A method for evaluating the potential of a parental combination for inducing double haploids in plant breeding

Abstract

The present application relates to the technical field of plant breeding, and particularly relates to a method for evaluating the double haploid induction potential of parent combinations in plant breeding. The present application constructs a method for evaluating the double haploid induction potential of parent combinations in plant breeding through genetic recombination simulation, a phenotype prediction model and a comprehensive evaluation function of multi-trait collaborative selection, can quantitatively predict the induction potential of different parent combinations before induction, can simultaneously consider multiple breeding objectives, realize collaborative optimization, and thus help to prioritize the allocation and scale planning of breeding resources. The present application can effectively save induction and identification resources, accelerate the breeding process of excellent varieties, and has a wide application prospect in the breeding of corn, rice, wheat and other plants.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of plant breeding, and particularly relates to a method for evaluating the double haploid induction potential of parent combinations in plant breeding. BACKGROUND

[0002] A double haploid (DH) line is a homozygous line formed by doubling the chromosomes of gametes through induction. By using double haploid induction technology, homozygous materials can be quickly obtained, and the breeding cycle can be shortened, which has become an important tool in modern breeding of crops such as corn, rice and wheat.

[0003] In traditional breeding practice, whether a parent combination can successfully induce an excellent double haploid line (DH line) often depends on field or greenhouse experiments, which is time-consuming, costly and resource-consuming. Specifically, different parent combinations show significant differences in the potential of double haploid induction, and the number of different parent combinations is huge, but resources are limited in breeding practice. Therefore, if there is no prior scientific evaluation, blind double haploid induction will lead to resource waste and low efficiency.

[0004] Most existing studies use genotype and trait information to predict the performance of the offspring of a combination, and lack prediction of the induction potential of the combination. Moreover, in conventional breeding, multiple traits are usually set as breeding goals as the standard for excellent materials. However, most methods only target a single trait when screening materials, and the classic multi-trait selection method relies on the experience of breeders and lacks intelligent selection of multiple traits associated with each other.

[0005] Genetic recombination simulation (e.g. Faux AM, Gorjanc G, Gaynor RC, Battagin M, Edwards SM Wilson DL, Hearne SJ, Gonen S, Hickey JM. (2016). AlphaSim: Software for Breeding Program Simulation. Plant Genome. 9(3) or Gaynor RC, Gorjanc G, Hickey JM. (2022). AlphaSimR: an R package for breeding program simulations. G3-Genes Genomes Genet. 11(2):jkaa017 or Moeinizade S, Han Y, Pham H, Hu GP, Wang LZ. (2021). A look-ahead Monte Carlo simulation method for improving parental selection in trait introgression. Sci Rep. 11(1):3918) and phenotypic prediction (e.g. Alemu A, Astrand J, Montesinos-Lopez OA, Sanchez JIY, Fernandez-Gonzalez J, Tadesse W, Vetukuri RR, Carlsson AS, Ceplitis A, Crossa J, Ortiz R, Chawade A. (2024). Genomic selection in plant breeding: Key factors shaping two decades of progress. Mol Plant. 17(4):552-578), multi-trait selection models (e.g. Moeinizade S, Kusmec A, Hu GP, Wang LZ, Schnable PS. (2020). Multi-trait Genomic Selection Methods for Crop Improvement. Genetics. 215(4):931-945 or Yang WY, Guo TT, Luo JY, Zhang RY, Zhao JR, Warburton ML, Xiao YJ, Yan JB.The development of Target-Oriented Prioritization: Targeted Selection Strategy by Integrating Organismal and Molecular Traits through Predictive Analytics in Breeding (2022) by Zhang et al. Genome Biology. 23(1):80 provides the possibility of constructing virtual predictive selection based on genotypic information. SUMMARY

[0006] The present application combines genetic recombination simulation, phenotype prediction, and multi-trait comprehensive evaluation, and provides an evaluation method for the double haploid induction potential of parent combinations in plant breeding. The evaluation method comprises:

[0007] Step (1): genetic recombination simulation is performed according to the genotypic data and genetic map data of each parent in the parent combination to be evaluated to generate a virtual double haploid population;

[0008] Step (2): a pre-established phenotype prediction model is used to predict the target traits of the test cross progeny obtained by test crossing the virtual double haploid population with a test variety;

[0009] Step (3): based on the breeding goal, a comprehensive evaluation function is used to score the results of the target trait prediction, and compared with a pre-set excellent threshold, and the induction potential value is calculated according to the proportion of virtual double haploid individuals reaching the pre-set excellent threshold.

