the MarkDuplicates function of the Picard toolkit (version 2.26.10) to identify and mark repeat sequences introduced by PCR amplification.

[0084] (4) Mutation detection: Use the HaplotypeCaller module of GATK software (version 4.2.6.1) to perform population SNP calling and generate the original mutation set.

[0085] (5) Strict filtering: The original SNPs were filtered using bcftools software (version 1.15.1) with the following filtering criteria: a) Only biallelic SNP sites were retained; b) Sites with a minor allele frequency greater than 0.05 were retained to ensure site polymorphism; c) Sites with a missing rate of less than 0.2 were retained to ensure data integrity.

[0086] After the above-described objective and repeatable standardized filtering process, a high-quality, genome-wide distributed set of 327,993 SNP markers was finally obtained. The distribution density of these SNPs on the 10 maize chromosomes is shown in [link to relevant documentation]. Figure 2 .

[0087] Example 3: Screening of Genome-wide Association Analysis (GWAS) and Core High-Effect SNP Marker Combinations

[0088] Genome-wide association analysis was performed based on the PV phenotype BLUP values ​​obtained in Example 1 and the genome-wide SNP genotype data obtained in Example 2.

[0089] 1. GWAS Analysis

[0090] The BLUP value of the PV phenotype obtained after inverse normal transformation in Example 1 was used as the target trait, and a genome-wide association analysis was performed with the genotype matrix of 327,993 SNPs. The analysis was performed using the FarmCPU model in the GAPIT software package (version 3.0). To control for false positives caused by population structure, the kinship matrix calculated based on genome-wide SNPs was included as a covariate in the model. The Manhattan plot shows the significance level of the association between SNPs and traits (…). Figure 3 ).

[0091] 2. Determination of Core SNP Marker Combinations

[0092] To achieve efficient and accurate genome-wide selection, this invention does not directly use all significant loci or all SNPs. Instead, based on GWAS analysis results, it proposes a marker screening strategy guided by genetic effect values. Specifically, the effect values ​​of all 327,993 SNP loci in the GWAS analysis are extracted. These SNP loci are then sorted in descending order according to the absolute value of their effect value (Beta value). The top 30 SNP loci in the sorting results are defined as the core SNP molecular marker combination claimed in this invention (Table 1). This combination concentrates the informational loci with the highest effect values ​​controlling ploidy fold variation.

[0093] Table 1 Information on 30 core SNP sites

[0094]

[0095] Example 4: Construction and Cross-Validation of a Genome-Wide Selection Prediction Model

[0096] To conduct rigorous model building and performance evaluation, 393 materials with complete phenotypic data were selected from the original natural population of 399 popcorn materials described in Examples 1 and 2 to form the model training and validation population. The remaining 6 materials were not included in this stage of analysis due to missing phenotypic data, to ensure the reliability of model training and validation. The purpose of this population was to objectively evaluate the prediction model based on core SNP markers through within-group cross-validation.

[0097] 1. Prediction Model Construction

[0098] The 393 popcorn inbred lines described in Example 1 (a subset of 399 natural populations with complete phenotypic and genotypic data) were used as the research material in this example. Each material contained:

[0099] (1) Target phenotypic data (y): the best linear unbiased prediction of grain expansion ratio obtained from Example 1 based on repeated multi-environment experiments.

[0100] (2) Core genotype data (X): that is, the genotype coding matrix of the 30 core SNP loci finally determined in Example 3.

[0101] 2. Five-fold cross-validation experimental design

[0102] To evaluate the robustness of the model and its ability to predict the phenotypes of unknown materials, a repeated random sampling-cross-validation strategy was adopted. The specific process is as follows:

[0103] (1) Dividing the training set and validation set: At the beginning of each round of validation, 314 materials (about 80%) are randomly selected from the above 393 model training and validation sets as the training set, which is used to build the prediction model for this round. The remaining 79 materials (about 20%) are used as the validation set, whose phenotypic data are considered unknown in this round of model prediction, and are specifically used to test the accuracy of the model in predicting new materials.

[0104] (2) Constructing a predictive model: For the training set (314 copies) of the current round, its BLUP phenotypic value (y) and the genotype data of 30 SNPs (X) are used as input. Using the rrBLUP package in R language, the kin.blup function (default parameters) is run to fit the best linear unbiased prediction model of ridge regression and obtain the effect size estimate of the 30 SNPs in this round. This is the specific prediction model constructed.

[0105] (3) Prediction and Recording: Substitute the genotype data (X_validation) of the validation set (79 samples) into the model constructed in step (2) to calculate the estimated genomic breeding value of each validation material, i.e., the predicted phenotypic value. ), and recorded these 79 predicted values.

[0106] (4) Calculate the accuracy of a single round: Calculate the accuracy of the 79 predicted values ​​obtained from the validation set of this round ( A correlation analysis is performed between the model and its corresponding actual BLUP value (y), and the Pearson correlation coefficient (r) is calculated. This correlation coefficient represents the prediction accuracy of the model under the current data partition.

[0107] (5) Repeated verification: In order to eliminate the randomness of a single random partition and obtain a stable and reliable accuracy estimate, the above steps (1) to (4) are executed independently 100 times. Each execution involves a completely new random partition, rebuilding the model, re-predicting and calculating the accuracy.

[0108] 3. Results

[0109] Based on the results of the GWAS analysis, the top 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 200, 300, and 500 SNP markers with the highest effects were selected for genome-wide selection analysis. Simultaneously, markers of a certain number were randomly selected from 327,993 loci as controls. Figure 4The results showed that the whole-genome selection prediction accuracy of the high-effect marker group was higher than that of the random group at different marker densities. Furthermore, the average prediction accuracy of the model constructed using the top 30 SNPs in the high-effect group reached 0.72. These results fully demonstrate that the 30 core SNP marker combinations screened in this invention can efficiently capture the genetic essence of ploidy traits, and the constructed whole-genome selection model has excellent predictive ability and reliability, fully meeting the needs of early and accurate material screening in breeding practice.

[0110] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for constructing a prediction model for the expansion ratio of popcorn kernels, characterized in that, Includes the following steps: Acquire phenotypic data of kernel expansion fold from multiple popcorn breeding materials; detect the genotype of SNP molecular marker combinations in the multiple popcorn breeding materials; and construct a predictive model for the expansion fold using the phenotypic data and genotype data through a genomic selection statistical model. The SNP molecular marker combination consists of 30 SNP loci located in the maize B73 reference genome version Zm-B73-REFERENCE-GRAMENE-4.0, as shown in the table below: 。 2. The method according to claim 1, characterized in that, The number of popcorn breeding materials shall be no less than 300 popcorn inbred lines.

3. The method according to claim 1, characterized in that, The phenotypic data are the best linear unbiased predictions corrected through repeated trials in multiple environments.

4. A method for predicting the expansion ratio of popcorn kernels, characterized in that, Includes the following steps: A popcorn kernel expansion ratio prediction model is constructed using the method described in any one of claims 1-3; the genotypes of SNP molecular marker combinations in the popcorn samples to be tested are detected; the genotype data are input into the constructed prediction model to calculate the estimated genomic breeding value of each sample; the samples are sorted according to the estimated genomic breeding value, and the samples with higher breeding values ​​are predicted to have a higher kernel expansion ratio.

Images