A critical evaluation of the effect of population size and phenotypic measurement on QTL detection and localization using a large F2 murine mapping population

ObjectiveThe determination of the location of quantitative trait loci (QTL) (i.e., QTL mapping) is essential for identifying new genes. Various statistical methods are being incorporated into different QTL mapping functions. However, statistical errors and limitations may often occur in a QTL mapping, implying the risk of false positive errors and/or failing to detect a true positive QTL effect. We simulated the power to detect four simulated QTL in tomato using cim() and stepwiseqtl(), widely adopted QTL mapping functions, and QTL.gCIMapping(), a derivative of the composite interval mapping method. While there is general agreement that those three functions identified simulated QTL, missing or false positive QTL were observed, which were prevalent when more realistic data (such as smaller population size) were provided.ResultsTo address this issue, we developed postQTL, a QTL mapping R workflow that incorporates (i) both cim() and stepwiseqtl(), (ii) widely used R packages developed for model selection, and (iii) automation to increase the accuracy, efficiency, and accessibility of QTL mapping. QTL mapping experiments on tomato F2 populations in which QTL effects were simulated or calculated showed advantages of postQTL in QTL detection.

The efficiency of quantitative trait locus (QTL) mapping methods needs to be investigated assuming high single nucleotide polymorphism (SNP) density and low heritability QTLs. This study assessed the efficiency of the least squares, maximum likelihood, and Bayesian approaches for QTL mapping assuming high SNP density and low heritability QTLs. We simulated 50 samples of 400 F2 individuals, which were genotyped for 1000 SNPs (average density of one SNP/centiMorgan) and phenotyped for three traits controlled by 12 QTLs and 88 minor genes. The genes were randomly distributed in the regions covered by the SNPs along ten chromosomes. The QTL heritabilities ranged from approximately 1–2% and the sample sizes were 200 and 400. The power of QTL detection ranged from 30 to 60%, the false discovery rate (FDR) ranged from only 0.5–1.2%, and the bias in the QTL position ranged from 4 to 6 cM. The QTL mapping efficiency was not influenced by the degree of dominance. The statistical approaches were comparable regarding the FDR. Regression-based and simple interval mapping methods showed equivalent power of QTL detection and mapping precision. Compared to interval mapping, the inclusive composite interval mapping provided slightly greater QTL detection power and mapping precision only for the intermediate and high heritability QTLs. By maximizing the prior number of QTLs, the Bayesian analysis provided the greatest power of QTL detection. No method proved to be superior.

The goal of quantitative trait loci (QTL) mapping in livestock is to find genes underlying traits of economic importance for genetic improvement through marker assisted selection (MAS). The studies presented in this thesis address several important issues in QTL detection and fine mapping using candidate gene analysis and linkage disequilibrium (LD) mapping using high density genotyping. Tests for candidate genes in F2 populations for QTL mapping were developed and evaluated. Results show that the extensive between-breed LD that is present in a cross can result in significant associations for candidate genes at considerable distances from the QTL. Tests that removed the impact of between-breed LD were not powerful in detecting candidate genes closely linked to the QTL, unless the candidate gene was the QTL. Therefore, candidate gene tests in QTL mapping populations must be interpreted with caution. Effectiveness of QTL mapping and MAS using LD in outbred populations depends on the extent of LD between markers and QTL which can differ between populations. Nine measures of LD between multi-allelic markers were evaluated as predictors of usable LD when LD is generated by drift. A standardized chi-square statistic ( ' ) was found to be the best predictor of usable LD of multi-allelic markers with QTL, while three other measures ( , 2 χ 2 df χ 2 r and ) were found to be good predictors of usable LD of single nucleotide polymorphisms (SNPs) with QTL. The effect of various factors on power and precision of QTL detection was evaluated and power and precision of regressionand identical by descent (IBD)-based LD mapping methods were compared. Power and precision of QTL detection increased with sample size, marker density and QTL effect. * D

