- Open Access
Accuracy of genotype imputation in Nelore cattle
© Carvalheiro et al.; licensee BioMed Central Ltd. 2014
- Received: 1 April 2014
- Accepted: 19 September 2014
- Published: 10 October 2014
Genotype imputation from low-density (LD) to high-density single nucleotide polymorphism (SNP) chips is an important step before applying genomic selection, since denser chips tend to provide more reliable genomic predictions. Imputation methods rely partially on linkage disequilibrium between markers to infer unobserved genotypes. Bos indicus cattle (e.g. Nelore breed) are characterized, in general, by lower levels of linkage disequilibrium between genetic markers at short distances, compared to taurine breeds. Thus, it is important to evaluate the accuracy of imputation to better define which imputation method and chip are most appropriate for genomic applications in indicine breeds.
Accuracy of genotype imputation in Nelore cattle was evaluated using different LD chips, imputation software and sets of animals. Twelve commercial and customized LD chips with densities ranging from 7 K to 75 K were tested. Customized LD chips were virtually designed taking into account minor allele frequency, linkage disequilibrium and distance between markers. Software programs FImpute and BEAGLE were applied to impute genotypes. From 995 bulls and 1247 cows that were genotyped with the Illumina® BovineHD chip (HD), 793 sires composed the reference set, and the remaining 202 younger sires and all the cows composed two separate validation sets for which genotypes were masked except for the SNPs of the LD chip that were to be tested.
Imputation accuracy increased with the SNP density of the LD chip. However, the gain in accuracy with LD chips with more than 15 K SNPs was relatively small because accuracy was already high at this density. Commercial and customized LD chips with equivalent densities presented similar results. FImpute outperformed BEAGLE for all LD chips and validation sets. Regardless of the imputation software used, accuracy tended to increase as the relatedness between imputed and reference animals increased, especially for the 7 K chip.
If the Illumina® BovineHD is considered as the target chip for genomic applications in the Nelore breed, cost-effectiveness can be improved by genotyping part of the animals with a chip containing around 15 K useful SNPs and imputing their high-density missing genotypes with FImpute.
- Imputation Method
- Genomic Prediction
- Pedigree Information
- Imputation Accuracy
- Genomic Estimate Breeding Value
Genomic information from dense single nucleotide polymorphism (SNP) chips provides the opportunity to increase the rate of genetic progress in breeding programs, if a sufficient number of markers and animals with phenotypes (or pseudo-phenotypes such as estimated breeding values, EBV) are genotyped . Because the cost of genotyping is high, alternative methods are necessary for cost-efficient genomic applications. A strategy that is used in dairy breeding programs is to genotype influential animals using a denser chip (e.g. Illumina® BovineSNP50 v2 - 50 K; Illumina Inc., San Diego, CA) and selection candidates and cows using a lower-density chip (e.g. Illumina® BovineLD - 7 K) and then to impute (i.e. predict) missing genotypes from lower to higher density before calculating genomic estimated breeding values (GEBV) . This cost-effective strategy provides reliabilities of GEBV that are similar to those obtained if selection candidates were genotyped with the higher-density chip ,.
The Nelore (indicine) breed is the most important beef cattle breed in Brazil . For this breed, the Illumina® BovineHD chip (HD) is used as the ¿gold standard¿ for research purposes, since a low level of linkage disequilibrium between adjacent markers is observed in lower-density chips (e.g. 50 K) ,. Profit margins from beef cattle operations are too low for the use of a HD chip at the commercial level. Thus, lower-density chips are required to overcome this limitation, which highlights the importance of assessing the accuracy of imputing genotypes in the Nelore breed.
The objective of this study was to assess the accuracy of genotype imputation in Nelore cattle, using different imputation methods, different commercial and customized SNP chips and sets of animals whose genotypes were to be imputed. The importance of relatedness between validation and reference animals was also evaluated for the different chips and methods.
