Improving power and accuracy of genome-wide association studies via a multi-locus mixed linear model methodology

Genome-Wide Association Study Reveals Both Overlapping and Independent Genetic Loci to Control Seed Weight and Silique Length in Brassica napus

Seed weight (SW) is one of three determinants of seed yield, which positively correlates with silique length (SL) in Brassica napus (rapeseed). However, the genetic mechanism underlying the relationship between seed weight (SW) and silique length (SL) is largely unknown at present. A natural population comprising 157 inbred lines in rapeseed was genotyped by whole-genome re-sequencing and investigated for SW and SL over four years. The genome-wide association study identified 20 SNPs in significant association with SW on A01, A04, A09, C02, and C06 chromosomes and the phenotypic variation explained by a single locus ranged from 11.85% to 34.58% with an average of 25.43%. Meanwhile, 742 SNPs significantly associated with SL on A02, A03, A04, A07, A08, A09, C01, C03, C04, C06, C07, and C08 chromosomes were also detected and the phenotypic variation explained by a single locus ranged from 4.01 to 48.02% with an average of 33.33%, out of which, more than half of the loci had not been reported in the previous studies. There were 320 overlapping or linked SNPs for both SW and SL on A04, A09, and C06 chromosomes. It indicated that both overlapping and independent genetic loci controlled both SW and SL in B. napus. On the haplotype block on A09 chromosome, the allele variants of a known gene BnaA.ARF18.a controlling both SW and SL were identified in the natural population by developing derived cleaved amplified polymorphic sequence (dCAPS) markers. These findings are valuable for understanding the genetic mechanism of SW and SL and also for rapeseed molecular breeding programs.

Detecting the QTL-Allele System of Seed Oil Traits Using Multi-Locus Genome-Wide Association Analysis for Population Characterization and Optimal Cross Prediction in Soybean

Soybean is one of the world's major vegetative oil sources, while oleic acid and linolenic acid content are the major quality traits of soybean oil. The restricted two-stage multi-locus genome-wide association analysis (RTM-GWAS), characterized with error and false-positive control, has provided a potential approach for a relatively thorough detection of whole-genome QTL-alleles. The Chinese soybean landrace population (CSLRP) composed of 366 accessions was tested under four environments to identify the QTL-allele constitution of seed oil, oleic acid and linolenic acid content (SOC, OAC, and LAC). Using RTM-GWAS with 29,119 SNPLDBs (SNP linkage disequilibrium blocks) as genomic markers, 50, 98, and 50 QTLs with 136, 283, and 154 alleles (2-9 per locus) were detected, with their contribution 82.52, 90.31, and 83.86% to phenotypic variance, corresponding to their heritability 91.29, 90.97, and 90.24% for SOC, OAC, and LAC, respectively. The RTM-GWAS was shown to be more powerful and efficient than previous single-locus model GWAS procedures. For each trait, the detected QTL-alleles were organized into a QTL-allele matrix as the population genetic constitution. From which the genetic differentiation among 6 eco-populations was characterized as significant allele frequency differentiation on 28, 56, and 30 loci for the three traits, respectively. The QTL-allele matrices were also used for genomic selection for optimal crosses, which predicted transgressive potential up to 24.76, 40.30, and 2.37% for the respective traits, respectively. From the detected major QTLs, 38, 27, and 25 candidate genes were annotated for the respective traits, and two common QTL covering eight genes were identified for further study.

The Application of Multi-Locus GWAS for the Detection of Salt-Tolerance Loci in Rice

