Genomewide study and validation of markers associated with production traits in German Landrace boars E. M. Strucken, A. O. Schmitt, U. Bergfeld, I. Jurke, M. Reissmann and G. A. Brockmann J ANIM SCI 2014, 92:1939-1944. doi: 10.2527/jas.2013-7247 originally published online March 26, 2014
The online version of this article, along with updated information and services, is located on the World Wide Web at: http://www.journalofanimalscience.org/content/92/5/1939
www.asas.org
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
Genomewide study and validation of markers associated with production traits in German Landrace boars1 E. M. Strucken,* A. O. Schmitt,† U. Bergfeld,‡ I. Jurke,* M. Reissmann,* and G. A. Brockmann*2 *Humboldt-Universität zu Berlin, Breeding Biology and Molecular Genetics, Invalidenstr. 42, 10115 Berlin, Germany; †Faculty of Science and Technology, Universitätsplatz 5, 39100 Bozen-Bolzano, Italy; and ‡Sächsisches Landesamt für Umwelt, Landwirtschaft und Geologie, Postfach 54 01 37, 01311 Dresden, Germany
Abstract: We present results from a genomewide association study (GWAS) and a single-marker association study. The GWAS was performed with the Illumina PorcineSNP60 BeadChip from which 5 markers were selected for a validation analysis. Genetic effects were estimated for feed intake, weight gain, and traits of fat and muscle tissue in German Landrace boars kept on performance test stations. The GWAS was performed in a population of 288 boars and the validation study for another 432 boars. No statistically significant effect was found in the GWAS after adjusting for multiple testing. Effects of 2 markers, which were significant genomewide before correction for multiple testing (P < 0.00005), could be confirmed in the validation study. The major allele of marker ALGA0056781 on SSC1 was positively associated with both higher weight gain and fat deposition. The effect on live-weight gain was 2.25 g/d in the GWAS (P = 0.0003) and 3.73 g/d in the validation study (P = 0.01) and for back fat thickness was 0.15 mm in the GWAS (P < 0.0001) and
0.20 mm in the validation study (P = 0.02). The marker had similar effects on test-day weight gain (GWAS: 3.85 g/d, P = 0.001; validation study: 6.80 g/d, P = 0.003) and back fat area (GWAS: 0.27 cm2, P < 0.0001; validation study: 0.35 cm2, P = 0.03). Marker ASGA0056782 on SSC13 was associated with live-weight gain. The major allele had negative effects in both studies (GWAS: –4.88 g/d, P < 0.0001; validation study: –3.75 g/d, P = 0.02). The effects of these 2 markers would have been excluded based on the GWAS alone but were shown to be significantly trait associated in the validation study indicating a false-negative result. The G protein-coupled receptor 126 (GPR126) gene approximately 200 kb downstream of marker ALGA0001781 was shown to be associated with human height and therefore might explain the association with weight gain in pigs. Several traits were affected in an economically desired direction by the minor allele of the markers, pointing to the possibility of improvement through further selection.
Key words: dominance, feed intake, genomic selection, GPR126, piglets, weight gain © 2014 American Society of Animal Science. All rights reserved. J. Anim. Sci. 2014.92:1939–1944 doi:10.2527/jas2013-7247 INTRODUCTION The pig industry uses terminal cross-breeding systems to increase production by combining lines of breeds with favorable paternal (growth) and maternal (reproduction, litter size, and mothering ability) characteristics.
1We thank Dr. Uwe Müller from the Humboldt-Universität zu Berlin for his help with the statistical analyses and Dr. Ulf Müller from the Sächsische Landesanstalt für Umwelt, Landwirtschaft und Geologie for providing phenotypic data. The project was supported by the Saxon Ministry for Environment and Agriculture. 2Corresponding author: [email protected] Received October 11, 2013. Accepted March 4, 2014.
While cross-breeding can exploit complementarity, permanent and cumulative improvement can only be achieved by selection. Until recently, selective breeding programs have mainly used conventional breeding values estimated from performance records of candidates for selection and their relatives. With the advent of genomewide markers, higher accuracies and shorter generation intervals may be achieved using genomic selection (e.g., Goddard and Hayes, 2007; Hayes et al., 2009). Obstacles impeding use of genomewide data in pig breeding are high costs and small calibration populations. Large calibration populations are needed to accurately estimate marker effects and verify genomic breeding values with highly accurate estimated breeding values (Tribout et al., 2012). However, large
1939 Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
1940
Strucken et al.
Table 1. Population description for corrected trait measurements in the genomewide association study (GWAS) and trait deviations in the validation study Trait Feed intake,1 kg Live weight gain, g/d Test-day weight gain, g/d Back muscle area, cm2 Back muscle thickness, mm Back fat area,1 cm2 Back fat thickness,1 mm Intramuscular fat content, % Drip loss,1 % pH 45 min postmortem
n 265 271 272 – 150 241 271 160 213 130
Mean ± SD 201 ± 43 689 ± 40 959 ± 70 – 53 ± 2 18 ± 4 13 ± 2 1.0 ± 0.2 4.0 ± 1.4 6.4 ± 0.15
GWAS Minimum 69 558 678 – 43 5 6 0.5 0.6 6.0
Maximum 330 841 1,247 – 59 36 23 1.8 9.2 6.9
n 433 433 433 433 – 433 433 433 433 433
Validation Study Mean ± SD Minimum 1.71 ± 6.2 –28 7.76 ± 17.6 –54 15.67 ± 28.2 –72 0.01 ± 2.38 –9.1 – – 0.75 ± 1.8 –7.5 0.17 ± 0.9 –5.8 0.003 ± 0.11 –0.34 –0.10 ± 0.45 –2.59 0.01 ± 0.06 –0.25
Maximum 30 90 116 14.3 – 7.5 5.0 0.84 1.68 0.20
1Trait deviations in the validation study are positive if feed intake, back fat area and thickness, and drip loss are lower than average because this is the favorable phenotype. The mean phenotypic value of the boars in the validation study deviated from the mean value since the whole German Landrace population was used for the correction of trait deviations for environmental effects.