[0010] The evaluation method of the present application realizes accurate prediction and quantification of the double haploid induction potential of parent combinations through genetic recombination simulation, phenotype prediction model, and comprehensive evaluation function of multi-trait collaborative selection, thereby solving the following defects existing in the prior art: (1) in the case of insufficient experience, multiple breeding goals cannot be considered in breeding; (2) the potential of a parent combination to produce excellent DH lines cannot be accurately predicted before induction.

[0011] Preferably, in step (1), the genetic map data includes genetic distance information data between each genetic marker.

[0012] In some embodiments, the genetic distance information data is usually expressed in units of cM (centiMorgan).

[0013] Preferably, the parent combination is a two-parent combination or a multi-parent hybrid combination (three-parent combination and above).

[0014] In some embodiments, the genotypic data comes from SNP chip sequencing, whole genome resequencing, or genotyping platforms.

[0015] Preferably, step (1) specifically includes: converting the genetic distance information data into recombination probability by using a mapping function, and then randomly simulating whether a recombination event occurs in each marked interval according to the recombination probability; recording the virtual gamete genotype generated after a successful recombination event, performing chromosome doubling operation on the virtual gamete genotype to obtain the virtual double haploid individual genotype, and finally generating the virtual double haploid population.

[0016] Preferably, recombination probability Through function Definition, mapping function Used to store genetic distance information data Convert to recombination probability .

[0017] Preferably, in each random simulation, a virtual gamete genotype is generated from the parental genotype, denoted as .

[0018] Preferably, the virtual gamete genotype is subjected to chromosome doubling to obtain the virtual double haploid individual genotype, denoted as , This indicates chromosome doubling of the virtual gamete genotype.

[0019] Preferably, the final generated virtual double haploid population is denoted as .in, This indicates the parents that produced the double haploid population. Indicates the number of parents; This indicates the size of the virtual double haploid population.

[0020] In the specific implementation process, the size of the virtual double haploid population It can be set as needed, such as 100, 500, 1000 or larger.

[0021] Preferably, in step (2), the phenotypic prediction model includes a linear mixture model, a Bayesian regression model, or a machine learning model.

[0022] Preferably, the general prediction formula used in the phenotypic prediction model is expressed as follows: ;

[0023] in, Indicates by the first Genotype of a virtual double haploid individual Genotype of the test species Genotypes of the offspring after a test cross; The phenotypic prediction value of the testcross progeny; is a prediction function. In practical breeding calculation, a one-time prediction is usually performed on a population containing test cross offspring. Let the genotype set of the test cross offspring population be , then the phenotype prediction value vector of the population is . The prediction process is realized by vector operation, denoted as .

[0024] Preferably, the prediction function has a specific mathematical form determined according to the model framework adopted, including but not limited to: when a linear mixed model is adopted, the corresponding linear function form; when a Bayesian regression model is adopted, the corresponding integral form based on the empirical distribution; or when a machine learning model is adopted, the corresponding nonlinear function form. Those skilled in the art can select according to the data characteristics and accuracy requirements in the actual breeding scene.

[0025] As a specific embodiment, when the prediction function is a linear mixed model, it can be specifically represented as:

[0026] ;

[0027] wherein, represents the phenotype prediction value vector of the test cross offspring population; represents the fixed effect design matrix of the training set; represents the fixed effect design matrix of the test cross offspring; represents the estimated value of the fixed effect; represents the kinship matrix between individuals in the training set; represents the kinship matrix between the test cross offspring and the individuals in the training set; is the estimated value of the additive genetic variance; is the estimated value of the residual variance; is the true phenotype value of the training set; is the unit matrix.

[0028] Preferably, in step (3), the comprehensive evaluation function includes a comprehensive selection index function based on economic weight or a goal-oriented optimization algorithm function.

[0029] Preferably, the comprehensive evaluation function is used for synergistic calculation of multiple target traits (for example target traits, 2) prediction values.

[0030] Preferably, the comprehensive evaluation function is a multi-trait synergistic selection function, and for a single test cross offspring genotype in the population, the calculation formula of the comprehensive selection index is represented as:

[0031] ;

[0032] in, Indicates by the first Genotype of a virtual double haploid individual Genotype of the test species Genotypes of the offspring after a test cross , To determine the number of offspring from a test cross; Indicates based on the first Genotype of a virtual double haploid individual Genotype of the test species The predicted number of offspring genotypes after a test cross A vector of phenotypic predicted values ​​for each target trait. ; To be Individual trait predictions are mapped to a comprehensive evaluation function.