Improving the salt-tolerance of direct-seeding rice at the seed germination stage is a major goal of breeders. Efficiently identifying salt tolerance loci will help researchers develop effective rice breeding strategies. In this study, six multi-locus genome-wide association studies (GWAS) methods (mrMLM, FASTmrMLM, FASTmrEMMA, pLARmEB, pKWmEB, and ISIS EM-BLASSO) were applied to identify quantitative trait nucleotides (QTNs) for the salt tolerance traits of 478 rice accessions with 162,529 SNPs at the seed germination stage. Among the 371 QTNs detected by the six methods, 56 were identified by at least three methods. Among these 56 QTNs, 12, 6, 7, 4, 13, 12, and 12 were found to be associated with SSI-GI, SSI-VI, SSI-MGT, SSI-IR-24h, SSI-IR-48h, SSI-GR-5d, and SSI-GR-10d, respectively. Additionally, 66 candidate genes were identified in the vicinity of the 56 QTNs, and two of these genes (LOC_Os01g45760 and LOC_Os10g04860) are involved in auxin biosynthesis according to the enriched GO terms and KEGG pathways. This information will be useful for identifying the genes responsible for rice salt tolerance. A comparison of the six methods revealed that ISIS EM-BLASSO identified the most co-detected QTNs and performed best, with the smallest residual errors and highest computing speed, followed by FASTmrMLM, pLARmEB, mrMLM, pKWmEB, and FASTmrEMMA. Although multi-locus GWAS methods are superior to single-locus GWAS methods, their utility for identifying QTNs may be enhanced by adding a bin analysis to the models or by developing a hybrid method that merges the results from different methods.

Genome-Wide Association Mapping of Starch Pasting Properties in Maize Using Single-Locus and Multi-Locus Models

Maize starch plays a critical role in food processing and industrial application. The pasting properties, the most important starch characteristics, have enormous influence on fabrication property, flavor characteristics, storage, cooking, and baking. Understanding the genetic basis of starch pasting properties will be beneficial for manipulation of starch properties for a given purpose. Genome-wide association studies (GWAS) are becoming a powerful tool for dissecting the complex traits. Here, we carried out GWAS for seven pasting properties of maize starch with a panel of 230 inbred lines and 145,232 SNPs using one single-locus method, genome-wide efficient mixed model association (GEMMA), and three multi-locus methods, FASTmrEMMA, FarmCPU, and LASSO. We totally identified 60 quantitative trait nucleotides (QTNs) for starch pasting properties with these four GWAS methods. FASTmrEMMA detected the most QTNs (29), followed by FarmCPU (19) and LASSO (12), GEMMA detected the least QTNs (7). Of these QTNs, seven QTNs were identified by more than one method simultaneously. We further investigated locations of these significantly associated QTNs for possible candidate genes. These candidate genes and significant QTNs provide the guidance for further understanding of molecular mechanisms of starch pasting properties. We also compared the statistical powers and Type I errors of the four GWAS methods using Monte Carlo simulations. The results suggest that the multi-locus method is more powerful than the single-locus method and a combination of these multi-locus methods could help improve the detection power of GWAS.

Metabolome-wide association studies for agronomic traits of rice

Identification of trait-associated metabolites will advance the knowledge and understanding of the biosynthetic and catabolic pathways that are relevant to the complex traits of interest. In the past, the association between metabolites (treated as quantitative traits) and genetic variants (e.g., SNPs) has been extensively studied using metabolomic quantitative trait locus (mQTL) mapping. Nevertheless, the research on the association between metabolites with agronomic traits has been inadequate. In practice, the regular approaches for QTL mapping analysis may be adopted for metabolites-phenotypes association analysis due to the similarity in data structure of these two types of researches. In the study, we compared four regular QTL mapping approaches, i.e., simple linear regression (LR), linear mixed model (LMM), Bayesian analysis with spike-slab priors (Bayes B) and least absolute shrinkage and selection operator (LASSO), by testing their performances on the analysis of metabolome-phenotype associations. Simulation studies showed that LASSO had the higher power and lower false positive rate than the other three methods. We investigated the associations of 839 metobolites with five agronomic traits in a collection of 533 rice varieties. The results implied that a total of 25 metabolites were significantly associated with five agronomic traits. Literature search and bioinformatics analysis indicated that the identified 25 metabolites are significantly involved in some growth and development processes potentially related to agronomic traits. We also explored the predictability of agronomic traits based on the 839 metabolites through cross-validation, which showed that metabolomic prediction was efficient and its application in plant breeding has been justified.

On the impact of relatedness on SNP association analysis

BACKGROUND

When testing for SNP (single nucleotide polymorphism) associations in related individuals, observations are not independent. Simple linear regression assuming independent normally distributed residuals results in an increased type I error and the power of the test is also affected in a more complicate manner. Inflation of type I error is often successfully corrected by genomic control. However, this reduces the power of the test when relatedness is of concern. In the present paper, we derive explicit formulae to investigate how heritability and strength of relatedness contribute to variance inflation of the effect estimate of the linear model. Further, we study the consequences of variance inflation on hypothesis testing and compare the results with those of genomic control correction. We apply the developed theory to the publicly available HapMap trio data (N=129), the Sorbs (a self-contained population with N=977 characterised by a cryptic relatedness structure) and synthetic family studies with different sample sizes (ranging from N=129 to N=999) and different degrees of relatedness.