calibration populations are not available for most pig breeds. Therefore, reported marker effects from genomewide association studies (GWAS) should be rigorously verified in different populations. Two problems arise from GWAS with small population sizes, the first one being an increased chance of false positive results in GWAS if multiple testing is not adequately considered by correction of the significance levels for marker effects or by permutation testing. Secondly, reducing the significance threshold has an inherent risk of introducing false negative results, especially if the population is small (i.e., low statistical power). As most methods to account for multiple testing are rather stringent, implementation may result in exclusion of true effects (Strucken et al., 2012). Our objective was to estimate GWAS marker effects in a small German Landrace population and subsequently verify the effects in another different population to ascertain the frequencies of type I and type II errors. MATERIAL AND METHODS Phenotypic Data Animals were anesthetized and slaughtered according to German animal protection laws (TierschutzSchlachtverordnung, BGBl. I S. 2982). Phenotypic data were collected within the national performance tests for German Landrace boars from 2009 to 2011 and provided by the breeding organization after correction for environmental effects (Zentralverband der Deutschen Schweineproduktion e.V., Bonn, Germany; www.zdsbonn.de/; Table 1.) All animals were kept under similar conditions on 2 national test stations and slaughtered at 105 kg BW. For the GWAS, 288 animals were selected for low pedigree relationships based on their kin-
ship coefficient. Sire and dam information was available for 268 animals showing that the population descended from 30 sires and 205 dams resulting in 105 full-sibs and 155 paternal and 2 maternal half-sibs. For the validation study, another 432 boars were available that descended from 56 sires and 233 dams resulting in 206 full-sibs and 132 paternal and 13 maternal half-sibs. Sires and dams overlapped by 45 and 43%, respectively, between the populations used in the GWAS and validation study. For all animals in the GWAS and in the validation study, the same phenotypes were measured, except for the back muscle, where the thickness was measured for the GWAS and the area for the validation study. For the GWAS, we used own performance data corrected for environmental effects; correction for environmental effects was done on the basis of animals that were in the test station simultaneously with the animals chosen for the GWAS (584 animals). For the validation study, we used trait deviations (TD) after correction for environmental effects, considering the kinship of the animals and the direction of effect for breeding (favorable or unfavorable). Trait deviations give the deviation of the phenotype of a particular boar in comparison to the population mean. All available animals with phenotypic data of the German Landrace population were used to correct for environmental effects in the validation study. Traits of the GWAS and the validation study were corrected for covariance between traits (live weight for back fat thickness and back muscle thickness, warm carcass weight for intramuscular fat content, back fat area, back muscle area, and drip loss, and live weight at test day for feed intake and test day gain) and fixed effect factors such as farm, year, quarter, sex, slaughter house, test number, test station, test group, ultrasound device, and measuring person. Furthermore, to correct for external factors, TD were adjusted to the direction of effects for breeding,
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
1941
Marker association in Landrace boars
Table 2. Information about most significant markers selected from the genomewide association study (GWAS) for the validation study (genome build Sscrofa 10.2) Traits in GWAS1 Major allele effect FE 0.27
Uncorrected2 P-value 2.4 × 10–5 (0.46)
Marker ALGA0001781 (rs81350882)
SSC 1
Position, bp 25,798,122
SNP A/G
Major allele A
ASGA0039629 (rs81403471) ASGA0056782 (rs81443902) H3GA0049322 (rs80901871) ASGA0077474 (rs81466984)
8
120,311,332
A/G
A
IMF
0.02
2.3 × 10–5 (0.51)
13
27,499,096
A/G
A
LWG
–4.88
2.3 × 10–6 (0.60)
17
56,149,830
A/C
C
DL
0.20
5.2 × 10–5 (0.51)
17
57,039,720
A/G
G
FI
–2.13
1.9 × 10–5 (0.45)
1FE
= fat area (cm2); IMF = intramuscular fat content (%); LWG = life weight gain (g/d); DL = drip loss (%); FI = feed intake (kg). are uncorrected for multiple testing; P-values after correction for multiple testing are given in parentheses.
2P-values
such that a positive value represents a desirable breeding direction and a negative value represents an undesirable breeding direction. As such for the traits feed intake, fat area, back fat thickness, and drip loss, a higher performance than the population mean was assigned a negative value to acknowledge an undesirable breeding direction and should be kept in mind when interpreting the results between the 2 studies. Feed intake and test-day weight gain were recorded for the period in which animal live weight grew from 30 to 105 kg. Live-weight gain represents the average daily weight gain from birth to slaughter. Back muscle and back fat thickness were measured on the live animal via ultrasound above the 13th thoracic vertebra. Back muscle and back fat area (cm2) as well as intramuscular fat content were measured at the loin between the 13th and 14th thoracic vertebra of the longissimus dorsi after slaughter. Intramuscular fat content was measured by chemical analysis. Further meat traits included the drip loss of the loin collected over 24 h at 4°C after slaughter and the pH value of the loin 45 min postmortem. Finally, all traits were tested for normal distribution using the Kolmogorov-Smirnov test. Intramuscular fat content was not normally distributed in the GWAS population and was therefore log-transformed. Genetic Data Genotyping of Animals in the Genomewide Association Study. Animals were genotyped with the 60K PorcineSNP60 BeadChip (Illumina, San Diego, CA). Quality control included a minor allele frequency of >0.01 and a call rate per marker and animal of >10%. In total 283 animals and 53,536 markers passed quality control. In addition, 16 animals were excluded because their corresponding phenotypic data was unavailable. No markers in the GWAS reached a significant P-value (α = 0.05) after permutation correction for multiple testing. Nevertheless, 5 markers were selected based
on lowest uncorrected P-values to evaluate whether the previous effects are valid (Table 2). Sus scrofa chromosomes 1, 8, and 13 each contained 1 marker, while 2 markers were located on SSC17 (Table 2). Genotyping of Animals in the Validation Study. Deoxyribonucleic acid was isolated from ear tissue collected from each boar. The Kompetitive allele-specific polymerase chain reaction (KASPar) SNP genotyping system (KBioscience Ltd., Maple Park, Hoddesdon, UK) was used to identify the genotypes of each animal for every marker. Assays were performed according to the protocol published by Kreuzer et al. (2013) and performed with allele-specific primers in a polymerase chain reaction (Supplementary Table 1). Statistical Analyses The GWAS was performed with the GenABEL package in R (Aulchenko et al., 2007). A fast score test, as implemented in the package, was used for the association analysis using principal components based on genomic data to account for population stratification (Price et al., 2006). Additionally, the lambda factor was used to correct P-values if population stratification was still indicated after correction. Multiple testing thresholds were determined based on 10,000 permutations. The validation study with the 5 selected markers was performed using the SAS package (version 9.3; SAS Inst. Inc., Cary, NC). A mixed model was fitted for the association between the SNP and the TD: yijkl = μ + SNPi + sirej(damk) + eijkl, in which yijkl is the TD of each animal, μ is the mean of the TD of the 432 boars, SNPi is the genotype of the ith marker (0, 1 or 2 for the additive genetic model and 0 or 1 for dominance model), damk nested within sirej was used to account for full- and half-sib relationships, and eijkl is the random residual error. Genotype was fitted as a numeric
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
1942
–0.06 (0.91)
–1.91 (0.16) 0.23 (0.90) –0.11 (0.86) 1.29 (0.33)
1.94 (0.15) –2.67 (0.25) –7.18 (0.0007) –3.07 (0.23) –4.89 (0.05) –0.34 (0.90) –0.24 (0.83) 1.34 (0.50)
3.73 (0.01) 0.99 (0.16) –2.84 (0.06) –4.88 (2.3 × 10–6) –3.75 (0.02)
ASGA0077474
H3GA0049322
ASGA0056782
ASGA0039629
1FI = feed intake (kg); LWG = live weight gain (g/d); TWG = test-day weight gain (g/d); FE = fat area (cm2); FT = fat thickness (mm); IMF = intramuscular fat content (%); DL = drip loss (%), PH = pH value 45 min postmortem. Back muscle area and thickness were not significantly associated in either studies and, therefore, are not shown. 2Effects of the validation study have been multiplied by –1 to make comparisons easier because a positive trait deviation value for feed intake, fat area and thickness, and drip loss indicates values that are lower than the average (see also Material and Methods). Bold typescript identifies effects that led to the selection of markers in the GWAS and validated significant effects in the validation study.