Experimental error control is very important in quantitative trait locus (QTL) mapping. Although numerous statistical methods have been developed for QTL mapping, a QTL detection model based on an appropriate experimental design that emphasizes error control has not been developed. Lattice design is very suitable for experiments with large sample sizes, which is usually required for accurate mapping of quantitative traits. However, the lack of a QTL mapping method based on lattice design dictates that the arithmetic mean or adjusted mean of each line of observations in the lattice design had to be used as a response variable, resulting in low QTL detection power. As an improvement, we developed a QTL mapping method termed composite interval mapping based on lattice design (CIMLD). In the lattice design, experimental errors are decomposed into random errors and block-within-replication errors. Four levels of block-within-replication errors were simulated to show the power of QTL detection under different error controls. The simulation results showed that the arithmetic mean method, which is equivalent to a method under random complete block design (RCBD), was very sensitive to the size of the block variance and with the increase of block variance, the power of QTL detection decreased from 51.3% to 9.4%. In contrast to the RCBD method, the power of CIMLD and the adjusted mean method did not change for different block variances. The CIMLD method showed 1.2- to 7.6-fold higher power of QTL detection than the arithmetic or adjusted mean methods. Our proposed method was applied to real soybean (Glycine max) data as an example and 10 QTLs for biomass were identified that explained 65.87% of the phenotypic variation, while only three and two QTLs were identified by arithmetic and adjusted mean methods, respectively.

The genomic region associated with the expression of a quantitative trait is referred to as quantitative trait locus (QTL), which may contain one or more genes. QTLs have been grouped into different categories on the basis of their effect size, the effect of environment on their expression, and the manner of their action. QTL mapping is generally based on biparental populations in which the marker genotype and trait phenotype data are analyzed to detect association between the two. A large number of QTL analysis approaches have been proposed based on regression analysis, maximum likelihood parameter estimation, or Bayesian models. Single QTL mapping methods detect single QTL at a time. Multiple QTL mapping combines multiple regression analysis with simple interval mapping to include all the significant QTLs in the genetic model. Composite interval mapping can be extended to deal with data coming from multiple cross populations and for joint analysis of multiple traits. Appropriate experimental designs and QTL analysis methods are available for the detection and estimation of QTL x QTL and QTL x environment interactions. Confirmation of QTL analysis results, i.e., QTL validation, consists of confirmation of marker-QTL association and QTL position in unrelated germplasm and the assessment of effects of the genetic background on QTL expression. Homozygous lines derived from near-isogenic lines (NILs) and intercross recombinant inbred lines have been used for fine mapping of QTL regions. QTL meta-analysis attempts to integrate the results from different QTL studies to determine the “actual” number of QTLs affecting a trait and to reduce the QTL confidence intervals. QTL mapping identifies markers flanking the QTL regions, which can be used for marker-assisted selection in breeding programs. The findings from QTL mapping studies are affected by several factors like genetic properties of QTL, genetic background, size of mapping population, and effect of environment and experimental error.

Blood pressure (BP) in any human population exhibits as a continuous variable that fits a bell-shaped curve. Hypertensive individuals are those whose BP is maintained at one extreme of the curve and above a defined cutoff. Despite progress made in identifying the mechanisms underlying certain rare monogenic forms of hypertension,1,2 the etiology and pathogenesis of essential hypertension remain poorly understood. Because existing human populations are genetically heterogeneous, and because environmental factors impacting on the pathogenesis of hypertension cannot be controlled in a given population, it is difficult to identify the molecular mechanisms that transduce the sequela of essential hypertension via direct human studies.3 To alleviate the drawbacks of human investigations, animal models, especially inbred rodents, have been developed and experimentally manipulated to identify quantitative trait loci (QTLs) for BP, because major confounding environmental factors, such as diet and genetic background, can be systematically controlled. Once identified in animal models, the molecular basis may be translated into physiological understandings of essential hypertension in humans. It is with this expectation that efforts have been launched to identify the molecular basis of BP QTLs in animal models. Because the identification of individual QTLs is primarily based on their chromosome locations unbiased by, or unrestricted to, their physiological roles, positional cloning is believed to be the most efficient strategy. Before we embark on discussions regarding QTL discovery, a definition is in order. Semantic arguments abound as to exactly what a QTL, that is, a locus,4 entails. Is it 1 gene or a collection of genes? As genetic mapping progresses from a large chromosome segment to an interval of submegabase, several regions initially thought to contain 1 BP QTL5 appear to harbor >1 in each of them,6–10 whereas several other regions turned out to harbor 1 QTL as expected.11,12 …