[0033] The comprehensive evaluation function Various comprehensive evaluation functions known in this field can be used.

[0034] As one implementation, the comprehensive evaluation function is a comprehensive selection index function, and its formula is as follows:

[0035] ;

[0036] in, Indicates the first The weight of each target trait; Indicates the first Phenotypic predictions for each target trait.

[0037] As another implementation method, the comprehensive evaluation function The goal-oriented optimization algorithm function is defined by the following formula:

[0038] ;

[0039] in, Indicates the first Phenotypic predictions for each target trait; This represents a vector of preset target values ​​for each trait. Indicates the first The weight of each target trait.

[0040] Preferably, in step (3), the formula for calculating the induction potential value is as follows:

[0041] ;

[0042] wherein, represents an induction potential value; represents a parent for producing a DH population, represents the number of parents; represents a preset excellent threshold value; is the number of simulations; is an indicator function, which takes 1 when the condition is true, and 0 otherwise; represents a virtual DH population produced in the i-th simulation; j represents a virtual DH population produced in the i-th simulation; is a comprehensive score value obtained by the comprehensive evaluation function.

[0043] Preferably, is set to 1000-10000 to ensure the stability of the statistical results.

[0044] In the present application, the induction potential value is used to measure the potential of a test parent combination to produce an excellent DH system.

[0045] Preferably, the evaluation method further comprises: ranking a plurality of parent combinations based on the induction potential value to determine a resource allocation scheme (for example, parent combinations with higher induction potential values are preferentially used for large-scale DH induction experiments, and parent combinations with lower induction potential values are reduced or not allocated experimental resources).

[0046] Alternatively, a statistical relationship model between the induction scale, the number of excellent DH systems, and the induction success probability is established based on the induction potential value.

[0047] Preferably, the general form of the statistical relationship model is represented as follows:

[0048] ;

[0049] wherein, represents the induction scale; represents the number of excellent DH systems to be obtained; represents the induction success probability; represents the induction potential value; represents a parent for producing a DH population, represents the number of parents; is a success probability calculation function.

[0050] The success probability calculation function includes but is not limited to construction based on binomial distribution model, Poisson distribution approximation, and Monte Carlo simulation.

[0051] As an implementation manner, the success probability calculation function Based on a binomial distribution model (Bernoulli trial model). The model regards each double haploid induction as an independent trial, and the induction potential value of generating one good double haploid is Under this model, the induction success probability of obtaining at least good double haploids is The calculation formula is:

[0052] ;

[0053] As other implementation manners, the success probability calculation function may also adopt other probability models or calculation methods, for example:

[0054] When the induction scale is large and the induction potential value is small, Poisson distribution can be used for approximate calculation.

[0055] Monte Carlo simulation method can also be used to statistically estimate the induction success probability by a large number of computer random experiments.

[0056] Preferably, the statistical relationship model is used for at least one of the following aspects:

[0057] (1) Given the induction scale and the induction potential value, estimate the success probability of achieving the quantity target of good double haploids;

[0058] (2) Given the induction success probability and the induction potential value, calculate the required induction scale;

[0059] (3) Given the induction scale and the quantity target of good double haploids, evaluate the induction success probability.

[0060] Further, the present application provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the evaluation method when executing the program.

[0061] Further, the present application provides a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the evaluation method.

[0062] Further, the present application provides a computer program product comprising a computer program, wherein the computer program is executable by a processor to implement the evaluation method.

[0063] Compared with the prior art, the present application has the following beneficial effects:

[0064] This invention constructs a method for assessing the induction potential of double haploid parental combinations in plant breeding by integrating genetic recombination simulation, phenotypic prediction models, and multi-trait synergistic selection evaluation functions. It can quantitatively predict the induction potential of different parental combinations before induction and simultaneously consider multiple breeding objectives, achieving synergistic optimization. This facilitates the priority allocation and large-scale planning of breeding resources. This invention can effectively save induction and identification resources, accelerate the breeding process of superior varieties, and has broad application prospects in plant breeding such as maize, rice, and wheat. Attached Figure Description

[0065] Figure 1 This is a verification and comparison of the induction potential of double haploid maize parental combinations.

[0066] Figure 2 It is the induction curve of a single parent combination of maize. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention. In the embodiments provided in this specification, where specific techniques or conditions are not specified, they are performed according to the techniques or conditions described in the literature in this field, or according to the product instructions. Reagents or instruments whose manufacturers are not specified are all conventional products that can be purchased through legitimate channels.