PH 0.03 (0.10) –0.01 (0.001) –0.01 (0.48) 0.005 (0.25) –0.04 (0.21) 0.005 (0.28) –0.03 (0.40) 0.001 (0.85) –0.002 (0.90) –0.002 (0.52) DL2 –0.01 (0.63) 0.08 (0.05) –0.003 (0.91) 0.005 (0.91) 0.0007 (0.99) –0.03 (0.50) 0.20 (5.2 × 10–5) 0.05 (0.31) 0.04 (0.05) –0.02 (0.50) IMF 0.01 (0.001) –0.01 (0.35) 0.02 (2.3 × 10–5) 0.01 (0.15) 0.002 (0.76) 0.005 (0.67) –0.008 (0.34) –0.004 (0.73) –0.002 (0.56) –0.001 (0.89) FT2 0.15 (3.5 ×10–5) 0.20 (0.02) 0.0005 (0.99) –0.25 (0.007) –0.12 (0.07) –0.28 (0.006) –0.11 (0.15) –0.12 (0.27) –0.10 (0.004) 0.12 (0.15) FE2 0.27 (2.4 × 10–5) 0.35 (0.03) –0.07 (0.36) –0.23 (0.17) –0.09 (0.44) –0.33 (0.07) –0.32 (0.02) –0.23 (0.25) –0.21 (0.002) 0.08 (0.59) TWG 3.85 (0.001) 6.80 (0.003) LWG 2.25 (0.0003)
FI2 1.15 (0.03) 0.80 (0.14) 0.32 (0.60) –0.71 (0.20) –0.72 (0.45) –1.24 (0.04) –0.14 (0.90) –0.62 (0.35) –2.13 (1.9 × 10–5)
Both studies showed comparable genotype and allele frequencies with both having the same major allele. The overall minor allele frequency in the GWAS was 0.27 ± 0.14 after quality control. The 5 markers that were selected for the validation study had a minor allele frequency (B) of >0.20 in the GWAS and >0.18 in the validation study (Supplementary Table 2). All selected markers were in Hardy-Weinberg equilibrium in both studies. While uncorrected P-values were as low as 0.00002 (Table 2), none of the marker effects of the GWAS were significant after correction for multiple testing. Therefore, all further comparisons were made for SNP that had the lowest uncorrected P-value for the examined traits in the GWAS. In the GWAS, the most significant effects of the chosen markers were for the traits fat deposition, weight gain, drip loss, and feed intake (Table 2). These genomewide allele effects were reproduced in the validation study for 2 markers: marker ALGA0001781 on SSC1 had validated effects on live-weight gain and testday weight gain as well as on back fat area and thickness, and marker ASGA0056782 on SSC13 showed effects on live-weight gain (Table 3). Allele effects were larger in the validation study; for example, the major allele of marker ALGA0001781 increased live-weight and test-day weight gain by 3.73 and 6.80 g/d, respectively, compared to 2.25 and 3.85 g/d in the GWAS, whereas the major allele of marker ASGA0056782 had a less negative impact on liveweight gain in the validation study with only –3.75 g/d compared to –4.88 g/d in the GWAS (Table 3). Estimates for genotype effects and the least square mean differences between the genotype classes showed a significantly higher weight gain the more copies of the major allele of marker ALGA0001781 were present in the genotype (Fig. 1). Marker ASGA0056782 indicated a heterosis effect for live-weight gain, but differences were only significant between the AA and AB genotypes (Fig. 1). The dominance effect of live weight gain was 5.59 ± 1.99 units (P = 0.006). Back fat thickness decreased the more copies of the major allele of marker ASGA0056782 were present but significantly increased for the major allele of marker ALGA0001781 (Fig. 1). Additionally, marker ASGA0056782 showed a positive
Study GWAS Validation GWAS Validation GWAS Validation GWAS Validation GWAS Validation
RESULTS
Marker ALGA0001781
variable in the additive allele model and as a categorical variable when differences between genotype classes were estimated. The Bonferroni correction was applied to account for multiple testing when differences between leastsquare means of genotype classes were tested. Neither the GWAS nor the validation study included further environmental effects because the traits used were already corrected by the breeder organization for masking factors as described in the phenotypic data.
Table 3. Additive effects of major alleles of tested markers in the genomewide association study (GWAS) and the validation study. Uncorrected P-values for GWAS and corrected P-values for the validation study are given in parentheses1
Strucken et al.
