Author's Accepted Manuscript

Genomic information in pig breeding: Science

meets industry needs
Noelia Ibañez-Escriche, Selma Forni, Jose Luis
Noguera, Luis Varona

www.elsevier.com/locate/livsci

PII: S1871-1413(14)00256-X
DOI: http://dx.doi.org/10.1016/j.livsci.2014.05.020
Reference: LIVSCI2463

To appear in: Livestock Science

Cite this article as: Noelia Ibañez-Escriche, Selma Forni, Jose Luis Noguera,
Luis Varona, Genomic information in pig breeding: Science meets industry
needs, Livestock Science, http://dx.doi.org/10.1016/j.livsci.2014.05.020

This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal
pertain.
Genomic information in pig breeding: science meets industry needs

Noelia Ibañez-Escriche1*, Selma Forni2, Jose Luis Noguera1, Luis Varona3

1
Genètica i Millora Animal, IRTA, Av. AlcaldeRoviraRoure, 191, 25198, Lleida,
Spain.noelia.ibanez@irta.es; joseluis.noguera@irta.es
2
PIC North America, 100 Bluegrass Commons Blvd., Suite 2200, Hendersonville, TN
37075, USA. Selma.Forni@genusplc.com
3
Unidad de Genética y Mejora Animal, Universidad de Zaragoza, C./ Miguel Servet,
177, 50013, Zaragoza, Spain. lvarona@unizar.es

*Corresponding autor.

Abstract

The inclusion of genomic information on prediction of breeding values has been

explored by the pig breeding industry as molecular techniques evolved.Except for a few

successful cases, for example the HAL-1843®, the use of genomic information for pig

breeding has been limited. The development of SNP chips with high density markers

across the genome coupled with new statistical methods allowed genomic selection

(GS) to become a common practice. GS application entailed important advantages for

pig breeding as it increased the accuracy of the breeding values for selection candidates

and offered an opportunity for the practice of new selection strategies. Nevertheless, the

application of GS is not straightforward for every target trait and breeding scheme.

Many efforts have been done to evaluate the new methods and strategies to efficiently

implement GS in pig breeding. Several aspects of the population and breeding

objectives must be considered before GS is applied. This paper reviews the current

status and challenges faced by pig breeders in the implementation of GS and the future

opportunities that may arise as molecular technologies advance.

1 
 
Key words: Genomic Selection; Pig Breeding; SNP; Accuracy; Crossbreeding; non-

additive inheritance.

1. A brief pig breeding history.

Since the domestication of pig, breeding strategies were applied to adapt the swine

populations to human demands. Initially, the application of breeding techniques was

intuitive and without any scientific criteria; the called genetics science did not emerge

until the beginning of the twenty century. During the last century, the advance of

population and quantitative genetic theories provided new tools for the prediction of

breeding values and allowed a remarkable increase in genetic progress. In the pork

industry, this “genetic” revolution started on the 40’s to 60’s with the application of the

quantitative genetics through selection indexes (Hazel, 1943) and systematic

crossbreeding (Dickerson 1952, 1974). Later on, it was followed by the advent of mixed

model procedures (Henderson, 1984) such as the BLUP (Best Linear Unbiased

Prediction). Until now, BLUP has been the method of choice to obtain predictors of

breeding values for the selection candidates in most pig breeding companies.

Since the late 80’s and 90’s, the availability of neutral molecular markers, such

as the microsatellites, encouraged the development of a plethora of experimental studies

to locate genomic regions and Quantitative Trait Loci (QTL) associated with traits of

economic interest (Bidanel and Rothschild, 2002). A summary of the results provided

by all these experiments are presented in www.animalgenome.org/QTLdb, showing

that, in January, 2014, a total of 9,862 QTL for 653 different traits have been identified.

Unfortunately, the promises that QTL discovery raised in the scientific community were

not followed with a massive identification of causal polymorphisms that could be

2 
 
directly applied by the pig breeding industry. A remarkable exception for that rule was

the RYR1 (Ryanodine Receptor 1) or HAL-1843®‘stress’ gene test in 1991 (Fujii et al.,

1991), which effects were well known even before the development of QTL

experiments. In addition, some other successful studies identified some other interesting

genes for pig breeding, for example, the Estrogen Receptor (ESR; Rothschild et al.,

1996) , the Insuline Growth Factor 2 (IGF2; Van Laere et al., 2003) or the Leptin

Receptor (Ovilo et al., 2005). However, as the number of relevant genes, or markers in

linkage disequilibrium with causal polymorphisms, was small, the application of gene

or marker assisted selection (Van Eenennaam et al., 2014) in pig breeding industry was

minor and it was only used as a complementary tool to the standard mixed model

genetic evaluation.