Pedigree information of genotyped animals
Individuals in pedigree
Individuals with progeny
Individuals with no progeny
Individuals with only known sire
Individuals with only known dam
Individuals with known sire and dam
Founders with no progeny
Genomic relationship statistics between reference and validation sets
Sire/sire (793; 202)
Sire/dam (793; 1247)
All sires and dams were genotyped with the HD chip, which contains approximately 777 K SNPs. Animals from the validation set had their HD genotypes masked, except for the genotypes of markers present on the LD chip under evaluation, thus mimicking a situation in which these animals were genotyped with LD chips.
Imputation from commercial LD panels to the HD chip was simulated by assuming that genotypes of animals from the validation set were available only for markers that were present on the HD and the following commercial chips: Illumina® BovineLD (7 K), Illumina® BovineSNP50 v2 (50 K) and GeneSeek® Genomic Profiler 20 K and 75 K for indicine breeds (GGP20Ki and GGP75Ki, respectively).
Eight customized (mimicked) 15 K LD chips were also tested, with varying densities and SNP selection criteria. Testing customized LD chips with less than 15 K SNPs was not relevant because their cost-effectiveness would not be attractive (Illumina®, personal communication). Using markers of the HD chip that passed quality control (see below), four 15 K chips were simulated based on selection of one marker from each window of 29 subsequent markers, according to the UMD v3.1 assembly. For the first 15 K chip (15 K_e), SNPs were evenly spaced by selecting the last marker from each window. For the second 15 K chip (15 K_em), the SNP with the highest minor allele frequency (MAF) was selected from each window, and for the third chip (15 K_el), the SNP with the highest average linkage disequilibrium with other SNPs from the same window was selected from each window. The r2 was adopted as the measure of linkage disequilibrium. In the fourth 15 K chip (15K_eml), the SNP with the highest value for the product between its MAF and its average r2 with other SNPs from the same window was selected from each window.
The remaining four customized LD chips were developed based on the add-on concept offered by Illumina®, where additional SNPs can be added to an existing commercial chip in a cost-effective way. The Illumina® Bovine LD chip (7 K) was used as the base chip. Additional SNPs were selected using the same criteria as used for the 15K_eml chip. Windows containing 39, 25, 16 and 9 subsequent markers were used to compose chips with densities around 18 K (11a7 K), 24 K (17a7 K), 34 K (27a7 K) and 55 K (48a7 K) markers.
Quality control of the genotypes
Quality control was performed for HD genotypes of the reference set, using the following criteria for excluding SNPs: (1) SNPs that were located in non-autosomal regions; (2) SNPs that had the same genomic coordinates, i.e. mapped to the same positions (just the replicates were removed); (3) SNPs with a p-value in the Hardy-Weinberg equilibrium z-test  less than or equal to 10−5; (4) SNPs with a MAF less than 0.02; and (5) SNPs with a call rate per SNP less than 0.98 (genotypes with a GenCall score less than 0.70 were considered missing when computing this statistic). After these edits, 439 595 SNPs remained. All samples from the reference set had a call rate per individual greater than 0.9 for SNPs passing quality control and were kept for the analyses.
Number (Nb) of SNPs shared with the HD chip, for different SNP chips
Nb common SNPs with HD
Nb common SNPs after QC2
Illumina® Bovine SNP50 v2
GeneSeek® Genomic Profiler 20 K - Indicine
GeneSeek® Genomic Profiler 75 K - Indicine
Customized 11K_eml add-on 7 K
Customized 17 K_eml add-on 7 K
Customized 27 K_eml add-on 7 K
Customized 48K_eml add-on 7 K
Imputation of genotypes from the LD chips to the HD chip was performed using the software packages BEAGLE v.3.3  and FImpute v.2.2 . BEAGLE is a commonly used population-based imputation program (i.e. it does not rely on pedigree information) that adopts a stochastic procedure based on a Hidden Markov Monte-Carlo process to infer the probabilities of each haplotype/genotype. We used the most likely genotype as the predicted genotype. FImpute uses a family and population-based algorithm, or only the population-based algorithm, if pedigree information is not available, to deterministically phase the haplotypes and impute the missing genotypes. To evaluate the performance when considering family information in FImpute, we used both approaches, i.e. with or without pedigree information. Both programs were run with default parameters ,.