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
Marker association in Landrace boars
Figure 1. Effect plots for selected markers and production traits from the validation study (A is the major allele). *P < 0.01; +P < 0.05. FT = fat thickness; LWG = life weight gain; TWG = test-day weight gain. †For a better understanding, effects for fat thickness have been multiplied by –1 because a positive trait deviation indicates values that are lower than the average.
dominance effect of 0.12 ± 0.12 units (P = 0.02) for back fat thickness. Even though marker ASGA0077474 on SSC17 showed significant effects in the GWAS on feed intake, fat traits, and drip loss, the results from the validation study were not only insignificant but also pointed in the opposite direction of effects (Table 3). Furthermore, 3 out of the 5 markers (ALGA0001781, ASGA0039629, and ASGA0056782) showed significant effects on several traits in the validation study that were not significant in the GWAS. In half of these cases, the direction of the marker effect was the same in both studies (Table 3). DISCUSSION The small sample size of the GWAS was insufficient considering that for the detection of markers with small effects on quantitative traits several thousands of individuals are required (Klein, 2007; Spencer et al., 2009). Thus, it was not surprising that no effects were significant after correction for multiple testing. However, a small sample size with low power also leads to an increased risk of false negative results and thus the rejection of possibly true associations. Therefore, we compared effects that were identified based on uncorrected significant P-values from the GWAS with effects found in the validation study. This resulted in 2 markers, ALGA0001781 and ASGA0056782, with similar effects on growth and fat traits in both studies. If we had based our results on the GWAS alone, we would
1943
have rejected the effects of these 2 markers. However, including the results of the validation study gives reason to argue that the effects were false negatives and should be considered as true effects. The effects of the other 3 markers that were chosen from the GWAS could not be confirmed in the validation study. These results would have been rejected correctly under consideration of multiple testing and can be regarded as false positive results. Since the validation study was performed in a population that is related to the GWAS population, the validation study confirms the SNP effects in the examined German Landrace population. Nevertheless, further studies are necessary to provide evidence that the SNP associated effects occur also in other Landrace populations. Furthermore, marker ASGA0056782 showed dominance effects on live-weight gain and back fat thickness. In pig breeding, line crosses are frequently used to harvest nonadditive genetic effects by crossing breeds that are genetically similar within the breed but highly diverse between breeds. Therefore, this marker might be particularly interesting for breeding purposes. By accumulating one allele in one pig line and the other allele in another line, dominance effects could be predicted across generations and used for breeding improvement in a terminal cross-breeding system. Regarding breeding goals for the Landrace pigs, marker ALGA0001781 raises a problem as it increases weight gain but also fat thickness and area. Fat thickness and area are unfavorable and the breeding goal is to reduce the subcutaneous fat depots but increase live weight. The detected association of marker ALGA0001781 could be caused by pleiotropic effects of one gene on both traits or by the cumulative effect of different genes that are in high linkage disequilibrium with the identified SNP. This locus cannot be used to separate growth and fat deposition if the effects are of pleiotropic nature. If different loci are responsible for the observed effects, recombinant animals could potentially be identified that carry haplotypes combining the favorable alleles of both high growth and low fat deposition. Finally, we looked at genes that surround the 2 validated markers 500 kb up- and downstream. We found genes with previously known functions that may explain some of the effects of the markers. The vacuolar protein sorting-associated protein VTA1 homolog (VTA1) gene is located approximately 200 kb downstream of marker ALGA0001781 on SSC1. VTA1 enhances Vps4 adenosin triphosphatase (ATPase) activity, which regulates the assembly and disassembly of the ESCRT-III complex (Lottridge et al., 2006). Thus, it regulates the intake of nutrients into the cell, which in turn might explain the effects on weight gain and fat area of marker ALGA0001781. Another candidate in close neighborhood of ALGA0001781 is the G protein-coupled receptor 126
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
1944
Strucken et al.
(GPR126) gene, which is involved in cell signaling. Recently, GPR126 has been associated with body height in Australian human families (Liu et al., 2010); therefore, this gene could explain the effects on weight gain in pigs. These 2 potential candidate genes also support the true effect on weight gains for marker ALGA0001781 found in both studies. Conclusion In a GWAS and a subsequent validation study, the genetic markers ALGA0001781 and ASGA0056782 showed consistent effects on growth and fat traits. Therefore, these markers are considered as potential candidates for use in selection. These results show that a less stringent P-value threshold could increase the detection of true associations in GWAS with small population size, which would have otherwise been dismissed as false negative results. However, as the allele effects of 3 out of 5 markers could not be reproduced, this shows the risk of false positive results and how important it is to validate results before their information can be introduced in practical genomic breeding programs of the target population. Since the marker ALGA0001781 is associated with both BW gain and high fat deposition, extended analyses are necessary to decide whether pleiotropic effects of 1 or the combined effect of several causative mutations are linked with the associated SNP. Additional studies are necessary to verify that the identified SNP are associated with the observed trait effects in other Landrace populations.
LITERATURE CITED Aulchenko, Y. S., S. Ripke, A. Isaacs, and C. M. van Duijn. 2007. GenABEL: An R library for genome-wide association analysis. Bioinformatics 23:1294–1296. Goddard, M. E., and B. J. Hayes. 2007. Genomic selection. J. Anim. Breed. Genet. 124:323–330. Hayes, B. J., P. J. Bowman, A. J. Chamberlain, and M. E. Goddard. 2009. Invited review: Genomic selection in dairy cattle: Progress and challenges. J. Dairy Sci. 92:433–443. Klein, R. J. 2007. Power analysis for genome-wide association studies. BMC Genet. 8:58. Kreuzer, S., M. Reissmann, and G. A. Brockmann. 2013. Gene test to elucidate the ETEC F4ab/F4ac receptor status in pigs. Vet. Microbiol. 162:293–295. Liu, J. Z., S. E. Medland, M. J. Wright, A. K. Henders, A. C. Heath, P. A. Madden, A. Duncan, G. W. Montgomery, N. G. Martin, and A. F. McRae. 2010. Genome-wide association study of height and body mass index in Australian twin families. Twin Res. Hum. Genet. 13:179–193. Lottridge, J. M., A. R. Flannery, J. L. Vincelli, and T. H. Stevens. 2006. Vta1p and Vps46p regulate the membrane association and ATPase activity of Vps4p at the yeast multivesicular body. Proc. Natl. Acad. Sci. USA 103:6202–6207. Price, A. L., N. J. Patterson, R. M. Plenge, M. E. Weinblatt, N. A. Shadick, and D. Reich. 2006. Principal components analysis corrects for stratification in genome-wide association studies. Nat. Genet. 38:904–909. Spencer, C. C., Z. Su, P. Donnelly, and J. Marchini. 2009. Designing genome-wide association studies: Sample size, power, imputation, and the choice of genotyping chip. PLoS Genet. 5:E1000477. Strucken, E. M., R. H. Bortfeldt, D. J. de Koning, and G. A. Brockmann. 2012. Genome-wide associations for investigating time-dependent genetic effects for milk production traits in dairy cattle. Anim. Genet. 43:375–382. Tribout, T., C. Larzul, and F. Phocas. 2012. Efficiency of genomic selection in a purebred pig male line. J. Anim. Sci. 90:4164–4176.