RESULTS

We derive explicit and easily to apply approximation formulae to estimate the impact of relatedness on the variance of the effect estimate of the linear regression model. Variance inflation increases with increasing heritability. Relatedness structure also impacts the degree of variance inflation as shown for example family structures. Variance inflation is smallest for HapMap trios, followed by a synthetic family study corresponding to the trio data but with larger sample size than HapMap. Next strongest inflation is observed for the Sorbs, and finally, for a synthetic family study with a more extreme relatedness structure but with similar sample size as the Sorbs. Type I error increases rapidly with increasing inflation. However, for smaller significance levels, power increases with increasing inflation while the opposite holds for larger significance levels. When genomic control is applied, type I error is preserved while power decreases rapidly with increasing variance inflation.

CONCLUSIONS

Stronger relatedness as well as higher heritability result in increased variance of the effect estimate of simple linear regression analysis. While type I error rates are generally inflated, the behaviour of power is more complex since power can be increased or reduced in dependence on relatedness and the heritability of the phenotype. Genomic control cannot be recommended to deal with inflation due to relatedness. Although it preserves type I error, the loss in power can be considerable. We provide a simple formula for estimating variance inflation given the relatedness structure and the heritability of a trait of interest. As a rule of thumb, variance inflation below 1.05 does not require correction and simple linear regression analysis is still appropriate.

An innovative procedure of genome-wide association analysis fits studies on germplasm population and plant breeding

KEY MESSAGE

The innovative RTM-GWAS procedure provides a relatively thorough detection of QTL and their multiple alleles for germplasm population characterization, gene network identification, and genomic selection strategy innovation in plant breeding. The previous genome-wide association studies (GWAS) have been concentrated on finding a handful of major quantitative trait loci (QTL), but plant breeders are interested in revealing the whole-genome QTL-allele constitution in breeding materials/germplasm (in which tremendous historical allelic variation has been accumulated) for genome-wide improvement. To match this requirement, two innovations were suggested for GWAS: first grouping tightly linked sequential SNPs into linkage disequilibrium blocks (SNPLDBs) to form markers with multi-allelic haplotypes, and second utilizing two-stage association analysis for QTL identification, where the markers were preselected by single-locus model followed by multi-locus multi-allele model stepwise regression. Our proposed GWAS procedure is characterized as a novel restricted two-stage multi-locus multi-allele GWAS (RTM-GWAS, https://github.com/njau-sri/rtm-gwas ). The Chinese soybean germplasm population (CSGP) composed of 1024 accessions with 36,952 SNPLDBs (generated from 145,558 SNPs, with reduced linkage disequilibrium decay distance) was used to demonstrate the power and efficiency of RTM-GWAS. Using the CSGP marker information, simulation studies demonstrated that RTM-GWAS achieved the highest QTL detection power and efficiency compared with the previous procedures, especially under large sample size and high trait heritability conditions. A relatively thorough detection of QTL with their multiple alleles was achieved by RTM-GWAS compared with the linear mixed model method on 100-seed weight in CSGP. A QTL-allele matrix (402 alleles of 139 QTL × 1024 accessions) was established as a compact form of the population genetic constitution. The 100-seed weight QTL-allele matrix was used for genetic characterization, candidate gene prediction, and genomic selection for optimal crosses in the germplasm population.

Genome-wide association study reveals genetic loci and candidate genes for average daily gain in Duroc pigs

Objective

Average daily gain (ADG) is an important target trait of pig breeding programs. We aimed to identify single nucleotide polymorphisms (SNPs) and genomic regions that are associated with ADG in the Duroc pig population.

Methods

We performed a genome-wide association study involving 390 Duroc boars and by using the PorcineSNP60K Beadchip and two linear models.

