Prediction of heterosis using genome-wide SNP-marker data: application to egg production traits in white Leghorn crosses

Prediction of heterosis has a long history with mixed success, partly due to low numbersof genetic markers and/or small data sets. We investigated the prediction of heterosisfor egg number, egg weight and survival days in domestic white Leghorns, using∼400 000 individuals from 47 crosses and allele frequencies on∼53 000 genome-wide single nucleotide polymorphisms (SNPs). When heterosis isdue to dominance, and dominance effects are independent of allele frequencies, heterosisis proportional to the squared difference in allele frequency (SDAF) between parental purelines (not necessarily homozygous). Under these assumptions, a linear model includingregression on SDAF partitions crossbred phenotypes into pure-line values and heterosis,even without pure-line phenotypes. We therefore used models where phenotypes of crossbredswere regressed on the SDAF between parental lines. Accuracy of prediction was determinedusing leave-one-out cross-validation. SDAF predicted heterosis for egg number and weightwith an accuracy of ∼0.5, but did not predict heterosis for survival days. Heterosispredictions allowed preselection of pure lines before field-testing, saving∼50% of field-testing cost with only 4% loss in heterosis. Accuraciesfrom cross-validation were lower than from the model-fit, suggesting that accuraciespreviously reported in literature are overestimated. Cross-validation also indicated thatdominance cannot fully explain heterosis. Nevertheless, the dominance model hadconsiderable accuracy, clearly greater than that of a general/specific combiningability model. This work also showed that heterosis can be modelled even when pure-linephenotypes are unavailable. We concluded that SDAF is a useful predictor of heterosis incommercial layer breeding.     

Mapping quantitative trait loci for egg production traits in an F2 intercross of Oh-Shamo and White Leghorn chickens

We performed quantitative trait locus (QTL) analyses for egg production traits, including age at first egg (AFE) and egg production rates (EPR) measured every 4 weeks from 22 to 62 weeks of hen age, in a population of 421 F₂ hens derived from an intercross between the Oh‐Shamo (Japanese Large Game) and White Leghorn breeds of chickens. Simple interval mapping revealed a main‐effect QTL for AFE on chromosome 1 and four main‐effect QTL for EPR on chromosomes 1 and 11 (three on chromosome 1 and one on chromosome 11) at the genome‐wide 5% levels. Among the three EPR QTL on chromosome 1, two were identified at the early stage of egg laying (26–34 weeks of hen age) and the remaining one was discovered at the late stage (54–58 weeks). The alleles at the two EPR QTL derived from the Oh‐Shamo breed unexpectedly increased the trait values, irrespective of the Oh‐Shamo being inferior to the White Leghorn in the trait. This suggests that the Oh‐Shamo, one of the indigenous Japanese breeds, is an untapped resource that is important for further improvement of current elite commercial laying chickens. In addition, six epistatic QTL were identified on chromosomes 2, 4, 7, 8, 17 and 19, where none of the above main‐effect QTL were located. This is the first example of detection of epistatic QTL affecting egg production traits. The main and epistatic QTL identified accounted for 4–8% of the phenotypic variance. The total contribution of all QTL detected for each trait to the phenotypic and genetic variances ranged from 4.1% to 16.9% and from 11.5% to 58.5%, respectively.     

DNA markers and crossbreeding scheme as means to select sires for heterosis in egg production of chickens

Genotypes for 24 microsatellite markers, dispersed across the chicken genome, were used to predict progeny performance and heterosis for egg production (number and mass) in 'layers' (egg-type chickens). These markers were used to evaluate genetic distance between each of 39 sires sampled from two-layer male-lines; Rhode Island Red (RIR) and White egg Leghorn (Leghorn), and a DNA pool of 30 randomly sampled females from a Brown-egg female line (Silver). Each sire was analysed for egg production across months in the laying period and cumulatively in each of three subperiods; onset (2 month), mid (9 month) and late (1 month). The average Reynolds' genetic distance between Leghorn sires and the Silver female line (theta;=0.6) was significantly higher than that between RIR sires and the Silver female line (theta;=0.5). Neither performance nor heterosis values in the RIR sire's daughters were associated with genetic distance values between sires and the Silver female line. On the other hand, performance as well as heterosis values of Leghorn's daughters were positively associated with genetic distance. This association was particularly evident in the mid-subperiod. If 25% of the most genetically distant Leghorn sires from the Silver female line had been selected in a single generation on the basis of DNA markers information only, average egg production of the crossbred daughters would have been improved by about nine eggs (3%). In principle, further improvement is possible if selection to increase genetic distance between the parental lines is carried on.     