2. Genomic selection age

The emerging of high throughput technologies (specially the SNP chips of high

density across genome) accelerated the discovery of genomic regions contributing to

genetic variability of complex traits (Genome Wide Association, GWAS) and, in the

last 10 years, some breeding companies included trait specific SNP panels in the genetic

evaluation procedures (Van Eenenaam et al., 2014). Further, the development of new

statistical methods for exploiting this information on genomic prediction (Meuwissen et

al., 2001) changed the landscape of the genetic evaluation and animal breeding and it

allowed a substantial improvement in breeding value prediction. The use of massive

genomic information on the genetic evaluations (called Genomic Selection: GS) entails

important advantages over traditional evaluation. It improves the annual response to

selection by increasing the accuracy of the non phenotyped selection candidates and/or

reducing the generation interval. However, GS require large investments in genotyping

and that can be not cost effective in some cases. A suitable genotyped and phenotyped

3 
 
population (reference population) is required for GS profitability (Meuwissen, 2009).

The appropriate reference population size depends on many parameters such as the

effective population size, the genetic architecture of the trait, the genome size and the

density of the SNP chip (Goddard, 2009). Thus, feasibility of the application of GS is

breeding scheme and population dependent.

2.1 Genomic selection in pig breeding

In the pig breeding industry GS had raised great interest. The accuracy of the

breeding values for the target traits, generally low compared to dairy cattle, can be

significantly increased (Table 1), preserving the generation interval and control of

inbreeding (Lillehammer et al., 2011; Forni et at., 2011). The effectiveness of genome

selection depends largely on the level of linkage-disequilibrium (LD) that can be

captured by markers within the target population. The higher the level of LD, the fewer

markers are needed to capture the genomic regions contributing to the phenotype. In a

survey of several swine lines, Deeb et al. (2010) found that a few thousand equally-

spaced SNP achieved average LD of 0.2 or higher in most pure lines and crosses

evaluated. Moreover, of the 62k markers on the commercially available PorcineSNP60

BeadChip, over 50k SNP had allele frequencies of 5% or higher in most pure-lines and

crosses, indicating that the commercial chip provided enough markers to be effectively

used for genomic selection of all line groups. On the other hand, GS also allowed new

selection schemes (e.g. crossbred use, directed mating) and the introduction of novel

phenotypes in the criteria of selection (e.g. disease resistance). The improvement of

genetic gain on the nucleus has an important impact on the large commercial

populations and can make GS economically feasible (Simianer, 2009), given the large

influence of elite individuals. Pig breeding schemes currently have developed a very

efficient data recording scheme which easily could include genomic information.

4 
 
However, in spite of these advantages, GS in pig has only recently be implemented

(Forni et al., 2011; Ostersen et al., 2011) and it is still not so common as in dairy cattle.

First, the rather recent availability of the SNP chip technology for pigs (Ramos et al.,

2009) has delayed its introduction. Second, the peculiarities of pig breeding schemes

(e.g. small nucleus size, diversity of breeding goals, pyramid system) made genomic

evaluation strategies not straightforward to implement. In the last years, many efforts

have been done to evaluate new methods and strategies to allow efficiently implement

GS on pigs.

2.1.1 Genomic evaluation methods

Genetic evaluation methods that incorporate the genomic information have become a

popular topic in animal breeding. The genomic evaluation could be divided in two main

strategies, the multi-step (VanRaden, 2008) and the single-step (Misztal et al., 2009)

approaches. In the first strategy, multi-step, the response variable is linked to marker

information and later the genomic prediction is merged with the parent average to

estimate a breeding value (EBV). Within this approximation, numerous statistical

approaches have been developed to calculate the prediction of the marker effects, such

as GBLUP, Bayes A, Bayes B, Bayes Cπ, Bayesian Lasso and several non-parametric

procedures, differing in the shrinkage strategy for the marker effects (Gianola, 2013).