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
References
This article cites 11 articles, 3 of which you can access for free at: http://www.journalofanimalscience.org/content/92/5/1939#BIBL
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
The online version of this article, along with updated information and services, is located on the World Wide Web at: http://www.journalofanimalscience.org/content/92/5/1939
www.asas.org
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
Genomewide study and validation of markers associated with production traits in German Landrace boars1 E. M. Strucken,* A. O. Schmitt,† U. Bergfeld,‡ I. Jurke,* M. Reissmann,* and G. A. Brockmann*2 *Humboldt-Universität zu Berlin, Breeding Biology and Molecular Genetics, Invalidenstr. 42, 10115 Berlin, Germany; †Faculty of Science and Technology, Universitätsplatz 5, 39100 Bozen-Bolzano, Italy; and ‡Sächsisches Landesamt für Umwelt, Landwirtschaft und Geologie, Postfach 54 01 37, 01311 Dresden, Germany
Abstract: We present results from a genomewide association study (GWAS) and a single-marker association study. The GWAS was performed with the Illumina PorcineSNP60 BeadChip from which 5 markers were selected for a validation analysis. Genetic effects were estimated for feed intake, weight gain, and traits of fat and muscle tissue in German Landrace boars kept on performance test stations. The GWAS was performed in a population of 288 boars and the validation study for another 432 boars. No statistically significant effect was found in the GWAS after adjusting for multiple testing. Effects of 2 markers, which were significant genomewide before correction for multiple testing (P < 0.00005), could be confirmed in the validation study. The major allele of marker ALGA0056781 on SSC1 was positively associated with both higher weight gain and fat deposition. The effect on live-weight gain was 2.25 g/d in the GWAS (P = 0.0003) and 3.73 g/d in the validation study (P = 0.01) and for back fat thickness was 0.15 mm in the GWAS (P < 0.0001) and
0.20 mm in the validation study (P = 0.02). The marker had similar effects on test-day weight gain (GWAS: 3.85 g/d, P = 0.001; validation study: 6.80 g/d, P = 0.003) and back fat area (GWAS: 0.27 cm2, P < 0.0001; validation study: 0.35 cm2, P = 0.03). Marker ASGA0056782 on SSC13 was associated with live-weight gain. The major allele had negative effects in both studies (GWAS: –4.88 g/d, P < 0.0001; validation study: –3.75 g/d, P = 0.02). The effects of these 2 markers would have been excluded based on the GWAS alone but were shown to be significantly trait associated in the validation study indicating a false-negative result. The G protein-coupled receptor 126 (GPR126) gene approximately 200 kb downstream of marker ALGA0001781 was shown to be associated with human height and therefore might explain the association with weight gain in pigs. Several traits were affected in an economically desired direction by the minor allele of the markers, pointing to the possibility of improvement through further selection.
Key words: dominance, feed intake, genomic selection, GPR126, piglets, weight gain © 2014 American Society of Animal Science. All rights reserved. J. Anim. Sci. 2014.92:1939–1944 doi:10.2527/jas2013-7247 INTRODUCTION The pig industry uses terminal cross-breeding systems to increase production by combining lines of breeds with favorable paternal (growth) and maternal (reproduction, litter size, and mothering ability) characteristics.
1We thank Dr. Uwe Müller from the Humboldt-Universität zu Berlin for his help with the statistical analyses and Dr. Ulf Müller from the Sächsische Landesanstalt für Umwelt, Landwirtschaft und Geologie for providing phenotypic data. The project was supported by the Saxon Ministry for Environment and Agriculture. 2Corresponding author: [email protected] Received October 11, 2013. Accepted March 4, 2014.
While cross-breeding can exploit complementarity, permanent and cumulative improvement can only be achieved by selection. Until recently, selective breeding programs have mainly used conventional breeding values estimated from performance records of candidates for selection and their relatives. With the advent of genomewide markers, higher accuracies and shorter generation intervals may be achieved using genomic selection (e.g., Goddard and Hayes, 2007; Hayes et al., 2009). Obstacles impeding use of genomewide data in pig breeding are high costs and small calibration populations. Large calibration populations are needed to accurately estimate marker effects and verify genomic breeding values with highly accurate estimated breeding values (Tribout et al., 2012). However, large
1939 Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
1940
Strucken et al.
Table 1. Population description for corrected trait measurements in the genomewide association study (GWAS) and trait deviations in the validation study Trait Feed intake,1 kg Live weight gain, g/d Test-day weight gain, g/d Back muscle area, cm2 Back muscle thickness, mm Back fat area,1 cm2 Back fat thickness,1 mm Intramuscular fat content, % Drip loss,1 % pH 45 min postmortem
n 265 271 272 – 150 241 271 160 213 130
Mean ± SD 201 ± 43 689 ± 40 959 ± 70 – 53 ± 2 18 ± 4 13 ± 2 1.0 ± 0.2 4.0 ± 1.4 6.4 ± 0.15
GWAS Minimum 69 558 678 – 43 5 6 0.5 0.6 6.0
Maximum 330 841 1,247 – 59 36 23 1.8 9.2 6.9
n 433 433 433 433 – 433 433 433 433 433
Validation Study Mean ± SD Minimum 1.71 ± 6.2 –28 7.76 ± 17.6 –54 15.67 ± 28.2 –72 0.01 ± 2.38 –9.1 – – 0.75 ± 1.8 –7.5 0.17 ± 0.9 –5.8 0.003 ± 0.11 –0.34 –0.10 ± 0.45 –2.59 0.01 ± 0.06 –0.25
Maximum 30 90 116 14.3 – 7.5 5.0 0.84 1.68 0.20
1Trait deviations in the validation study are positive if feed intake, back fat area and thickness, and drip loss are lower than average because this is the favorable phenotype. The mean phenotypic value of the boars in the validation study deviated from the mean value since the whole German Landrace population was used for the correction of trait deviations for environmental effects.