[0068] Example 1: Evaluation of the induction potential of double haploids in maize single parental combinations

[0069] (1) Corn Taking a parental combination consisting of two parents as an example, based on marker genetic distance information from SNP data and genetic map data... Determine the recombination probability between adjacent tags. Recombination probability Through mapping function Definition. Subsequently, based on the recombination probability, a random simulation is performed within each labeled interval to determine whether a recombination event occurs. The genotype of the virtual gametes generated after a successful recombination event is recorded, denoted as . Chromosomal doubling is performed on the virtual gamete genotype to obtain the virtual double haploid individual genotype, denoted as , This indicates that chromosome doubling is performed on the virtual gamete genotypes, ultimately generating the virtual double haploid population, denoted as […]. Genetic recombination simulations were conducted twice, with a population size of 1000 each time.

[0070] (2) Using the linear mixed model established in advance to predict the target traits of the test cross offspring obtained by the test cross between the virtual diploid population and the test species; the prediction formula of the phenotypic prediction model is expressed as:

[0071] ;

[0072] wherein, represents the phenotypic prediction value vector of the test cross offspring population; represents the fixed effect design matrix of the training set; represents the fixed effect design matrix of the test cross offspring; represents the estimated value of the fixed effect; represents the kinship matrix between the individuals of the training set; represents the kinship matrix between the test cross offspring and the individuals of the training set; is the estimated value of the additive genetic variance; is the estimated value of the residual variance; is the true phenotypic value of the training set; is the unit matrix.

[0073] (3) Based on trait 1, trait 2, …, trait 11, multi-trait coordinated selection is carried out, and a comprehensive evaluation function is used to score the results of the target trait prediction, wherein the comprehensive evaluation function is a multi-trait coordinated selection function, and the formula is expressed as:

[0074] ;

[0075] wherein, , represents the genotype of the test cross offspring formed after the genotype of the i-th virtual diploid individual and the genotype of the test species are crossed represents the predicted phenotypic prediction value vector of the above 11 target traits; represents the preset target value vector of each trait; represents the weight of the i-th trait. The following Table 1 shows the comprehensive score calculation process for a specific test cross offspring genotype . Table 1

[0076]

[0077] Table 1

[0078]

[0079] According to the above table data and calculation process, the comprehensive score of the test cross offspring genotype​​​ The overall score is 0.01.

[0080] The comprehensive score result of the comprehensive evaluation function is compared with the preset excellent threshold. By comparison, the induction potential value is calculated based on the proportion of virtual double haploid individuals that reach the preset excellent threshold. The formula for calculating the induced potential value is as follows:

[0081] ;

[0082] in, Indicates the potential value; This represents the preset excellent threshold (in this embodiment, ); For the number of simulations (in this embodiment, ); This is an indicator function that takes the value 1 when the condition is true and 0 otherwise. Indicates the first j The virtual double haploid population generated by this simulation; This is the comprehensive score value obtained from the comprehensive evaluation function.

[0083] Simulation results: The number of individuals meeting the threshold was 53 in the first simulation and 60 in the second. Therefore, The result is 0.057, indicating that under a certain scale, this parental combination has a 5.7% probability of producing a superior DH line.

[0084] Example 2: Verification and Comparison of the Induction Potential of Double Haploids in Maize Parental Combinations

[0085] To verify the accuracy and superiority of the evaluation method of this invention, a real maize breeding population was used as an example, and the populations were ranked according to their induction potential. The method was based on the literature Bernardo, Rex., "Genomewide selection of parental inbreds: classes of loci and virtual biparental populations." Crop Science 54.6 (2014): 2586-2595. The method (denoted as the BASE method) and the method of this invention (denoted as the TOP_Order method) are used to evaluate the excellent double haploid recognition efficiency of the above population. The recognition efficiency curves under different selection pressures are plotted and the area under the curve (AUC) is calculated as a quantitative indicator (the higher the AUC value, the better the recognition efficiency).

[0086] The results are as follows Figure 1As shown, the evaluation method of the present application can effectively distinguish the low-value group and the high-value group, and the AUC value of TOP_order reaches 0.688, while the AUC value of BASE method as a control is 0.618.