However, relevant differences have not been found between different prediction

methods in field data (Hayes et al., 2009a), in contrast with simulated data (Meuwisen

et al., 2001). Particularly, in pigs, Ostersen et al. (2011) did not find differences

between prediction methods for daily gain and feed conversion. The authors also

showed that the variable response (e.g. deregressed EBV) could be used for pig

genomic evaluations. Nevertheless, one of the main limitations of the multi-step

procedures is that the genomic prediction are only available for the genotyped

5 
 
individual and they are not immediately analogous with EBV calculated from the

standard mixed model procedures (Legarra et al., 2009). A single step strategy was

developed (Aguilar et al., 2010; Christensen and Lund, 2010) to avoid this shortcoming.

It consists of a modification of the classical mixed model BLUP in which the additive

relationship matrix (A) is extended including the genomic relationship matrix (G) for

the genotyped individuals. For animals with genotypes, predictions are equivalent to the

GBLUP approach under the multi-step strategy (Christensen et al., 2012), but it

provides direct breeding values predictions for all animals in the pedigree (genotyped

and not genotyped). Besides the computation of a new relationship matrix, no further

modifications in models and software are required and the method has become popular

for computational straightforwardness; especially in pig breeding where information is

accumulated and genetic evaluation is computed as often as weekly (Forni et al., 2011,

Christensen et al., 2012). Recently, Fernando et al. (2013) proposed an extension of the

single-step method that allows to model markers effects with alternative shrinkage

methods (e.g. Bayes B, Bayes C, Bayesian Lasso, etc.) that can be potentially

advantageous for traits controlled by few genes of large effect (Resende et al., 2012 ).

2.1.2. Genotyping cost

One of the main limitations of the GS implementation in pig breeding is the

genotyping cost. The number of candidates to selection for genotyping can be large and

their economic value is considerable lower than dairy young bulls.Generation interval in

pig herds is also smaller and forces a constant increase of genotypes and phenotypes

from the reference population. Favourably, the LD in commercial pig lines is much

larger than in the cattle herds (Veronezeet al., 2013) and relatively smaller reference

populations than dairy cattle can be used. Simulation studies showed that after some

generations of selection the genomic prediction accuracy falls quickly (Ibáñez-Escriche

6 
 
and Blasco, 2011). Results of genomic evaluation in pig reproductive traits (Cleveland

et al., 2010) reported a large loss of accuracy when the training and validation data sets

were separated by several generations (based on birth year). The establishment of

genotyping strategies to return the economic investment is a key point for the practical

GS in pig breeding. In a simulation study following the Norwegian Landrace maternal

breeding scheme (Lillehamer et al., 2013), a base reference population of 1,200

genotyped animals per year was sufficient large to improve traditional selection. The

authors did not found great differences on cost-benefit between genotyping strategies

(from 1,200 to 4,800 genotyped animals per year), since a higher number of genotyped

individual increased both the investment and the genetic gain. Nevertheless, they

recommended increasing the base reference population (1,200 genotyped males per

year) with some genotyped females instead of more males per litter. This strategy

reduced the rate of inbreeding providing a similar or even greater genetic response.

When male breeding schemes were simulated (Tribout et al., 2012), profitability results

were heterogeneous, depending on the availability of recorded phenotypes. Traits with a

systematic recording on the candidates of selection (both sex) required a large

genotyped reference population (~13.000 genotypes per year) to improve gains from

traditional genetic evaluation. However, traits which only a small number of relatives

were phenotyped (e.g. meat quality) required a relative small genotyped reference

population (1,000 animals) to improve the expected genetic gain. The strategy that

maximizes the return on the investment depends directly on the size of the production

strata in the pyramidal scheme and the definition of optimum strategies cannot be

generalized. Additionally, when the comparison between GS and traditional BLUP is

made at same rate of additional investment, Tribout et al. (2013) found that in a pig

male line GS take advantages over BLUP starting from medium-high investments.These

7 
 
authors also evaluate the use of a low-cost genotype strategy combined with genotype

imputation and found that it was one of the most profitable strategies.

Imputation is a process that uses empirical or statistical rules to fill out missing

genotypes because either the marker was not on the genotyping panel or was not called.

Imputation tests performed on pigs (Huagh et al 2012; Cleveland and Hickey 2013)

showed that the cost of genotyping could be greatly reduced when genotyping selection

candidates for a small panel and sires and grandsires for the full PorcineSNP60 with a

small reduction in accuracy of GEBV (Genomic Estimated Breeding Values). Adding

low-density genotypes for dams offered also small additional increases at reduced cost.