Analyses 1 to 12 were used to compare SNP chips. Results from analyses 13 to 16 were compared to those of their counterpart analyses (1, 2, 8 and 9, respectively) to evaluate the gain in accuracy when considering family information with FImpute. Results from analyses 17 to 23 were compared to those from analyses 1 to 4, 8, 9 and 12, to evaluate imputation accuracy when using different sets of animals to be imputed. Finally, analyses 24 to 29 and 30 to 35 were performed to infer accuracy of the imputed genotypes of Nelore young sires and dams using BEAGLE and different LD chips.
Two criteria were used to assess imputation accuracy. The first was the percentage of correctly imputed genotypes (PERC). For incorrectly imputed genotypes, either one or both alleles can be imputed incorrectly. To distinguish between these two cases, the Pearson¿s correlation (CORR) between imputed and observed genotypes (coded as 0, 1 or 2 copies of the B allele) was also computed, as in Hickey et al. . Both PERC and CORR were calculated by individual and by SNP, for imputed SNPs only. Since both BEAGLE and FImpute imputed all missing genotypes, statistics on the proportion of imputed SNPs were not needed.
The impact of genetic relatedness between validation and reference animals on imputation accuracy was assessed by regressing CORR on the average of the top 10 genomic relationships between each animal in the validation set with all the animals in the reference set .
Comparison of lower-density SNP chips
Average (standard deviation) imputation accuracy, for different imputation analyses using FImpute
Nb (%) SNPs to be imputed
Imputation accuracy was greater than 0.97 for all chips except for the 7 K chip. Considering the proportion of SNPs to be imputed (99.1%), the imputation accuracy was high even for the 7 K chip. As documented in the literature ,, imputation accuracy increases with a decreasing proportion of SNPs to be imputed. However, the gain in accuracy from the 11a7 K to the 48a7 K chip, for example, was small (0.0108) because the accuracy obtained with the 11a7 K chip was already high.
Among the commercial chips, imputation accuracies with the 50 K and GGP20Ki chips were similar and outperformed that of the 7 K chip, while the GGP75Ki chip had the best accuracy. Among the virtual 15 K chips, selecting SNPs based on MAF (15K_em) rather than on linkage disequilibrium (15K_el), in addition of being evenly spaced, tended to result in slightly higher imputation accuracies. The highest accuracy was observed when both criteria and even spacing were combined to define the SNP content of a virtual chip (15K_eml). However, the increase in accuracy was nominal compared to the 15K_e evenly spaced chip. Imputation accuracy of the 11a7 K chip was comparable to that of 15K_eml, with the potential benefit of the former being cheaper to be manufactured/acquired by adding SNPs on the existing 7 K chip.
Although commercial and customized chips resulted in similar imputation accuracies, the customized chips that had the highest accuracies outperformed commercial chips with a similar density (after quality control). For instance, the average accuracy of the 15K_eml chip was 0.6% and 0.7% higher than those of the 50 K and GGP20Ki chips, respectively.
