Genome Wide Association Study For 17 Agronomic Traits In Maize And The Genome Assembly Of Mo17 And One Teosinte

i Genome-wide association study for 17 agronomic traits in maize There is no doubt that maize is one of the most important crop in the world.Planted under different climate conditions,maize shows abundantly amazing phenotype diversities.Researches about the natural variations that control the changes of phenotype could help us understand the genetic architecture of the traits and improve the nutritional quality,stress tolerance and yield of maize by molecular breeding.Obtaining high density genetic markers by genotype imputation have been widely used in genome-wide association studies to enhance the power and mapping precision.Here we develop a two-steps imputation methods to meet the challenge of large proportion missing genotypes and enlarge our association panel from 368 to 513 lines.And conduct the GWAS analysis for 17 agronomic traits by MLM and A-D test.The results are as follows:1.Genotype imputation and the validation:The association panel comprising 513 lines is genotyped by Maize SNP50 array with 56,110 SNPs(data setl).368 lines among the association panel have 556,809 high density SNPs(data set2).The high density SNPs of the left 145 lines are imputed by taking the sharing 49,728 SNPs between data set 1 and 2 as frame SNPs.Through the simulation using chromosome 1 data,the optimal parameters combination are determined reducing the missing rate from 91.6%to 12.8%with 96.6%accuracy rate.GWAS using imputed genotypes(data set 3)has higher power than using data set 1 and 2 by comparing the results from analyzing the published kernel oil concentration.2.Genome wide association studies:The phenotypes of 17 agronomic traits are obtained from filed trails in five environments in two years.Only 5 traits with 10 significant loci are detected by MLM at a threshold of P<1.80E-6(1/556809)and no significant loci is detected at a stricter threshold of P<8.98E-8(0.05/556809).However,a nonparametric model A-D test could detect 310 significant loci of 17 traits even at a stricter threshold of P<8.98E-8(0.05/556809).A-D test improves the power of GWAS for the complex agronomic traits.3.Evaluation of GWAS statistical models based on phenotype simulation:Through phenotype simulation,we analyse the relationship between statistical power and the effects of QTL,allele frequencies and phenotype distribution and conclude that A-D test is more applicable to detect moderate QTNs and more robust to the phenotype with abnormal distribution,while MLM always has higher or similar power in major QTNs detection.4.The validation of association results and candidate genes prediction:we used three independent RIL populations to conduct linkage analysis for kernel width,ear length and kernel number per row.9 significant loci detected by association analysis are within the QTL confidence interval,and we identify the corresponding candidate genes.Above all,the IBD-KNN two steps imputation method is applicative to impute the large proportion missing genotypes.A-D test is an alternative method for GWAS analysis.Combing genotype imputation and A-D test,we successfully explore the genetic architecture of 17 agronomic traits in maize.All of these supply the valuable resource for marker-assisted breeding in the future.ii The genome assembly of Mo17 and mexicanaAnother area interested in our research is the genome assembly of Mo 17 and a wild maize relative.The maize reference genome-B73 which had been published in 2009 have greatly accelerated the researches about maize genomics and maize functional genomics and brought enormous scientific and economic benefits over the past years.Nowadays,the sequencing technologies,omics and genome edit technologies are rapidly developing,it is gaining importance to identify more elite alleles in the magic maize.But a single B73 reference genome is far from enough.Therefore we are aimed to assemble two available Mo17 and mexicana genomes.These results also supply the sequence foundation for maize pan-genome and maize wild relatives researches.The main results are as follows:1.Genome assembly:Draft genomes for Mo 17(a modern and widely-used maize inbred)and teosinte(Zea mays ssp.mexicana,hereafter mexicana)were assembled using a novel meta-assembly strategy based on genomic sequencing of ten selected lines derived from a maize-teosinte cross.The final assembled genomes were 2.04 Gb for Mo17 and 1.20 Gb for mexicana with scaffolds N50 of 3 Mb and 106 kb,respectively.2.Structure variations:Comparative analyses revealed the high level of diversity between Mo 17,B73 and mexicana,including hundreds of thousands of PAVs,megabase-size structural variations and a nearly-30Mb inversion.3.Identification of structure variations:Using the assembly of genomes,we identified 310 positive selective genes in modern maize and mexicana.The orthologs of two positive selective genes in mexicana have been proven to be related with drought,salt and cold tolerance in rice.This could be explained by the special highland living conditions of mexicana.4.The pattern of spontaneous mutation in maize:The pattern of spontaneous mutation in maize:Spontaneous mutation rate was estimated to be 3.87×10-8 per site per generation in maize,with a nonrandom distribution across the genome.A higher deleterious mutation rate was observed in the pericentromeric regions than elsewhere in the genome,which might be caused by differences in recombination frequency.5.Introgression:The mexicana genome contributed significantly to maize adaptation and improvement with 3.1%and 2.2%putative introgression regions identified in the B73 and Mo 17 genomes,respectively.