calibration populations are not available for most pig breeds. Therefore, reported marker effects from genomewide association studies (GWAS) should be rigorously verified in different populations. Two problems arise from GWAS with small population sizes, the first one being an increased chance of false positive results in GWAS if multiple testing is not adequately considered by correction of the significance levels for marker effects or by permutation testing. Secondly, reducing the significance threshold has an inherent risk of introducing false negative results, especially if the population is small (i.e., low statistical power). As most methods to account for multiple testing are rather stringent, implementation may result in exclusion of true effects (Strucken et al., 2012). Our objective was to estimate GWAS marker effects in a small German Landrace population and subsequently verify the effects in another different population to ascertain the frequencies of type I and type II errors. MATERIAL AND METHODS Phenotypic Data Animals were anesthetized and slaughtered according to German animal protection laws (TierschutzSchlachtverordnung, BGBl. I S. 2982). Phenotypic data were collected within the national performance tests for German Landrace boars from 2009 to 2011 and provided by the breeding organization after correction for environmental effects (Zentralverband der Deutschen Schweineproduktion e.V., Bonn, Germany; www.zdsbonn.de/; Table 1.) All animals were kept under similar conditions on 2 national test stations and slaughtered at 105 kg BW. For the GWAS, 288 animals were selected for low pedigree relationships based on their kin-
ship coefficient. Sire and dam information was available for 268 animals showing that the population descended from 30 sires and 205 dams resulting in 105 full-sibs and 155 paternal and 2 maternal half-sibs. For the validation study, another 432 boars were available that descended from 56 sires and 233 dams resulting in 206 full-sibs and 132 paternal and 13 maternal half-sibs. Sires and dams overlapped by 45 and 43%, respectively, between the populations used in the GWAS and validation study. For all animals in the GWAS and in the validation study, the same phenotypes were measured, except for the back muscle, where the thickness was measured for the GWAS and the area for the validation study. For the GWAS, we used own performance data corrected for environmental effects; correction for environmental effects was done on the basis of animals that were in the test station simultaneously with the animals chosen for the GWAS (584 animals). For the validation study, we used trait deviations (TD) after correction for environmental effects, considering the kinship of the animals and the direction of effect for breeding (favorable or unfavorable). Trait deviations give the deviation of the phenotype of a particular boar in comparison to the population mean. All available animals with phenotypic data of the German Landrace population were used to correct for environmental effects in the validation study. Traits of the GWAS and the validation study were corrected for covariance between traits (live weight for back fat thickness and back muscle thickness, warm carcass weight for intramuscular fat content, back fat area, back muscle area, and drip loss, and live weight at test day for feed intake and test day gain) and fixed effect factors such as farm, year, quarter, sex, slaughter house, test number, test station, test group, ultrasound device, and measuring person. Furthermore, to correct for external factors, TD were adjusted to the direction of effects for breeding,
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
1941
Marker association in Landrace boars
Table 2. Information about most significant markers selected from the genomewide association study (GWAS) for the validation study (genome build Sscrofa 10.2) Traits in GWAS1 Major allele effect FE 0.27
Uncorrected2 P-value 2.4 × 10–5 (0.46)
Marker ALGA0001781 (rs81350882)
SSC 1
Position, bp 25,798,122
SNP A/G
Major allele A
ASGA0039629 (rs81403471) ASGA0056782 (rs81443902) H3GA0049322 (rs80901871) ASGA0077474 (rs81466984)
8
120,311,332
A/G
A
IMF
0.02
2.3 × 10–5 (0.51)
13
27,499,096
A/G
A
LWG
–4.88
2.3 × 10–6 (0.60)
17
56,149,830
A/C
C
DL
0.20
5.2 × 10–5 (0.51)
17
57,039,720
A/G
G
FI
–2.13
1.9 × 10–5 (0.45)
1FE
= fat area (cm2); IMF = intramuscular fat content (%); LWG = life weight gain (g/d); DL = drip loss (%); FI = feed intake (kg). are uncorrected for multiple testing; P-values after correction for multiple testing are given in parentheses.
2P-values
such that a positive value represents a desirable breeding direction and a negative value represents an undesirable breeding direction. As such for the traits feed intake, fat area, back fat thickness, and drip loss, a higher performance than the population mean was assigned a negative value to acknowledge an undesirable breeding direction and should be kept in mind when interpreting the results between the 2 studies. Feed intake and test-day weight gain were recorded for the period in which animal live weight grew from 30 to 105 kg. Live-weight gain represents the average daily weight gain from birth to slaughter. Back muscle and back fat thickness were measured on the live animal via ultrasound above the 13th thoracic vertebra. Back muscle and back fat area (cm2) as well as intramuscular fat content were measured at the loin between the 13th and 14th thoracic vertebra of the longissimus dorsi after slaughter. Intramuscular fat content was measured by chemical analysis. Further meat traits included the drip loss of the loin collected over 24 h at 4°C after slaughter and the pH value of the loin 45 min postmortem. Finally, all traits were tested for normal distribution using the Kolmogorov-Smirnov test. Intramuscular fat content was not normally distributed in the GWAS population and was therefore log-transformed. Genetic Data Genotyping of Animals in the Genomewide Association Study. Animals were genotyped with the 60K PorcineSNP60 BeadChip (Illumina, San Diego, CA). Quality control included a minor allele frequency of >0.01 and a call rate per marker and animal of >10%. In total 283 animals and 53,536 markers passed quality control. In addition, 16 animals were excluded because their corresponding phenotypic data was unavailable. No markers in the GWAS reached a significant P-value (α = 0.05) after permutation correction for multiple testing. Nevertheless, 5 markers were selected based
on lowest uncorrected P-values to evaluate whether the previous effects are valid (Table 2). Sus scrofa chromosomes 1, 8, and 13 each contained 1 marker, while 2 markers were located on SSC17 (Table 2). Genotyping of Animals in the Validation Study. Deoxyribonucleic acid was isolated from ear tissue collected from each boar. The Kompetitive allele-specific polymerase chain reaction (KASPar) SNP genotyping system (KBioscience Ltd., Maple Park, Hoddesdon, UK) was used to identify the genotypes of each animal for every marker. Assays were performed according to the protocol published by Kreuzer et al. (2013) and performed with allele-specific primers in a polymerase chain reaction (Supplementary Table 1). Statistical Analyses The GWAS was performed with the GenABEL package in R (Aulchenko et al., 2007). A fast score test, as implemented in the package, was used for the association analysis using principal components based on genomic data to account for population stratification (Price et al., 2006). Additionally, the lambda factor was used to correct P-values if population stratification was still indicated after correction. Multiple testing thresholds were determined based on 10,000 permutations. The validation study with the 5 selected markers was performed using the SAS package (version 9.3; SAS Inst. Inc., Cary, NC). A mixed model was fitted for the association between the SNP and the TD: yijkl = μ + SNPi + sirej(damk) + eijkl, in which yijkl is the TD of each animal, μ is the mean of the TD of the 432 boars, SNPi is the genotype of the ith marker (0, 1 or 2 for the additive genetic model and 0 or 1 for dominance model), damk nested within sirej was used to account for full- and half-sib relationships, and eijkl is the random residual error. Genotype was fitted as a numeric
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
1942
–0.06 (0.91)
–1.91 (0.16) 0.23 (0.90) –0.11 (0.86) 1.29 (0.33)
1.94 (0.15) –2.67 (0.25) –7.18 (0.0007) –3.07 (0.23) –4.89 (0.05) –0.34 (0.90) –0.24 (0.83) 1.34 (0.50)
3.73 (0.01) 0.99 (0.16) –2.84 (0.06) –4.88 (2.3 × 10–6) –3.75 (0.02)
ASGA0077474
H3GA0049322
ASGA0056782
ASGA0039629
1FI = feed intake (kg); LWG = live weight gain (g/d); TWG = test-day weight gain (g/d); FE = fat area (cm2); FT = fat thickness (mm); IMF = intramuscular fat content (%); DL = drip loss (%), PH = pH value 45 min postmortem. Back muscle area and thickness were not significantly associated in either studies and, therefore, are not shown. 2Effects of the validation study have been multiplied by –1 to make comparisons easier because a positive trait deviation value for feed intake, fat area and thickness, and drip loss indicates values that are lower than the average (see also Material and Methods). Bold typescript identifies effects that led to the selection of markers in the GWAS and validated significant effects in the validation study.