Importance of pedigree information
Average (standard deviation) imputation accuracy, using FImpute with or without pedigree (Ped) information
1 and 13
2 and 14
8 and 15
9 and 16
Comparison of validation sets
Average (standard deviation) imputation accuracy, using dams or young sires as validation set
17 and 1
18 and 2
19 and 3
20 and 4
21 and 8
22 and 9
23 and 12
Comparison of imputation methods
Summary statistics of imputation accuracy, using BEAGLE and FImpute
Importance of genomic relatedness for imputation accuracy
SNP-wise imputation accuracy
Imputation methods rely partially on linkage disequilibrium between markers to infer unobserved genotypes. The Nelore breed has lower levels of linkage disequilibrium at short distances than taurine breeds ,. Nevertheless, the imputation accuracies obtained in this study are comparable to accuracies reported in the literature for taurine breeds -. For instance, Ma et al.  imputed genotypes from the 50 K to the HD chip in a population of Swedish and Finnish Red cattle and found similar accuracies (around 0.97) to those reported here. As documented in the literature e.g. ,,, imputation accuracy increased with increasing density of the LD chip. Increasing the density of the LD chip from 7 K to 15 K resulted in a greater increase in imputation accuracy than an increase in density from 15 K to 75 K, because imputation accuracy was already high (>0.97) for the 15 K chip. This result is consistent, for example, with those of Khatkar et al.  who reported on the imputation of 50 K genotypes of Australian Holstein-Friesian cattle and observed a relatively greater increase in imputation accuracy when the density of the LD chip increased from 3 K to 7 K than from 7 K to 10 K.
Imputation accuracy has a large influence on the reliability of genomic predictions ,,. Mulder et al.  derived a deterministic equation to predict the accuracy of GEBV based on imputation accuracy (measured as a correlation) and observed that it increased linearly with increasing imputation correlation. Daetwyler et al.  suggested that the decline in accuracy of GEBV was actually slightly lower than the decline in accuracy of imputation. In a scenario with low-density genotypes (14 SNPs/Morgan), these authors observed that 87.8% of missing genotypes were correctly imputed but 95% of the accuracy of GEBV obtained with high-density SNP genotypes (1500 SNPs/Morgan) was achieved. Although reliability of GEBV was not evaluated in the present study, it is plausible to assume that the reliability of GEBV of Nelore cattle based on imputed genotypes from a chip with approximately 15 K SNPs, for which the imputation accuracy was around 0.98, would be similar to that of GEBV obtained with the HD chip. Nevertheless, a more thorough analysis on this subject is needed.
Different (mimicked) customized LD chips were tested in order to evaluate changes in imputation accuracy when criteria for SNP selection were modified. As in Mulder et al. , selecting SNPs based on MAF (15K_em) and even spacing across the genome had little impact on imputation accuracy compared to selecting SNPs based only on even spacing. A small favorable difference was observed when SNPs were selected based on MAF (15K_em) rather than on linkage disequilibrium (15K_el), in addition to being evenly spaced. A slightly better accuracy was observed when both criteria (MAF and linkage disequilibrium), in addition to even spacing, were combined to select the SNPs for the chip (15K_eml). Increasing SNP density in the telomere regions of the chromosomes is expected to further increase the imputation performance of the customized chips . We did not use refined algorithms to optimize the imputation accuracy of the customized chips since the development of an LD chip was outside the scope of our study. Results obtained with the 11a7 K chip suggests that if a new optimized LD chip was to be developed, adding SNPs to the existing commercial 7 K chip would be a good strategy since this would be less costly and provide an imputation accuracy that is comparable to that of a completely customized chip with similar density.
The customized chips that showed the highest imputation accuracy slightly outperformed the commercial chips with an equivalent density. It is important to mention that commercial and customized chips cannot be properly compared, since the design of the customized chips used information on genotypes from the same population than that to be imputed. However, it does highlight the importance of using population-specific information to design LD chips.
Imputation accuracy was not as much affected by pedigree information as by using different imputation methods. FImpute resulted in higher imputation accuracies for the different chips and validation sets than BEAGLE. Ma et al.  found that FImpute slightly outperformed BEAGLE when imputing Swedish and Finnish Red cattle genotypes from 50 K to HD, but BEAGLE outperformed FImpute when imputing from 3 K to 50 K. Sun et al.  also observed a slightly better imputation performance of BEAGLE compared to FImpute when imputing Angus genotypes from 7 K to 50 K. These results indicate that the choice of the imputation method depends on the chip and population, i.e. there is no single method that provides higher imputation accuracy for all scenarios. However, an outstanding advantage of FImpute over BEAGLE is its computational efficiency. As reported by Ma et al. , processing time of the analysis with FImpute was much shorter than with BEAGLE (data not shown).