[0087] Example 3: Ranking a plurality of parent combinations based on induction potential values to determine a resource allocation scheme

[0088] Taking three parent combinations C1, C2 and C3 of corn as examples, the induction potential values of the three parent combinations are calculated according to the steps of Example 1, and the values are 0.001, 0.18 and 0.30, respectively. Under the condition of limited total resources, C3 is preferentially induced, followed by C2, and C1 can be selected not to be induced.

[0089] Example 4: Induction suggestion for a single parent combination of corn

[0090] Taking parent combination C4 of corn as an example, the induction potential value of the parent combination is calculated according to the steps of Example 1, and the value is 0.0085. Based on the induction potential value, a statistical relationship model is established between the induction scale, the target number of excellent dihaploid systems and the induction success probability, and the model is constructed based on the principle of binomial distribution, and the expression is as follows:

[0091] ;

[0092] Wherein, represents the induction scale; represents the target number of excellent dihaploid systems obtained; represents the induction success probability; represents the induction potential value.

[0093] Based on the above statistical relationship model, when is 3, is 100, the final induction success probability is 5%; corn parent combination C4 needs to induce at least 989 DH lines to achieve a final induction success probability of 99% (as shown in Figure 2 ).

[0094] Finally, it should be pointed out that the above examples are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for evaluating the induction potential of double haploid combinations in plant breeding, characterized in that, include: Step (1): Genetic recombination simulation is performed based on the genotype data and genetic map data of each parent in the parental combination to be evaluated to generate a virtual double haploid population; Step (2): The target trait is predicted for the testcross progeny obtained by the testcross between the virtual double haploid population and the test species using a pre-established phenotypic prediction model; the phenotypic prediction model is a linear mixture model; Step (3): Based on the breeding objectives, the results of the target trait prediction are scored using a comprehensive evaluation function and compared with a preset excellence threshold. The induction potential value is calculated based on the proportion of virtual double haploid individuals that reach the preset excellence threshold. The comprehensive evaluation function is a multi-trait co-selection function. Step (1) specifically includes: converting genetic distance information data into recombination probability by using a mapping function, and then randomly simulating whether a recombination event occurs in each labeled interval according to the recombination probability; recording the virtual gamete genotype generated after a successful recombination event, performing chromosome doubling operation on the virtual gamete genotype to obtain the virtual double haploid individual genotype, and finally generating the virtual double haploid population; In step (2), the prediction formula of the phenotypic prediction model is expressed as: ； in, This represents a vector of predicted phenotypic values ​​for the testcross offspring population. Represents the fixed effects design matrix of the training set; Represents the fixed-effects design matrix for testcross offspring; Represents the estimated value of the fixed effects; This represents the kinship matrix among individuals in the training set. This represents the kinship matrix between testcross offspring and individuals in the training set; It is an estimate of the additive genetic variance; It is an estimate of the residual variance; These are the true phenotypic values ​​of the training set; It is an identity matrix.

2. The evaluation method according to claim 1, characterized in that, In step (1), the genetic map data includes genetic distance information between each genetic marker; And / or, the parental combination is a biparental combination or a multiparental hybrid combination.

3. The evaluation method according to claim 1, characterized in that, In step (3), the formula for the comprehensive evaluation function is expressed as: ； in, Indicates by the first Genotype of a virtual double haploid individual Genotype of the test species Genotypes of the offspring after a test cross , To determine the number of offspring from a test cross; Indicates based on the first Genotype of a virtual double haploid individual Genotype of the test species The predicted number of offspring genotypes after a test cross A vector of phenotypic predicted values ​​for each target trait. ; To be Individual trait predictions are mapped to a comprehensive evaluation function.

4. The evaluation method according to any one of claims 1 to 3, characterized in that, In step (3), the formula for calculating the induction potential value is as follows: ； in, Indicates the potential value; Indicates the parents that produced the double haploid population; , M Indicates the number of parents; This indicates the preset optimal threshold. For the number of simulations; This is an indicator function that takes the value 1 when the condition is true and 0 otherwise. Indicates the first j The virtual double haploid population generated by this simulation; This is the comprehensive score value obtained from the comprehensive evaluation function.

5. The evaluation method according to claim 1, characterized in that, Also includes: Multiple parental combinations are ranked based on the induced potential values ​​to determine a resource allocation scheme; Alternatively, a statistical relationship model can be established based on the induction potential value to connect the induction scale, the target number of superior double haploid systems, and the probability of successful induction.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the evaluation method as described in any one of claims 1 to 5.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the evaluation method as described in any one of claims 1 to 5.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the evaluation method as described in any one of claims 1 to 5.

Images