PH 0.03 (0.10) –0.01 (0.001) –0.01 (0.48) 0.005 (0.25) –0.04 (0.21) 0.005 (0.28) –0.03 (0.40) 0.001 (0.85) –0.002 (0.90) –0.002 (0.52) DL2 –0.01 (0.63) 0.08 (0.05) –0.003 (0.91) 0.005 (0.91) 0.0007 (0.99) –0.03 (0.50) 0.20 (5.2 × 10–5) 0.05 (0.31) 0.04 (0.05) –0.02 (0.50) IMF 0.01 (0.001) –0.01 (0.35) 0.02 (2.3 × 10–5) 0.01 (0.15) 0.002 (0.76) 0.005 (0.67) –0.008 (0.34) –0.004 (0.73) –0.002 (0.56) –0.001 (0.89) FT2 0.15 (3.5 ×10–5) 0.20 (0.02) 0.0005 (0.99) –0.25 (0.007) –0.12 (0.07) –0.28 (0.006) –0.11 (0.15) –0.12 (0.27) –0.10 (0.004) 0.12 (0.15) FE2 0.27 (2.4 × 10–5) 0.35 (0.03) –0.07 (0.36) –0.23 (0.17) –0.09 (0.44) –0.33 (0.07) –0.32 (0.02) –0.23 (0.25) –0.21 (0.002) 0.08 (0.59) TWG 3.85 (0.001) 6.80 (0.003) LWG 2.25 (0.0003)
FI2 1.15 (0.03) 0.80 (0.14) 0.32 (0.60) –0.71 (0.20) –0.72 (0.45) –1.24 (0.04) –0.14 (0.90) –0.62 (0.35) –2.13 (1.9 × 10–5)
Both studies showed comparable genotype and allele frequencies with both having the same major allele. The overall minor allele frequency in the GWAS was 0.27 ± 0.14 after quality control. The 5 markers that were selected for the validation study had a minor allele frequency (B) of >0.20 in the GWAS and >0.18 in the validation study (Supplementary Table 2). All selected markers were in Hardy-Weinberg equilibrium in both studies. While uncorrected P-values were as low as 0.00002 (Table 2), none of the marker effects of the GWAS were significant after correction for multiple testing. Therefore, all further comparisons were made for SNP that had the lowest uncorrected P-value for the examined traits in the GWAS. In the GWAS, the most significant effects of the chosen markers were for the traits fat deposition, weight gain, drip loss, and feed intake (Table 2). These genomewide allele effects were reproduced in the validation study for 2 markers: marker ALGA0001781 on SSC1 had validated effects on live-weight gain and testday weight gain as well as on back fat area and thickness, and marker ASGA0056782 on SSC13 showed effects on live-weight gain (Table 3). Allele effects were larger in the validation study; for example, the major allele of marker ALGA0001781 increased live-weight and test-day weight gain by 3.73 and 6.80 g/d, respectively, compared to 2.25 and 3.85 g/d in the GWAS, whereas the major allele of marker ASGA0056782 had a less negative impact on liveweight gain in the validation study with only –3.75 g/d compared to –4.88 g/d in the GWAS (Table 3). Estimates for genotype effects and the least square mean differences between the genotype classes showed a significantly higher weight gain the more copies of the major allele of marker ALGA0001781 were present in the genotype (Fig. 1). Marker ASGA0056782 indicated a heterosis effect for live-weight gain, but differences were only significant between the AA and AB genotypes (Fig. 1). The dominance effect of live weight gain was 5.59 ± 1.99 units (P = 0.006). Back fat thickness decreased the more copies of the major allele of marker ASGA0056782 were present but significantly increased for the major allele of marker ALGA0001781 (Fig. 1). Additionally, marker ASGA0056782 showed a positive
Study GWAS Validation GWAS Validation GWAS Validation GWAS Validation GWAS Validation
RESULTS
Marker ALGA0001781
variable in the additive allele model and as a categorical variable when differences between genotype classes were estimated. The Bonferroni correction was applied to account for multiple testing when differences between leastsquare means of genotype classes were tested. Neither the GWAS nor the validation study included further environmental effects because the traits used were already corrected by the breeder organization for masking factors as described in the phenotypic data.
Table 3. Additive effects of major alleles of tested markers in the genomewide association study (GWAS) and the validation study. Uncorrected P-values for GWAS and corrected P-values for the validation study are given in parentheses1
Strucken et al.