Another factor that influenced imputation accuracy was the level of relatedness between imputed and reference animals. In agreement with the literature ,,,, imputation accuracy tended to increase as the relatedness between imputed and reference animals increased for both imputation methods. The influence of relatedness on imputation accuracy decreased with increasing SNP density of the LD chip. For the Nelore cattle population, using the denser LD chips (GGP75Ki and 48a7 K) resulted in high accuracies (>0.90) with FImpute even for animals that were poorly related to the reference set.
For some genomic applications (e.g. genome-wide association (GWA) studies), SNP-wise imputation accuracy is relevant to prevent the propagation of genotyping errors. As in Erbe et al.  and VanRaden et al. , some regions of the genome contained markers that presented an erratic pattern of linkage disequilibrium, which suggests potential mapping and reference genome assembly problems. These regions had a negligible effect on imputation accuracy by individual but can potentially affect GWA studies. While the origin of this erratic linkage disequilibrium pattern is unknown, a precautious strategy would be to exclude markers from these regions as a quality control criterion for GWA studies. To facilitate this, a supplementary table [See Additional file 1: Table S1] summarizes all the observed regions that presented poor imputation performance.
Finally, it is not clear if the Illumina® BovineHD chip should be considered as the target high-density chip for genomic applications in the Nelore breed. Recent genomic prediction results (not published) have revealed that the 50 K and HD chips share similar predictive abilities for different traits in Nelore cattle. At present, it is not clear to what extent the size and composition of the reference population influence these results. As reported by VanRaden et al. , the benefit of using denser chips for genomic prediction becomes more evident as the reference population increases. Denser chips are also preferred for genomic applications that aim at identifying and subsequently using information from causal mutations ,. Except for the most valuable breeding stock (e.g. influential sires and potential donor cows), genotyping the animals with dense chips is prohibitive for most beef cattle operations. Thus, genotyping strategies need to be further investigated to allow the incorporation of genomic information in beef cattle breeding programs in a cost-effective way. The results presented here show that a strategy of genotyping dams and young sire candidates with LD chips to predict missing HD genotypes by imputation is feasible. Future studies are needed to better identify the proper densities of genotyping chips to be used for each category of animals and in which proportion they should be genotyped for each application.
Our results indicate that if the HD chip is considered as the target chip for genomic applications in the Nelore breed, cost-effectiveness can be improved by genotyping part of the economically marginal animals with an LD chip that contains around 15 K useful SNPs and imputing the missing HD genotypes. A denser LD chip (50 K useful SNPs) is recommended for animals that are poorly related to the reference population. For the current Nelore population, FImpute is preferred over BEAGLE for imputation of missing genotypes.
JFG and TSS conceived the study and JFG led its coordination. RC, SAB and HHRN led the analyses and the manuscript preparation. All authors contributed to design the study, discuss the results and review the manuscript. All authors read and approved the final version of the manuscript.
We acknowledge the technical assistance of Guilherme Penteado Coelho Filho and Daniel Biluca and the financial support of Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), process 560922/2010-8. Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by authors and the US Department of Agriculture. The USDA is an equal opportunity provider and employer.