Results

After quality control, we detected 3,5971 SNPs, which included seven SNPs that are significantly associated with the ADG of pigs. We identified six quantitative trait loci (QTL) regions for ADG. These QTLs included four previously reported QTLs on Sus scrofa chromosome (SSC) 1, SSC5, SSC9, and SSC13, as well as two novel QTLs on SSC6 and SSC16. In addition, we selected six candidate genes (general transcription factor 3C polypeptide 5, high mobility group AT-hook 2, nicotinamide phosphoribosyltransferase, oligodendrocyte transcription factor 1, pleckstrin homology and RhoGEF domain containing G4B, and ENSSSCG00000031548) associated with ADG on the basis of their physiological roles and positional information. These candidate genes are involved in skeletal muscle cell differentiation, diet-induced obesity, and nervous system development.

Conclusion

This study contributes to the identification of the casual mutation that underlies QTLs associated with ADG and to future pig breeding programs based on marker-assisted selection. Further studies are needed to elucidate the role of the identified candidate genes in the physiological processes involved in ADG regulation.

Substantial contribution of genetic variation in the expression of transcription factors to phenotypic variation revealed by eRD-GWAS

BACKGROUND

There are significant limitations in existing methods for the genome-wide identification of genes whose expression patterns affect traits.

RESULTS

The transcriptomes of five tissues from 27 genetically diverse maize inbred lines were deeply sequenced to identify genes exhibiting high and low levels of expression variation across tissues or genotypes. Transcription factors are enriched among genes with the most variation in expression across tissues, as well as among genes with higher-than-median levels of variation in expression across genotypes. In contrast, transcription factors are depleted among genes whose expression is either highly stable or highly variable across genotypes. We developed a Bayesian-based method for genome-wide association studies (GWAS) in which RNA-seq-based measures of transcript accumulation are used as explanatory variables (eRD-GWAS). The ability of eRD-GWAS to identify true associations between gene expression variation and phenotypic diversity is supported by analyses of RNA co-expression networks, protein-protein interaction networks, and gene regulatory networks. Genes associated with 13 traits were identified using eRD-GWAS on a panel of 369 maize inbred lines. Predicted functions of many of the resulting trait-associated genes are consistent with the analyzed traits. Importantly, transcription factors are significantly enriched among trait-associated genes identified with eRD-GWAS.

CONCLUSIONS

eRD-GWAS is a powerful tool for associating genes with traits and is complementary to SNP-based GWAS. Our eRD-GWAS results are consistent with the hypothesis that genetic variation in transcription factor expression contributes substantially to phenotypic diversity.

Whole genome sequencing-based association study to unravel genetic architecture of cooked grain width and length traits in rice

In this study, we used 2.9 million single nucleotide polymorphisms (SNP) and 393,429 indels derived from whole genome sequences of 591 rice landraces to determine the genetic basis of cooked and raw grain length, width and shape using genome-wide association study (GWAS). We identified a unique fine-mapped genetic region GWi7.1 significantly associated with cooked and raw grain width. Additionally, GWi7.2 that harbors GL7/GW7 a cloned gene for grain dimension was found. Novel regions in chromosomes 10 and 11 were also found to be associated with cooked grain shape and raw grain width, respectively. The indel-based GWAS identified fine-mapped genetic regions GL3.1 and GWi5.1 that matched synteny breakpoints between indica and japonica. GL3.1 was positioned a few kilobases away from GS3, a cloned gene for cooked and raw grain lengths in indica. GWi5.1 found to be significantly associated with cooked and raw grain width. It anchors upstream of cloned gene GW5, which varied between indica and japonica accessions. GWi11.1 is present inside the 3'-UTR of a functional gene in indica that corresponds to a syntenic break in chromosome 11 of japonica. Our results identified novel allelic structural variants and haplotypes confirmed using single locus and multilocus SNP and indel-based GWAS.

Whole genome sequencing-based association study to unravel genetic architecture of cooked grain width and length traits in rice

In this study, we used 2.9 million single nucleotide polymorphisms (SNP) and 393,429 indels derived from whole genome sequences of 591 rice landraces to determine the genetic basis of cooked and raw grain length, width and shape using genome-wide association study (GWAS). We identified a unique fine-mapped genetic region GWi7.1 significantly associated with cooked and raw grain width. Additionally, GWi7.2 that harbors GL7/GW7 a cloned gene for grain dimension was found. Novel regions in chromosomes 10 and 11 were also found to be associated with cooked grain shape and raw grain width, respectively. The indel-based GWAS identified fine-mapped genetic regions GL3.1 and GWi5.1 that matched synteny breakpoints between indica and japonica. GL3.1 was positioned a few kilobases away from GS3, a cloned gene for cooked and raw grain lengths in indica. GWi5.1 found to be significantly associated with cooked and raw grain width. It anchors upstream of cloned gene GW5, which varied between indica and japonica accessions. GWi11.1 is present inside the 3'-UTR of a functional gene in indica that corresponds to a syntenic break in chromosome 11 of japonica. Our results identified novel allelic structural variants and haplotypes confirmed using single locus and multilocus SNP and indel-based GWAS.

