Heterosis Prediction Model Of Yield Traits Of Brassica Napus And Liquid Chip

The heterosis is a phenomenon that the first generation of hybrid is superior to the parent in one or more traits,which is widespread in the biological world.Utilization of crop heterosis is one of the most effectual methods to increase crop yield and improve crop quality.With the decrease of the cost of sequencing technology,genomics research has entered a new era of high precision and high throughput.The development and application of gene chips based on molecular markers can quickly select excellent hybrid parent combinations,significantly improve the utilization rate of heterosis,accelerate the breeding process of rapeseed,and the breeding process is accelerated.In this study,35 samples of Brassica napus were used as parental materials,and306 hybrid combinations were formulated by incomplete diallel crossing,the correlations and heterozygosity of 9 yield-related traits such as parental and hybrid progeny plants were analyzed.Base on resequencing of 35 parent of Brassica napus and phenotypic data of all materials,and through the difference analysis and collinearity analysis between groups with the marker type as the fixed factor,screening out specific SNP marker loci related to traits,the effect size and properties of the specific loci were estimated.By analyzing the relationship between effect value and performance of hybrids and their parents,and a molecular marker prediction model for hybrid performance was constructed using stepwise regression analysis.Moreover,a liquid chip was developed for SNP markers included in the prediction model and some related work was done.The specific research results are as follows:1.Analysis of 9 traitsANOVA was conducted on 9 yield-related traits including plant height,first effective branching node,first effective branch number of 306 Brassica napus hybrids F1,the results showed that first effective branch number,principal effective length,first effective branching node,silique length,number of berries per silique,thousand-grain weight and yield per plant,they were significant or extremely significant differences in306 hybrids F1.But the principal effective silique number was not significantly different among the 306 F1 hybrids.The variance analysis was performed on 340 materials including parental and hybrid F1 using the same method,the results showed that plant height,first effective branching node,first effective branch number,principal effective length,silique length,number of berries per silique,thousand-grain weight and yield per plant,these 8 traits had significantly or extremely significant differences in340 materials including parents and hybrid F1,but principal effective silique number was not significant difference among the 340 materials.2.Screening of specific marker loci35 parental materials whole gene were resequenced,got a total of about 2,518.906 Mb clean reads and 752.94 Gb clean data,the average sequencing coverage averages12 X.A total of 609,094 SNP loci with polymorphisms were obtained using Zhongshuang 11 of genome as the reference sequence.The marker status of the parents and the predicted hybrid progeny was divided into six marker types,and the marker type was used as a fixed factor for grouping,and 2,421～ 360,970 different numbers of specific SNPs loci were screened out for each trait.After collinear analysis,the retention rate of SNP loci of each trait ranged from 41.06%to 60.29%,the number of loci with different properties was as follows: dominant loci >additive loci > additive-dominant loci.For each character,the dominant effect is greater than the additive effect.3.Hybrid performanc prediction modelThe effect values of relevant loci in research materials were estimated according to the types and effects of specific loci.Taking the principal effective silique numbe as an example,the additive effect size was between-35.15 and 37.724,and the dominant effect value was between-30.507 and 33.168,the ratio of additive effect to dominant effect of 9 traits was between 0.116 and 0.830,except for principal effective silique number,each trait was more strongly influenced by dominant effects than additive effects.The best prediction models for the nine traits were obtained using stepwise regression analysis with the marker effect values of the nine trait-specific SNP marker loci of 340 materials as the independent variables and the trait phenotype values as the dependent variables,and the coefficients of determination were higher for eight traits except for first effective branch number.This shows that the accuracy of the prediction models are very high.The optimal prediction model of 9 traits was obtained through 10-fold crossvalidation which was repeated three times,these optimal prediction models involve marker loci and numbers that are consistent with the results of the best prediction models obtained by stepwise regression analysis.Except for first effective branch number,the optimal prediction models of the other eight traits have good stability and high accuracy,which have high reference value for application in breeding work.4.Development of liquid chipBased on the 357 SNP loci(solid-phase chip detection)included in the prediction model of 9 traits constructed in the early stage of our laboratory,using Geno Plexs technology to develop liquid-phase chip,only 249 SNP loci are single copy in the genome of Brassica napus.And the microarray detection was performed with seven materials,and the number of effective loci detected was less than half of the number of loci involved in the prediction model.This detection effect is not satisfactory,so the chip needs further improvement before it can be applied.5.R package development of heterosis utilizationDevelopment of the R package(pre＿heterosis package)for hybrid performance prediction using the Rstudio of Windows platform.The pre＿heterosis package consists of three functions,eff＿mar(),Multicollinearity(),and value＿eff＿mar(),and they can realize functions such as significant locus screening and effect estimation,collinearity analysis elimination,and gene effect value matrix estimation.The development of this R package can fill in the blank of some calculation programs in heterosis prediction research,and play a certain role in promoting related research in the future.