Predicting unobserved phenotypes for complex traits from whole-genome SNP data

Genome-wide association studies (GWAS) for quantitative traits and disease in humans and other species have shown that there are many loci that contribute to the observed resemblance between relatives. GWAS to date have mostly focussed on discovery of genes or regulatory regions habouring causative polymorphisms, using single SNP analyses and setting stringent type-I error rates. Genome-wide marker data can also be used to predict genetic values and therefore predict phenotypes. Here, we propose a Bayesian method that utilises all marker data simultaneously to predict phenotypes. We apply the method to three traits: coat colour, %CD8 cells, and mean cell haemoglobin, measured in a heterogeneous stock mouse population. We find that a model that contains both additive and dominance effects, estimated from genome-wide marker data, is successful in predicting unobserved phenotypes and is significantly better than a prediction based upon the phenotypes of close relatives. Correlations between predicted and actual phenotypes were in the range of 0.4 to 0.9 when half of the number of families was used to estimate effects and the other half for prediction. Posterior probabilities of SNPs being associated with coat colour were high for regions that are known to contain loci for this trait. The prediction of phenotypes using large samples, high-density SNP data, and appropriate statistical methodology is feasible and can be applied in human medicine, forensics, or artificial selection programs.     

A genome-wide SNP scan reveals novel loci for egg production and quality traits in white leghorn and brown-egg dwarf layers

Availability of the complete genome sequence as well as high-density SNP genotyping platforms allows genome-wide association studies (GWAS) in chickens. A high-density SNP array containing 57,636 markers was employed herein to identify associated variants underlying egg production and quality traits within two lines of chickens, i.e., White Leghorn and brown-egg dwarf layers. For each individual, age at first egg (AFE), first egg weight (FEW), and number of eggs (EN) from 21 to 56 weeks of age were recorded, and egg quality traits including egg weight (EW), eggshell weight (ESW), yolk weight (YW), eggshell thickness (EST), eggshell strength (ESS), albumen height(AH) and Haugh unit(HU) were measured at 40 and 60 weeks of age. A total of 385 White Leghorn females and 361 brown-egg dwarf dams were selected to be genotyped. The genome-wide scan revealed 8 SNPs showing genome-wise significant (P<1.51E-06, Bonferroni correction) association with egg production and quality traits under the Fisher's combined probability method. Some significant SNPs are located in known genes including GRB14 and GALNT1 that can impact development and function of ovary, but more are located in genes with unclear functions in layers, and need to be studied further. Many chromosome-wise significant SNPs were also detected in this study and some of them are located in previously reported QTL regions. Most of loci detected in this study are novel and the follow-up replication studies may be needed to further confirm the functional significance for these newly identified SNPs.     

Quantitative trait loci segregating in crosses between New Hampshire and White Leghorn chicken lines: I. egg production traits

A genome scan was performed to detect chromosomal regions that affect egg production traits in reciprocal crosses between two genetically and phenotypically extreme chicken lines: the partially inbred line New Hampshire (NHI) and the inbred line White Leghorn (WL77). The NHI line had been selected for high growth and WL77 for low egg weight before inbreeding. The result showed a highly significant region on chromosome 4 with multiple QTL for egg production traits between 19.2 and 82.1 Mb. This QTL region explained 4.3 and 16.1% of the phenotypic variance for number of eggs and egg weight in the F₂ population, respectively. The egg weight QTL effects are dependent on the direction of the cross. In addition, genome‐wide suggestive QTL for egg weight were found on chromosomes 1, 5, and 9, and for number of eggs on chromosomes 5 and 7. A genome‐wide significant QTL affecting age at first egg was mapped on chromosome 1. The difference between the parental lines and the highly significant QTL effects on chromosome 4 will further support fine mapping and candidate gene identification for egg production traits in chicken.     