The genetic architecture of water-soluble protein content and its genetic relationship to total protein content in soybean

Water-soluble protein content (WSPC) is a critical factor in both soybean protein quality and functionality. However, the underlying genetic determinants are unclear. Here, we used 219 soybean accessions and 152 recombinant inbred lines genotyped with high-density markers and phenotyped in multi-environments to dissect the genetic architectures of WSPC and protein content (PC) using single- and multi-locus genome-wide association studies. In the result, a total of 32 significant loci, including 10 novel loci, significantly associated with WSPC and PC across multi-environments were identified, which were subsequently validated by linkage mapping. Among these loci, only four exhibited pleiotropic effects for PC and WSPC, explaining the low correlation coefficient between the two traits. The largest-effect WSPC-specific loci, GqWSPC8, was stably identified across all six environments and tagged to a linkage disequilibrium block comprising two promising candidate genes AAP8 and 2 S albumin, which might contribute to the high level of WSPC in some soybean varieties. In addition, two genes, Glyma.13G123500 and Glyma.13G194400 with relatively high expression levels at seed development stage compared with other tissues were regarded as promising candidates associated with the PC and WSPC, respectively. Our results provide new insights into the genetic basis of WSPC affecting soybean protein quality and yield.

Prediction and association mapping of agronomic traits in maize using multiple omic data

Genomic selection holds a great promise to accelerate plant breeding via early selection before phenotypes are measured, and it offers major advantages over marker-assisted selection for highly polygenic traits. In addition to genomic data, metabolome and transcriptome are increasingly receiving attention as new data sources for phenotype prediction. We used data available from maize as a model to compare the predictive abilities of three different omic data sources using eight representative methods for six traits. We found that the best linear unbiased prediction overall performs better than other methods across different traits and different omic data, and genomic prediction performs better than transcriptomic and metabolomic predictions. For the same maize data, we also conducted genome-wide association study, transcriptome-wide association studies and metabolome-wide association studies for the six agronomic traits using both the genome-wide efficient mixed model association (GEMMA) method and a modified least absolute shrinkage and selection operator (LASSO) method. The new LASSO method has the ability to perform statistical tests. Simulation studies show that the modified LASSO performs better than GEMMA in terms of high power and low Type 1 error.

pLARmEB: integration of least angle regression with empirical Bayes for multilocus genome-wide association studies

Multilocus genome-wide association studies (GWAS) have become the state-of-the-art procedure to identify quantitative trait nucleotides (QTNs) associated with complex traits. However, implementation of multilocus model in GWAS is still difficult. In this study, we integrated least angle regression with empirical Bayes to perform multilocus GWAS under polygenic background control. We used an algorithm of model transformation that whitened the covariance matrix of the polygenic matrix K and environmental noise. Markers on one chromosome were included simultaneously in a multilocus model and least angle regression was used to select the most potentially associated single-nucleotide polymorphisms (SNPs), whereas the markers on the other chromosomes were used to calculate kinship matrix as polygenic background control. The selected SNPs in multilocus model were further detected for their association with the trait by empirical Bayes and likelihood ratio test. We herein refer to this method as the pLARmEB (polygenic-background-control-based least angle regression plus empirical Bayes). Results from simulation studies showed that pLARmEB was more powerful in QTN detection and more accurate in QTN effect estimation, had less false positive rate and required less computing time than Bayesian hierarchical generalized linear model, efficient mixed model association (EMMA) and least angle regression plus empirical Bayes. pLARmEB, multilocus random-SNP-effect mixed linear model and fast multilocus random-SNP-effect EMMA methods had almost equal power of QTN detection in simulation experiments. However, only pLARmEB identified 48 previously reported genes for 7 flowering time-related traits in Arabidopsis thaliana.