- Meuwissen THE, Hayes BJ, Goddard ME: Prediction of total genetic value using genome-wide dense marker maps. Genetics. 2001, 157: 1819-1829.PubMed CentralPubMedGoogle Scholar
- Schefers JM, Weigel KA: Genomic selection in dairy cattle: integration of DNA testing into breeding programs. Anim Front. 2012, 2: 4-9. 10.2527/af.2011-0032.View ArticleGoogle Scholar
- Khatkar MS, Moser G, Hayes BJ, Raadsma HW: Strategies and utility of imputed SNP genotypes for genomic analysis in dairy cattle. BMC Genomics. 2012, 13: 538-10.1186/1471-2164-13-538.PubMed CentralView ArticlePubMedGoogle Scholar
- Mulder HA, Calus MPL, Druet T, Schrooten C: Imputation of genotypes with low-density chips and its effect on reliability of direct genomic values in Dutch Holstein cattle. J Dairy Sci. 2012, 95: 876-889. 10.3168/jds.2011-4490.View ArticlePubMedGoogle Scholar
- Ferraz JBS, Felício PE: Production systems ¿ an example from Brazil. Meat Sci. 2010, 84: 238-243. 10.1016/j.meatsci.2009.06.006.View ArticlePubMedGoogle Scholar
- Espigolan R, Baldi F, Boligon AA, Souza FR, Gordo DG, Tonussi RL, Cardoso DF, Oliveira HN, Tonhati H, Sargolzaei M, Schenkel FS, Carvalheiro R, Ferro JA, Albuquerque LG: Study of whole genome linkage disequilibrium in Nellore cattle. BMC Genomics. 2013, 14: 305-10.1186/1471-2164-14-305.PubMed CentralView ArticlePubMedGoogle Scholar
- Pérez O¿Brien AM, Mészáros G, Utsunomiya YT, Sonstegard TS, Garcia JF, Van Tassel CP, Carvalheiro R, Silva MVGB, Sölkner J: Linkage disequilibrium levels in Bos indicus and Bos Taurus cattle using medium and high density SNP chip data and different minor allele frequency distributions. Livest Sci. 2014, 166: 121-132. 10.1016/j.livsci.2014.05.007.View ArticleGoogle Scholar
- Pausch H, Aigner B, Emmerling R, Edel C, Götz KU, Fries R: Imputation of high-density genotypes in the Fleckvieh cattle population. Genet Sel Evol. 2013, 45: 3-10.1186/1297-9686-45-3.PubMed CentralView ArticlePubMedGoogle Scholar
- VanRaden PM: Efficient methods to compute genomic predictions. J Dairy Sci. 2008, 91: 4414-4423. 10.3168/jds.2007-0980.View ArticlePubMedGoogle Scholar
- Hill WG, Robertson A: Linkage disequilibrium in finite populations. Theor Appl Genet. 1968, 38: 226-231. 10.1007/BF01245622.View ArticlePubMedGoogle Scholar
- Clayton D: snpStats: SnpMatrix and XSnpMatrix Classes and Methods. In Edited by R package version 1.14.0; 2013. , [http://www.bioconductor.org/packages/release/bioc/html/snpStats.html]
- Browning BL, Browning SR: A unified approach to genotype imputation and haplotype-phase inference for large data sets of trios and unrelated individuals. Am J Hum Genet. 2009, 84: 210-223. 10.1016/j.ajhg.2009.01.005.PubMed CentralView ArticlePubMedGoogle Scholar
- Sargolzaei M, Chesnais JP, Schenkel FS: A new approach for efficient genotype imputation using information from relatives. BMC Genomics. 2014, 15: 478-10.1186/1471-2164-15-478.PubMed CentralView ArticlePubMedGoogle Scholar
- Hickey JM, Crossa J, Babu R, Campos G: Factors affecting the accuracy of genotype imputation in populations from several maize breeding programs. Crop Sci. 2012, 52: 654-663. 10.2135/cropsci2011.07.0358.View ArticleGoogle Scholar