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
Marker association in Landrace boars
Figure 1. Effect plots for selected markers and production traits from the validation study (A is the major allele). *P < 0.01; +P < 0.05. FT = fat thickness; LWG = life weight gain; TWG = test-day weight gain. †For a better understanding, effects for fat thickness have been multiplied by –1 because a positive trait deviation indicates values that are lower than the average.
dominance effect of 0.12 ± 0.12 units (P = 0.02) for back fat thickness. Even though marker ASGA0077474 on SSC17 showed significant effects in the GWAS on feed intake, fat traits, and drip loss, the results from the validation study were not only insignificant but also pointed in the opposite direction of effects (Table 3). Furthermore, 3 out of the 5 markers (ALGA0001781, ASGA0039629, and ASGA0056782) showed significant effects on several traits in the validation study that were not significant in the GWAS. In half of these cases, the direction of the marker effect was the same in both studies (Table 3). DISCUSSION The small sample size of the GWAS was insufficient considering that for the detection of markers with small effects on quantitative traits several thousands of individuals are required (Klein, 2007; Spencer et al., 2009). Thus, it was not surprising that no effects were significant after correction for multiple testing. However, a small sample size with low power also leads to an increased risk of false negative results and thus the rejection of possibly true associations. Therefore, we compared effects that were identified based on uncorrected significant P-values from the GWAS with effects found in the validation study. This resulted in 2 markers, ALGA0001781 and ASGA0056782, with similar effects on growth and fat traits in both studies. If we had based our results on the GWAS alone, we would
1943
have rejected the effects of these 2 markers. However, including the results of the validation study gives reason to argue that the effects were false negatives and should be considered as true effects. The effects of the other 3 markers that were chosen from the GWAS could not be confirmed in the validation study. These results would have been rejected correctly under consideration of multiple testing and can be regarded as false positive results. Since the validation study was performed in a population that is related to the GWAS population, the validation study confirms the SNP effects in the examined German Landrace population. Nevertheless, further studies are necessary to provide evidence that the SNP associated effects occur also in other Landrace populations. Furthermore, marker ASGA0056782 showed dominance effects on live-weight gain and back fat thickness. In pig breeding, line crosses are frequently used to harvest nonadditive genetic effects by crossing breeds that are genetically similar within the breed but highly diverse between breeds. Therefore, this marker might be particularly interesting for breeding purposes. By accumulating one allele in one pig line and the other allele in another line, dominance effects could be predicted across generations and used for breeding improvement in a terminal cross-breeding system. Regarding breeding goals for the Landrace pigs, marker ALGA0001781 raises a problem as it increases weight gain but also fat thickness and area. Fat thickness and area are unfavorable and the breeding goal is to reduce the subcutaneous fat depots but increase live weight. The detected association of marker ALGA0001781 could be caused by pleiotropic effects of one gene on both traits or by the cumulative effect of different genes that are in high linkage disequilibrium with the identified SNP. This locus cannot be used to separate growth and fat deposition if the effects are of pleiotropic nature. If different loci are responsible for the observed effects, recombinant animals could potentially be identified that carry haplotypes combining the favorable alleles of both high growth and low fat deposition. Finally, we looked at genes that surround the 2 validated markers 500 kb up- and downstream. We found genes with previously known functions that may explain some of the effects of the markers. The vacuolar protein sorting-associated protein VTA1 homolog (VTA1) gene is located approximately 200 kb downstream of marker ALGA0001781 on SSC1. VTA1 enhances Vps4 adenosin triphosphatase (ATPase) activity, which regulates the assembly and disassembly of the ESCRT-III complex (Lottridge et al., 2006). Thus, it regulates the intake of nutrients into the cell, which in turn might explain the effects on weight gain and fat area of marker ALGA0001781. Another candidate in close neighborhood of ALGA0001781 is the G protein-coupled receptor 126
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
1944
Strucken et al.
(GPR126) gene, which is involved in cell signaling. Recently, GPR126 has been associated with body height in Australian human families (Liu et al., 2010); therefore, this gene could explain the effects on weight gain in pigs. These 2 potential candidate genes also support the true effect on weight gains for marker ALGA0001781 found in both studies. Conclusion In a GWAS and a subsequent validation study, the genetic markers ALGA0001781 and ASGA0056782 showed consistent effects on growth and fat traits. Therefore, these markers are considered as potential candidates for use in selection. These results show that a less stringent P-value threshold could increase the detection of true associations in GWAS with small population size, which would have otherwise been dismissed as false negative results. However, as the allele effects of 3 out of 5 markers could not be reproduced, this shows the risk of false positive results and how important it is to validate results before their information can be introduced in practical genomic breeding programs of the target population. Since the marker ALGA0001781 is associated with both BW gain and high fat deposition, extended analyses are necessary to decide whether pleiotropic effects of 1 or the combined effect of several causative mutations are linked with the associated SNP. Additional studies are necessary to verify that the identified SNP are associated with the observed trait effects in other Landrace populations.
LITERATURE CITED Aulchenko, Y. S., S. Ripke, A. Isaacs, and C. M. van Duijn. 2007. GenABEL: An R library for genome-wide association analysis. Bioinformatics 23:1294–1296. Goddard, M. E., and B. J. Hayes. 2007. Genomic selection. J. Anim. Breed. Genet. 124:323–330. Hayes, B. J., P. J. Bowman, A. J. Chamberlain, and M. E. Goddard. 2009. Invited review: Genomic selection in dairy cattle: Progress and challenges. J. Dairy Sci. 92:433–443. Klein, R. J. 2007. Power analysis for genome-wide association studies. BMC Genet. 8:58. Kreuzer, S., M. Reissmann, and G. A. Brockmann. 2013. Gene test to elucidate the ETEC F4ab/F4ac receptor status in pigs. Vet. Microbiol. 162:293–295. Liu, J. Z., S. E. Medland, M. J. Wright, A. K. Henders, A. C. Heath, P. A. Madden, A. Duncan, G. W. Montgomery, N. G. Martin, and A. F. McRae. 2010. Genome-wide association study of height and body mass index in Australian twin families. Twin Res. Hum. Genet. 13:179–193. Lottridge, J. M., A. R. Flannery, J. L. Vincelli, and T. H. Stevens. 2006. Vta1p and Vps46p regulate the membrane association and ATPase activity of Vps4p at the yeast multivesicular body. Proc. Natl. Acad. Sci. USA 103:6202–6207. Price, A. L., N. J. Patterson, R. M. Plenge, M. E. Weinblatt, N. A. Shadick, and D. Reich. 2006. Principal components analysis corrects for stratification in genome-wide association studies. Nat. Genet. 38:904–909. Spencer, C. C., Z. Su, P. Donnelly, and J. Marchini. 2009. Designing genome-wide association studies: Sample size, power, imputation, and the choice of genotyping chip. PLoS Genet. 5:E1000477. Strucken, E. M., R. H. Bortfeldt, D. J. de Koning, and G. A. Brockmann. 2012. Genome-wide associations for investigating time-dependent genetic effects for milk production traits in dairy cattle. Anim. Genet. 43:375–382. Tribout, T., C. Larzul, and F. Phocas. 2012. Efficiency of genomic selection in a purebred pig male line. J. Anim. Sci. 90:4164–4176.
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
References
This article cites 11 articles, 3 of which you can access for free at: http://www.journalofanimalscience.org/content/92/5/1939#BIBL
Downloaded from www.journalofanimalscience.org at The University of British Columbia Library on November 24, 2014