They reported imputation accuracies exceeding 0.96 for a panel of 384 SNP when

parents and grandparents were genotyped at high density. Accuracies decreased with

declining levels of genotyping of close relatives but remained above 0.90 when at least

sires and grandsires were genotyped. Nevertheless, the accuracy of GEBV is

considerably affected for the size of high-density reference population (Cleveland and

Hickey 2013). Low-cost genotyping coupled with imputation is an alternative to

genotyping the large number of selection candidates as well as the females in pig

populations. However, research efforts are still needed to explain deeply the relation

between individual imputation accuracy and overall genetic gain. Nevertheless, the use

of trait-line specific low-density panels is a very attractive tool for traits of economic

importance that are difficult to measure as disease resistance (Van Eenenaam et al.,

2014).

Another alternative strategy to reduce the cost of genotyping cost is to share

genotypes from different breeding companies (e.g. Eurogenomics; Lund et al., 2011) or

populations (e.g. Norwegian Red breed; Brøndum et al., 2011) to a joint genomic

evaluation. Some published results in dairy and beef cattle indicated that the accuracy of

8 
 
multibreed genomic evaluations depends on the genetic distance among populations and

the marker density (Hayes et al., 2009b; Kizilkaya et al., 2010). However, this will not

be an easy task on pig breeding industry because there are many breeding lines with

different breeding goals (within and between companies) and they can be genetically

distant.

2.2 Handle with dominance and epistatic effects

Traditionally, genetic evaluation of pigs has been focused on the prediction of

breeding values as components of the additive genetic variance. However, under the

assumption of the infinitesimal model, the genetic components of the phenotype also

include dominant and epistatic (or interaction) effects (Falconer, 1989). The use of these

non-additive genetic effects emerges as an interesting alternative since the availability

of massive genotyping techniques. Estimates of dominance variance before the

upcoming of genomic information were rare but Culbertson et al.(1998) and Norris et

al. (2010) found that dominance represent a relevant percentage of genetic variation in

some traits of interest in pig breeding. Further, there are several evidences that epistatic

interactions can play an important role on the genetic background of reproductive

(Noguera et al., 2009), meat quality and carcass traits (Groβe-Brinkhaus et al., 2010).

Estimation of dominance or epistatic variance has become easier with the

availability of genomic information, as it can provide information for a wider range of

genetic relationships between phenotyped individuals. A first attempt of estimation of

dominance and epistatic variance with genomic data was presented by Su et al. (2012)

for daily gain in a Danish Duroc population. The authors found that the dominance and

additive x additive variation accounted for 5.6% and 9.5% of the total phenotypic

variance, respectively. For the dominance variance, the approach presented by Su et al.

9 
 
(2012) used a single step genomic evaluation (Aguilar et al., 2011) and introduced the

dominant (co) variance between individuals. However, this approach presents a

genotypic parameterization of the additive and the dominance effects. As a

consequence, the estimates obtained with this model cannot be directly compared with

the ones generated by the classical quantitative genetic approach. Recently, Vitezica et

al. (2013) presented an alternative that avoid this shortcoming. Their model expressed

the genetic variation in terms of breeding values and dominance deviations, leading to

higher estimates of relative dominance variation.

A direct application of a genomic evaluation that incorporates dominant, or even

epistatic variation, is the prediction of genetic merit of future matings. Under a simple

additive and dominant model, Toro and Varona (2010) found that the potential increase

of response in a first generation of selection ranges between 6 to 22% depending of the

heritability and the percentage of dominance variation. However, gains do not

accumulate across generations (Toro and Varona, 2010). Alternatively, non-linear

models have been also proposed as an alternative to capture interaction effects such as

dominance or epistasis (Gianola and van Kaam, 2008). Tusell et al. (2012) found that

nonlinear neural networks were the best method to predict future performance of

crossbred populations for litter size in pigs. Nevertheless, the differences from the

genomic linear models were not substantial.

Crossbreeding is widely used in pork production to exploit heterosis and breed

complementarity (Visscher et al., 2000). A suitable joint genomic evaluation of

purebreds could be performed using crossbred pigs as reference population (Ibáñez-

Escriche et al., 2009). Kinghorn et al. (2010) and Zeng et al. (2013) proposed an

expansion of this approach that includes the introduction of non-additive effects.