Quantitative trait loci analysis of egg and meat production traits in a red junglefowlxWhite Leghorn cross

Egg and production traits are of considerable economic importance in chickens. Using a White Leghorn x red junglefowl F(2) intercross, standard production measures of liver weight and colour, egg size, eggshell thickness, egg taste and meat quality were taken. A total of 160 markers covering 29 autosomes and the Z chromosome were genotyped on 175-243 individuals, depending on the trait under consideration. A total of nine significant quantitative trait loci (QTL) and three suggestive QTL were found on chicken chromosomes 1, 2, 4, 5, 7, 8, 10, 12, E47W24 and E22C19W28.     

A genome-wide association study for feed efficiency-related traits in a crossbred pig population

Feed efficiency (FE) is one of the most important traits in pig production. However, it is difficult and costly to measure it, limiting the collection of large amount of data for an accurate selection for better FE. Therefore, the identification of single-nucleotide polymorphisms (SNPs) associated with FE-related traits to be used in the genetic evaluation is of great interest of pig breeding programs for increasing the prediction accuracy and the genetic progress of these traits. The objective of this study was to identify SNPs significantly associated with FE-related traits: average daily gain (ADG), average daily feed intake (ADFI) and feed conversion ratio (FCR). We also aimed to identify potential candidate genes for these traits. Phenotypic information recorded on a population of 2386 three-way crossbreed pigs that were genotyped for 51 468 SNPs was used. We identified three loci of quantitative trait (QTL) regions associated with ADG and three QTL regions associated with ADFI; however, no significant association was found for FCR. A false discovery rate (FDR) ≤ 0.005 was used as the threshold for declaring an association as significant. The QTL regions associated with ADG on Sus scrofa chromosome (SSC) 1 were located between 177.01 and 185.47 Mb, which overlaps with the QTL regions for ADFI on SSC1 (173.26 and 185.47 Mb). The other QTL region for ADG was located on SSC12 (2.87 and 3.22 Mb). The most significant SNPs in these QTL regions explained up to 3.26% of the phenotypic variance of these traits. The non-identification of genomic regions associated with FCR can be explained by the complexity of this trait, which is a ratio between ADG and ADFI. Finally, the genes CDH19, CDH7, RNF152, MC4R, PMAIP1, FEM1B and GAA were the candidate genes found in the 1 Mb window around the QTL regions identified in this study. Among them, the MC4R gene (SSC1) has a well-known function related to ADG and ADFI. In this study, we identified three QTL regions for ADG (SSC1 and SSC12) and three for ADFI (SSC1). These regions were previously described in purebred pig populations; however, to our knowledge, this is the first study to confirm the relevance of these QTL regions in a crossbred pig population. The potential use of the SNPs and genes identified in this study in prediction models that combine genomic selection and marker-assisted selection should be evaluated for increasing the prediction accuracy of these traits in this population.     

Detecting inbreeding depression for reproductive traits in Iberian pigs using genome-wide data

Background

The current availability of genotypes for very large numbers of single nucleotide polymorphisms (SNPs) is leading to more accurate estimates of inbreeding coefficients and more detailed approaches for detecting inbreeding depression. In the present study, genome-wide information was used to detect inbreeding depression for two reproductive traits (total number of piglets born and number of piglets born alive) in an ancient strain of Iberian pigs (the Guadyerbas strain) that is currently under serious danger of extinction.

Methods

A total of 109 sows with phenotypic records were genotyped with the PorcineSNP60 BeadChip v1. Inbreeding depression was estimated using a bivariate animal model in which the inbreeding coefficient was included as a covariate. We used two different measures of genomic inbreeding to perform the analyses: inbreeding estimated on a SNP-by-SNP basis and inbreeding estimated from runs of homozygosity. We also performed the analyses using pedigree-based inbreeding.