We have extended the combined use of the "pseudo-testcross" mapping strategy and RAPD markers to map quantitative trait loci (QTLs) controlling traits related to vegetative propagation in Eucalyptus. QTL analyses were performed using two different interval mapping approaches, MAPMAKER-QTL (maximum likelihood) and QTL-STAT (non-linear least squares). A total of ten QTLs were detected for micropropagation response (measured as fresh weight of shoots, FWS), six for stump sprouting ability (measured as # stump sprout cuttings, #Cutt) and four for rooting ability (measured as % rooting of cuttings, %Root). With the exception of three QTLs, both interval-mapping methods yielded similar results in terms of QTL detection. Discrepancies in the most likely QTL location were observed between the two methods. In 75% of the cases the most likely position was in the same, or in an adjacent, interval. Standardized gene substitution effects for the QTLs detected were typically between 0.46 and 2.1 phenotypic standard deviations (σp), while differences between the family mean and the favorable QTL genotype were between 0.25 and 1.07 (σp). Multipoint estimates of the total genetic variation explained by the QTLs (89.0% for FWS, 67.1 % for#Cutt, 62.7% for %Root) indicate that a large proportion of the variation in these traits is controlled by a relatively small number of major-effect QTLs. In this cross, E. grandis is responsible for most of the inherited variation in the ability to form shoots, while E. urophylla contributes most of the ability in rooting. QTL mapping in the pseudo-testcross configuration relies on withinfamily linkage disequilibrium to establish marker/trait associations. With this approach QTL analysis is possible in any available full-sib family generated from undomesticated and highly heterozygous organisms such as forest trees. QTL mapping on two-generation pedigrees opens the possibility of using already existing families in retrospective QTL analyses to gather the quantitative data necessary for marker-assisted tree breeding.

Identification of the polymorphisms controlling quantitative traits remains a challenge for plant geneticists. Multiparent advanced generation intercross (MAGIC) populations offer an alternative to traditional linkage or association mapping populations by increasing the precision of quantitative trait loci (QTL) mapping. Here, we present the first tomato MAGIC population and highlight its potential for the valorization of intraspecific variation, QTL mapping and causal polymorphism identification. The population was developed by crossing eight founder lines, selected to include a wide range of genetic diversity, whose genomes have been previously resequenced. We selected 1536 SNPs among the 4 million available to enhance haplotype prediction and recombination detection in the population. The linkage map obtained showed an 87% increase in recombination frequencies compared to biparental populations. The prediction of the haplotype origin was possible for 89% of the MAGIC line genomes, allowing QTL detection at the haplotype level. We grew the population in two greenhouse trials and detected QTLs for fruit weight. We mapped three stable QTLs and six specific of a location. Finally, we showed the potential of the MAGIC population when coupled with whole genome sequencing of founder lines to detect candidate SNPs underlying the QTLs. For a previously cloned QTL on chromosome 3, we used the predicted allelic effect of each founder and their genome sequences to select putative causal polymorphisms in the supporting interval. The number of candidate polymorphisms was reduced from 12284 (in 800 genes) to 96 (in 54 genes), including the actual causal polymorphism. This population represents a new permanent resource for the tomato genetics community.

Kernel methods for phenotyping complex plant architecture

The study of quantitative trait effects is of great significance for molecular marker-assisted breeding. The accuracy of quantitative trait loci (QTL) mapping is the key factor affecting marker-assisted breeding, and is extremely significant. The effect of different heritability rates (10, 30, 50, 70, and 90%) on the accuracy of QTL mapping of five recombinant inbred lines (RILs) were analyzed via computer simulation. RILs display additive and epistatic genetic effects. The QTLs were analyzed using four different mapping procedures: multiple QTL model (MQM), composite interval mapping (CIM), multiple interval mapping (MIMR), and inclusive composite interval mapping (ICIM). The results revealed an increase in the QTL mapping accuracy and QTL detection power, and a decrease in the QTL interval range with the increase in heritability; conversely, an irregular number of false positive QTLs were generated. CIM and MQM only screen the additive and dominant effects; MIMR and ICIM screen the additive, dominant, and epistatic effects. The highest QTL detection power obtained using MQM and CIM was only 75%, while MIMR and ICIM showed a detection power of 100%. At heritability rates of more than 50 and less than 10%, the detection powers of the MIMR and ICIM procedures were >95 and <35%, respectively. QTL mapping has no significance at heritability rates <10%. The results of this study suggest that QTL mapping has significance at a heritability rate >30% (at least >10%) for practical marker-assisted breeding.