- Daetwyler HD, Calus MPL, Pong-Wong R, Campos G, Hickey JM: Genomic prediction in animals and plants: simulation of data, validation, reporting, and benchmarking. Genetics. 2013, 193: 347-365. 10.1534/genetics.112.147983.PubMed CentralView ArticlePubMedGoogle Scholar
- Badke YM, Bates RO, Ernst CW, Schwab C, Fix J, Van Tassell CP, Steibel JP: Methods of tagSNP selection and other variables affecting imputation accuracy in swine. BMC Genet. 2013, 14: 8-10.1186/1471-2156-14-8.PubMed CentralView ArticlePubMedGoogle Scholar
- Erbe M, Hayes BJ, Matukumalli LK, Goswami S, Bowman PJ, Reich CM, Mason BA, Goddard ME: Improving accuracy of genomic predictions within and between dairy cattle breeds with imputed high-density single nucleotide polymorphism panels. J Dairy Sci. 2012, 95: 4114-4129. 10.3168/jds.2011-5019.View ArticlePubMedGoogle Scholar
- VanRaden PM, Null DJ, Sargolzaei M, Wiggans GR, Tooker ME, Cole JB, Sonstegard TS, Connor EE, Winters M, van Kaam JBCHM, Valentini A, Van Doormaal BJ, Faust MA, Doak GA: Genomic imputation and evaluation using high-density Holstein genotypes. J Dairy Sci. 2013, 96: 668-678. 10.3168/jds.2012-5702.View ArticlePubMedGoogle Scholar
- Ma P, Brøndum RF, Zhang Q, Lund MS, Su G: Comparison of different methods for imputing genome-wide marker genotypes in Swedish and finnish red cattle. J Dairy Sci. 2013, 96: 4666-4677. 10.3168/jds.2012-6316.View ArticlePubMedGoogle Scholar
- VanRaden PM, O¿Connell JR, Wiggans GR, Weigel KA: Genomic evaluations with many more genotypes. Genet Sel Evol. 2011, 43: 10-10.1186/1297-9686-43-10.PubMed CentralView ArticlePubMedGoogle Scholar
- Daetwyler HD, Wiggans GR, Hayes BJ, Woolliams JA, Goddard ME: Imputation of missing genotypes from sparse to high density using long-range phasing. Genetics. 2011, 189: 317-327. 10.1534/genetics.111.128082.PubMed CentralView ArticlePubMedGoogle Scholar
- Boichard D, Chung H, Dassonneville R, David X, Eggen A, Fritz S, Gietzen KJ, Hayes BJ, Lawley CT, Sonstegard TS, Van Tassell CP, VanRaden PM, Viaud-Martinez KA, Wiggans GR: Design of a bovine low-density SNP array optimized for imputation. PLoS One. 2012, 7: e34130-10.1371/journal.pone.0034130.PubMed CentralView ArticlePubMedGoogle Scholar
- Sun C, Wu XL, Weigel KA, Rosa GJM, Bauck S, Woodward BW, Schnabel RD, Taylor JF, Gianola D: An ensemble-based approach to imputation of moderate-density genotypes for genomic selection with application to angus cattle. Genet Res. 2012, 94: 133-150. 10.1017/S001667231200033X.View ArticleGoogle Scholar
- Zhang Z, Druet T: Marker imputation with low-density marker panels in Dutch Holstein cattle. J Dairy Sci. 2010, 93: 5487-5494. 10.3168/jds.2010-3501.View ArticlePubMedGoogle Scholar
- Berry DP, Kearney JF: Imputation of genotypes from low- to high-density genotyping platforms and implications for genomic selection. Animal. 2011, 5: 1162-1169. 10.1017/S1751731111000309.View ArticlePubMedGoogle Scholar
- Visscher PM, Brown MA, McCarthy MI, Yang J: Five years of GWAS discovery. Am J Hum Genet. 2012, 90: 7-24. 10.1016/j.ajhg.2011.11.029.PubMed CentralView ArticlePubMedGoogle Scholar
- Meuwissen T, Hayes B, Goddard M: Accelerating improvement of livestock with genomic selection. Annu Rev Anim Biosci. 2013, 1: 221-237. 10.1146/annurev-animal-031412-103705.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.