Results

Significant inbreeding depression was detected for both traits using all three measures of inbreeding. Genome-wide information allowed us to identify one region on chromosome 13 associated with inbreeding depression. This region spans from 27 to 54 Mb and overlaps with a previously detected quantitative trait locus and includes the inter-alpha-trypsin inhibitor gene cluster that is involved with embryo implantation.

Conclusions

Our results highlight the value of high-density SNP genotyping for providing new insights on where genes causing inbreeding depression are located in the genome. Genomic measures of inbreeding obtained on a SNP-by-SNP basis or those based on the presence/absence of runs of homozygosity represent a suitable alternative to pedigree-based measures to detect inbreeding depression, and a useful tool for mapping studies. To our knowledge, this is the first study in domesticated animals using the SNP-by-SNP inbreeding coefficient to map specific regions within chromosomes associated with inbreeding depression.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-014-0081-5) contains supplementary material, which is available to authorized users.     

Predicting allele frequencies in DNA pools using high density SNP genotyping data

To evaluate the ability to use DNA pools with the Illumina Infinium genotyping platform, two sets of gradient pools were created using two pairs of highly inbred chicken lines. Replicate pools containing 0%, 10%, 20%, 40%, 60%, 80%, 90% and 100% of DNA from line A vs. B or line C vs. D were created, for a total of 28 pools. All pools were genotyped for 12 046 SNPs. Three frequency estimation methods proposed in the literature (standard, heterozygote‐corrected and normalized) were compared with three alternate methods proposed herein based on mean square error (MSE), bias and variance of estimated vs. true allele frequencies and the fit of regression of estimated on true frequencies. The three new methods had average square root MSE of 4.6%, 4.6% and 4.7% compared to 5.2%, 5.5% and 11.2% for the three literature methods. Average absolute biases of the literature methods were 2.4%, 2.7% and 8.2% compared to 2.4% for all new methods. Standard deviations of estimates were also smaller for the new methods, at 3.1%, 3.2% and 3.2% compared to 3.5%, 4.0% and 5.0% for previously reported methods. In conclusion, intensity data from the Illumina Infinium Assay can be efficiently used to estimate allele frequencies in pools, in particular using any of the new methods proposed herein.     

Genome‐wide association study reveals novel variants for growth and egg traits in Dongxiang blue‐shelled and White Leghorn chickens

This study was designed to investigate the genetic basis of growth and egg traits in Dongxiang blue‐shelled chickens and White Leghorn chickens. In this study, we employed a reduced representation sequencing approach called genotyping by genome reducing and sequencing to detect genome‐wide SNPs in 252 Dongxiang blue‐shelled chickens and 252 White Leghorn chickens. The Dongxiang blue‐shelled chicken breed has many specific traits and is characterized by blue‐shelled eggs, black plumage, black skin, black bone and black organs. The White Leghorn chicken is an egg‐type breed with high productivity. As multibreed genome‐wide association studies (GWASs) can improve precision due to less linkage disequilibrium across breeds, a multibreed GWAS was performed with 156 575 SNPs to identify the associated variants underlying growth and egg traits within the two chicken breeds. The analysis revealed 32 SNPs exhibiting a significant genome‐wide association with growth and egg traits. Some of the significant SNPs are located in genes that are known to impact growth and egg traits, but nearly half of the significant SNPs are located in genes with unclear functions in chickens. To our knowledge, this is the first multibreed genome‐wide report for the genetics of growth and egg traits in the Dongxiang blue‐shelled and White Leghorn chickens.     

Application of genome-wide SNP data for uncovering pairwise relationships and quantitative trait loci

The genetic analysis of quantitative traits in humans is changing as a result of the availability of whole-genome SNP data. Heritability analysis can make use of actual genetic sharing between pairs of individuals estimated from the genotype data, rather than the expected genetic sharing implied by their family relationship. This could provide more accurate heritability estimates and help to overcome the equal environment assumption. Quantitative trait locus (QTL) linkage mapping can make use of local genetic sharing inferred from very dense local genotype data from pedigree members or individuals not previously known to be related. This approach may be particularly suited for detecting loci that contain rare variants with major effect on the phenotype. Finally, whole-genome SNP data can be used to measure the genetic similarity between individuals to provide matched sets for association studies, in order to avoid spurious association from population stratification.     