Local properties of the genome can bias QTL analyses

Comparing power and precision of within-breed and multibreed genome-wide association studies of production traits using whole-genome sequence data for 5 French and Danish dairy cattle breeds.

Increasing the number of mice used to calculate recombinant inbred (RI) strain means increases the accuracy of determining the phenotype associated with each genotype (strain), which in turn enhances quantitative trait locus (QTL) detection and mapping. The purpose of this paper is to examine quantitatively the effect of within-strain sample size (n) on additive QTL mapping efficiency and to make comparisons with F2 and backcross (BC) populations where each genotype is represented by only a single mouse. When 25 RI strains are used, the estimated equivalent number of F2 mice yielding the same power to detect WTLs varies inversely as a function of the heritability of the trait in the RI population (hRI2). For example, testing 25 strains with n = 10 per strain is approximately equivalent to 160 F2 mice when hRI2 = 0.2, but only 55 when hRI2 = 0.6. While increasing n is always beneficial, the gain in power as n increases is greatest when hRI2 is low and is much diminished at high hRI2 values. Thus, hRI2 is high, there is little advantage of large n, even when n approaches infinity. A cost analysis suggested that RI populations are more cost-effective than conventional selectively genotyped F2 populations at hRI2 values likely to be seen in behavioral studies. However, with DNA pooling, this advantage is greatly reduced and may be reversed depending on the values of hRI2 and n.

BackgroundDeveloping new population types based on interspecific introgressions has been suggested by several authors to facilitate the discovery of novel allelic sources for traits of agronomic importance. Chromosome segment substitution lines from interspecific crosses represent a powerful and useful genetic resource for QTL detection and breeding programs.ResultsWe built a set of 64 chromosome segment substitution lines carrying contiguous chromosomal segments of African rice Oryza glaberrima MG12 (acc. IRGC103544) in the genetic background of Oryza sativa ssp. tropical japonica (cv. Caiapó). Well-distributed simple-sequence repeats markers were used to characterize the introgression events. Average size of the substituted chromosomal segments in the substitution lines was about 10 cM and covered the whole donor genome, except for small regions on chromosome 2 and 4. Proportions of recurrent and donor genome in the substitution lines were 87.59% and 7.64%, respectively. The remaining 4.78% corresponded to heterozygotes and missing data. Strong segregation distortion was found on chromosomes 3 and 6, indicating the presence of interspecific sterility genes. To illustrate the advantages and the power of quantitative trait loci (QTL) detection using substitution lines, a QTL detection was performed for scored traits. Transgressive segregation was observed for several traits measured in the population. Fourteen QTLs for plant height, tiller number per plant, panicle length, sterility percentage, 1000-grain weight and grain yield were located on chromosomes 1, 3, 4, 6 and 9. Furthermore, a highly significant QTL controlling resistance to the Rice stripe necrosis virus was located between SSR markers RM202-RM26406 (44.5-44.8 cM) on chromosome 11.ConclusionsDevelopment and phenotyping of CSSL libraries with entire genome coverage represents a useful strategy for QTL discovery. Mapping of the RSNV locus represents the first identification of a genetic factor underlying resistance to this virus. This population is a powerful breeding tool. It also helps in overcoming hybrid sterility barriers between species of rice.

A major goal of today's biology is to understand the genetic basis of quantitative traits. This can be achieved by statistical methods that evaluate the association between molecular marker variation and phenotypic variation in different types of mapping populations. The objective of this work was to evaluate the statistical power of quantitative trait loci (QTL) detection of various multi-parental mating designs, as well as to assess the reasons for the observed differences. Our study was based on an empirical data of 20 Arabidopsis thaliana accessions, which have been selected to capture the maximum genetic diversity. The examined mating designs differed strongly with respect to the statistical power to detect QTL. We observed the highest power to detect QTL for the diallel cross with random mating design. The results of our study suggested that performing sibling mating within subpopulations of joint-linkage mapping populations has the potential to considerably increase the power for QTL detection. Our results, however, revealed that using designs in which more than two parental alleles segregate in each subpopulation increases the power even more.