Dissecting the genetic architecture of Fusarium verticillioides seed rot resistance in maize by combining QTL mapping and genome-wide association analysis

Fusarium verticillioides can be transmitted via seeds and cause systemic infection in maize (Zea mays L.); its mycotoxin has harmful effects on animal and human health. We combined QTL mapping in recombinant inbred line (RIL) populations with a genome-wide association study (GWAS) of 217 diverse maize lines using 224,152 single nucleotide polymorphisms (SNPs) under controlled conditions to determine the genetic architecture of F. verticillioides seed rot (FSR) resistance. Our study identified 8 quantitative trait loci (QTLs) and 43 genes associated with 57 SNPs that were correlated with FSR resistance through linkage mapping and GWAS, respectively. Among these, there were three candidate genes, namely GRMZM2G0081223, AC213654.3_FG004, and GRMZM2G099255, which were detected in both linkage mapping and GWAS. Furthermore, the near-isogenic lines (NILs) containing GRMZM2G0081223, which also had a susceptible parent background, were found to have a significantly improved level of resistance. In addition, the expression profile of the three candidate genes revealed that they all respond to the infection following inoculation with F. verticillioides. These genetic analyses indicate that FSR resistance is controlled by loci with minor effect, and the polymerization breeding of lines with beneficial alleles and candidate genes could improve FSR resistance in maize.

Iterative sure independence screening EM-Bayesian LASSO algorithm for multi-locus genome-wide association studies

Genome-wide association study (GWAS) entails examining a large number of single nucleotide polymorphisms (SNPs) in a limited sample with hundreds of individuals, implying a variable selection problem in the high dimensional dataset. Although many single-locus GWAS approaches under polygenic background and population structure controls have been widely used, some significant loci fail to be detected. In this study, we used an iterative modified-sure independence screening (ISIS) approach in reducing the number of SNPs to a moderate size. Expectation-Maximization (EM)-Bayesian least absolute shrinkage and selection operator (BLASSO) was used to estimate all the selected SNP effects for true quantitative trait nucleotide (QTN) detection. This method is referred to as ISIS EM-BLASSO algorithm. Monte Carlo simulation studies validated the new method, which has the highest empirical power in QTN detection and the highest accuracy in QTN effect estimation, and it is the fastest, as compared with efficient mixed-model association (EMMA), smoothly clipped absolute deviation (SCAD), fixed and random model circulating probability unification (FarmCPU), and multi-locus random-SNP-effect mixed linear model (mrMLM). To further demonstrate the new method, six flowering time traits in Arabidopsis thaliana were re-analyzed by four methods (New method, EMMA, FarmCPU, and mrMLM). As a result, the new method identified most previously reported genes. Therefore, the new method is a good alternative for multi-locus GWAS.

Genome-wide association study is concerned with the associations between markers and traits of interest so as to identify all the significantly associated markers. In genome-wide association studies, hundreds of thousands of markers are genotyped for several hundreds of individuals. Usually, only a very minor subset of these markers is associated with the trait. Most penalization methods fail when the number of markers is much larger than the sample size. Based on this fact, we have developed an algorithm that proceeds in two stages. In the first stage (screening), we reduced the number of markers via correlation learning to a moderate size. We then used a moderate-scale variable selection method to select variables in the reduced model. Conditional on the selected variables, we repeated the screening procedure and chose another set of variables. In the second stage (estimation), all the above-selected variables are accurately estimated in a multi-locus model. Our approach is simple, accurate in estimation, fast and shows high statistical power of detecting relevant markers on simulated data. We have also used this method to identify relevant genes in real data analysis. We recommend our approach for conducting a multi-locus genome-wide association study.