Genome-wide association study of milk production traits in a crossbred dairy sheep population using three statistical models

Milk production is one of the most important characteristics of dairy sheep, and the identification of genes affecting milk production traits is critical to understanding the genetics and improve milk production in future generations. Three statistical techniques, namely GWAS, ridge-regression BLUP and BayesC , were used to identify SNPs in significant association with three milk production traits (milk yield, fat yield and protein yield) in a crossbred dairy sheep population. The results suggested that chromosomes 1, 3, 4, 5, 7 and 11 were likely to harbor genes important to milk production because these chromosomes had the greatest top-100-SNP variance contributions on the three milk production traits. The GWAS analysis identified between 74 and 288 genome-wide significant SNP (P < 0.05) whereas the BayesCπ model revealed between six and 63 SNPs, each with >95% posterior probability of inclusion as having a non-zero association effect on at least one of the three milk production traits. Positional candidate genes for milk production in sheep were searched, based on the sheep genomic assembly OAR version 3.1, such as those which map position coincided with or was located within 0.1 Mbp of a genome-wide suggestive or significant SNP. These identified SNPs and candidate genes supported some previous findings and also added new information about genetic markers for genetic improvement of lactation in dairy sheep, but keeping in mind that the majority of these positional candidate genes are not necessarily true causative loci for these traits and future validations are thus necessary.     

A genome-wide association study using international breeding-evaluation data identifies major loci affecting production traits and stature in the Brown Swiss cattle breed

Background

Fusarium graminearum sensu stricto (s.s.) is an ubiquitous pathogen of cereals. The economic impact of Fusarium head blight (FHB) is characterized by crop losses and mycotoxin contamination. Our objective was to associate SNP diversity within candidate genes with phenotypic traits. A total of 77 F. graminearum s.s. isolates was tested for severity of fungal infection (= aggressiveness) and deoxynivalenol (DON) production in an inoculated field experiment at two locations in each of two years. For seven genes known to control fungal growth (MetAP1, Erf2) or DON production (TRI1, TRI5, TRI6 TRI10 and TRI14) single nucleotides polymorphic sites (SNPs) were determined and evaluated for the extent of linkage disequilibrium (LD). Associations of SNPs with both phenotypic traits were tested using linear mixed models.

Results

Decay of LD was in most instances fast. Two neighboring SNPs in MetAP1 and one SNP in Erf2 were significantly (P < 0.05) associated with aggressiveness explaining proportions of genotypic variance (p _G ) of 25.6%, 0.5%, and 13.1%, respectively. One SNP in TRI1 was significantly associated with DON production (p _G = 4.4).

Conclusions

We argue that using the published sequence information of Fusarium graminearum as a template to amplify comparative sequence parts of candidate genes is an effective method to detect quantitative trait loci. Our findings underline the potential of candidate gene association mapping approaches to identify functional SNPs underlying aggressiveness and DON production for F. graminearum s.s populations.     

Intake, growth and carcass traits in male progeny of sires differing in genetic merit for beef production