Development of a multiple-hybrid population for genome-wide association studies: theoretical consideration and genetic mapping of flowering traits in maize

Various types of populations have been used in genetics, genomics and crop improvement, including bi- and multi-parental populations and natural ones. The latter has been widely used in genome-wide association study (GWAS). However, inbred-based GWAS cannot be used to reveal the mechanisms involved in hybrid performance. We developed a novel maize population, multiple-hybrid population (MHP), consisting of 724 hybrids produced using 28 temperate and 23 tropical inbreds. The hybrids can be divided into three subpopulations, two diallels and NC (North Carolina Design) II. Significant genetic differences were identified among parents, hybrids and heterotic groups. A cluster analysis revealed heterotic groups existing in the parental lines and the results showed that MHPs are well suitable for GWAS in hybrid crops. MHP-based GWAS was performed using 55 K SNP array for flowering time traits, days to tassel, days to silk, days to anthesis and anthesis-silking interval. Two independent methods, PEPIS developed for hybrids and TASSEL software designed for inbred line populations, revealed highly consistent results with five overlapping chromosomal regions identified and used for discovery of candidate genes and quantitative trait nucleotides. Our results indicate that MHPs are powerful in GWAS for hybrid-related traits with great potential applications in the molecular breeding era.

Genome-Wide Association Mapping Reveals the Genetic Control Underlying Branch Angle in Rapeseed (Brassica napus L.)

Plant architecture is vital not only for crop yield, but also for field management, such as mechanical harvesting. The branch angle is one of the key factors determining plant architecture. With the aim of revealing the genetic control underlying branch angle in rapeseed (Brassica napus L.), the positional variation of branch angles on individual plants was evaluated, and the branch angle increased with the elevation of branch position. Furthermore, three middle branches of individual plants were selected to measure the branch angle because they exhibited the most representative phenotypic values. An association panel with 472 diverse accessions was estimated for branch angle trait in six environments and genotyped with a 60K Brassica Infinium® SNP array. As a result of association mapping, 46 and 38 significantly-associated loci were detected using a mixed linear model (MLM) and a multi-locus random-SNP-effect mixed linear model (MRMLM), which explained up to 62.2 and 66.2% of the cumulative phenotypic variation, respectively. Numerous highly-promising candidate genes were identified by annotating against Arabidopsis thaliana homologous, including some first found in rapeseed, such as TAC1, SGR1, SGR3, and SGR5. These findings reveal the genetic control underlying branch angle and provide insight into genetic improvements that are possible in the plant architecture of rapeseed.

Mapping small-effect and linked quantitative trait loci for complex traits in backcross or DH populations via a multi-locus GWAS methodology

Composite interval mapping (CIM) is the most widely-used method in linkage analysis. Its main feature is the ability to control genomic background effects via inclusion of co-factors in its genetic model. However, the result often depends on how the co-factors are selected, especially for small-effect and linked quantitative trait loci (QTL). To address this issue, here we proposed a new method under the framework of genome-wide association studies (GWAS). First, a single-locus random-SNP-effect mixed linear model method for GWAS was used to scan each putative QTL on the genome in backcross or doubled haploid populations. Here, controlling background via selecting markers in the CIM was replaced by estimating polygenic variance. Then, all the peaks in the negative logarithm P-value curve were selected as the positions of multiple putative QTL to be included in a multi-locus genetic model, and true QTL were automatically identified by empirical Bayes. This called genome-wide CIM (GCIM). A series of simulated and real datasets was used to validate the new method. As a result, the new method had higher power in QTL detection, greater accuracy in QTL effect estimation, and stronger robustness under various backgrounds as compared with the CIM and empirical Bayes methods.

EcoTILLING revealed SNPs in GhSus genes that are associated with fiber- and seed-related traits in upland cotton

Cotton is the most important textile crop in the world due to its cellulose-enriched fibers. Sucrose synthase genes (Sus) play pivotal roles in cotton fiber and seed development. To mine and pyramid more favorable alleles for cotton molecular breeding, single nucleotide polymorphisms (SNPs) of GhSus family genes were investigated across 277 upland cotton accessions by EcoTILLING. As a result, a total of 24 SNPs in the amplified regions of eight GhSus genes were identified. These SNPs were significantly associated with at least one fiber- or seed-related trait measured in Nanjing, Anyang and Kuche in 2007-2009. Four main-effect quantitative trait nucleotides (QTNs) and five epistatic QTNs, with 0.76-3.56% of phenotypic variances explained by each QTN (PVE), were found to be associated with yield-related traits; six epistatic QTNs, with the 0.43-3.48% PVE, were found to be associated with fiber quality-related traits; and one main-effect QTN and one epistatic QTN, with the PVE of 1.96% and 2.53%, were found to be associated with seed oil content and protein content, respectively. Therefore, this study provides new information for molecular breeding in cotton.