Validation of economic indexes under a controlled experimental environment, can aid in their acceptance and use as breeding tools to increase herd profitability. The objective of this study was to compare intake, growth and carcass traits in bull and steer progeny of high and low ranking sires, for genetic merit in an economic index. The Beef Carcass Index (BCI; expressed in euro (€) and based on weaning weight, feed intake, carcass weight, carcass conformation and fat scores) was generated by the Irish Cattle Breeding Federation as a tool to compare animals on genetic merit for the expected profitability of their progeny at slaughter. A total of 107 male suckler herd progeny, from 22 late-maturing 'continental' beef sires of high (n = 11) or low (n = 11) BCI were compared under either a bull or steer production system, and slaughtered at approximately 16 and 24 months of age, respectively. All progeny were purchased after weaning at approximately 6 to 8 months of age. Dry matter (DM) intake and live-weight gain in steer progeny offered grazed grass or grass silage alone, did not differ between the two genetic groups. Similarly, DM intake and feed efficiency did not differ between genetic groups during an ad libitum concentrate-finishing period on either production system. Carcasses of progeny of high BCI sires were 14 kg heavier (P < 0.05) than those of low BCI sires. In a series of regression analyses, increasing sire BCI resulted in increases in carcass weight (P < 0.01) and carcass conformation (P = 0.051) scores, and decreases in carcass fat (P < 0.001) scores, but had no effect on weaning weight or DM intake of the progeny. Each unit increase in sire expected progeny difference led to an increase in progeny weaning weight, DM intake, carcass weight, carcass conformation score and carcass fat score of 1.0 (s.e. = 0.53) kg, 1.1 (s.e. = 0.32) kg, 1.3 (s.e. = 0.31) kg, 0.9 (s.e. = 0.32; scale 1 to 15) and 1.0 (s.e. = 0.25; scale 1 to 15), respectively, none of which differed from the theoretical expectation of unity. The expected difference in profitability at slaughter between progeny of the high and low BCI sires was €42, whereas the observed phenotypic profit differential of the progeny was €53 in favour of the high BCI sires. Results from this study indicate that the BCI is a useful tool in the selection of genetically superior sires, and that actual progeny performance under the conditions of this study is within expectations for both bull and steer beef production systems.     

Parallel and serial computing tools for testing single-locus and epistatic SNP effects of quantitative traits in genome-wide association studies

Background  

Genome-wide association studies (GWAS) using single nucleotide polymorphism (SNP) markers provide opportunities to detect epistatic SNPs associated with quantitative traits and to detect the exact mode of an epistasis effect. Computational difficulty is the main bottleneck for epistasis testing in large scale GWAS.     

A genome-wide association study of production traits in a commercial population of Large White pigs: evidence of haplotypes affecting meat quality

Background

Numerous quantitative trait loci (QTL) have been detected in pigs over the past 20 years using microsatellite markers. However, due to the low density of these markers, the accuracy of QTL location has generally been poor. Since 2009, the dense genome coverage provided by the Illumina PorcineSNP60 BeadChip has made it possible to more accurately map QTL using genome-wide association studies (GWAS). Our objective was to perform high-density GWAS in order to identify genomic regions and corresponding haplotypes associated with production traits in a French Large White population of pigs.

Methods

Animals (385 Large White pigs from 106 sires) were genotyped using the PorcineSNP60 BeadChip and evaluated for 19 traits related to feed intake, growth, carcass composition and meat quality. Of the 64 432 SNPs on the chip, 44 412 were used for GWAS with an animal mixed model that included a regression coefficient for the tested SNPs and a genomic kinship matrix. SNP haplotype effects in QTL regions were then tested for association with phenotypes following phase reconstruction based on the Sscrofa10.2 pig genome assembly.

Results

Twenty-three QTL regions were identified on autosomes and their effects ranged from 0.25 to 0.75 phenotypic standard deviation units for feed intake and feed efficiency (four QTL), carcass (12 QTL) and meat quality traits (seven QTL). The 10 most significant QTL regions had effects on carcass (chromosomes 7, 10, 16, 17 and 18) and meat quality traits (two regions on chromosome 1 and one region on chromosomes 8, 9 and 13). Thirteen of the 23 QTL regions had not been previously described. A haplotype block of 183 kb on chromosome 1 (six SNPs) was identified and displayed three distinct haplotypes with significant (0.0001 < P < 0.03) associations with all evaluated meat quality traits.

Conclusions

GWAS analyses with the PorcineSNP60 BeadChip enabled the detection of 23 QTL regions that affect feed consumption, carcass and meat quality traits in a LW population, of which 13 were novel QTL. The proportionally larger number of QTL found for meat quality traits suggests a specific opportunity for improving these traits in the pig by genomic selection.