Kinghorn et al. (2010) suggested the use of allelic frequencies of either the pure or the

10 
 
crossbred populations depending if the objective is to select individuals to increase

performance in one or the other. When the allelic frequencies of the complementary

populations are used, the procedure mimics the goals of the classical reciprocal

recursive selection. The used of allelic frequencies from the crossbred animals could

allow greater accuracy of genetic additive effects for traits under dominance inheritance

(Kinghorn et al. 2010; Zeng et al., 2013). This strategy might lead to a more effective

selection for performance in the field, specially when important traits are not

measurable in nucleus such as disease resistance (Dekkers, 2007). However, as Ibañez-

Escriche and Recio (2010) discussed, this strategy may impact accuracy negatively and

that should be carefully considered. The reliability of field records can be low and

recording system in the populations must be well designed and implemented.

Additionally, there is a generation lag between crossbreeding and selection candidates

that is difficult to reduce and it could seriously impact the overall genetic gain.

Pig breeding is based on a classical pyramidal structure where several elite lines

provide genetic material to a large base of production animals. The elite lines are

sometimes composed by a small number of individuals and restriction of inbreeding is a

key element on the genetic management of the populations under selection. Marker

information can be used not only for genomic prediction of breeding values but also for

the prediction of the molecular coancestry of future matings. Sonneson et al. (2010)

generalized the optimum contribution strategy developed by Meuwissen (1997) by

using molecular information to calculate the coancestry of a potential candidate to

selection with the remaining members of the population. The author showed that

genomic selection requires a molecular control of inbreeding, mainly in populations of a

limited size as pig populations. Recently, Sun et al. (2013) suggested combining

11 
 
dominance predictions of future matings with control of inbreeding to determine an

optimum mating design by linear programming.

Both the prediction of non-additive genetic performance and coancestry leads to

specific ranking of sires and dams for each particular mate. The practical consequences

of using non-additive genetic variation in pig breeding are important as the ranking of

selection candidates is not limited to a list of animals ordered by their additive genetic

prediction. The ranking of individuals depends also on its future mate and with the

application of new technologies will allow creating a specific ranking of sires and dams

depending of the male or female candidate for mating, aided through automatic web

calculations.

2.3. Practical aspects of the implementation of genomic selection

The implementation of genomic selection imposed several challenges for pig

breeding companies, mainly related to the logistic and storage of tissue samples and

genotype information. All genotype information must be available for inclusion in the

genetic evaluation before selection to maximize the benefit of molecular information on

accuracy. In the current genotype schemas, this includes the genotypes of the selection

candidates and of their relatives at the genetic nucleus level. The process of capturing

genomic information begins in the nucleus farms with the collection of tissue sample for

DNA extraction. Collection of tissue is usually done on pigs typically within 24 hours

of birth, at the same time the birth is recorded and animals are individually tagged.

Tissue samples, generally the tail, are collected and placed into uniquely bar-coded

tubes linked with the piglet tag and recorded directly into a database using hand held

computers. On arrival at the tissue storage facility, labels are scanned and freezer

location is recorded to facilitate sample identification at the moment of DNA extraction.

12 
 
The DNA extraction for genotyping, the genotyping process itself and the resulting

manipulation of the genomic data, including quality control, must happen within 8 to 10

weeks of age. Breeding companies were required to develop automated processes and

invest in new facilities, hardware and employee training to implement GS (McLaren et

al, 2013).

Most pig breeding companies reported to be investing and implementing GS in

the last 5 years; a few major players have it fully implemented in their selection

schemes for all traits and lines. The single step method (Aguilar et al., 2010;

Christensen and Lund, 2010) has been the most used strategy because it is simpler to

compute. Breeding values in commercial pig breeding need to be updated weekly.

Improvements in accuracy of selection with the single step method were reported by

Forni et al. (2011) and Christensen et al. (2012). Overall return on investments to

implement genomic selection is positive, especially in maternal lines that are strongly

selected based on reproductive performance. Genomic information has a greater impact

on reproductive traits and piglet robustness (Table 1).

The increasing number of genotyped animals in the genetic evaluation is the

biggest challenge that the industry will have to address in the near future. Computing

time for the genetic evaluation tends to increase exponentially with the number of

animals genotyped. High-performance computing methods and data storage systems

need to be developed to handle the number of genotypes that pig breeding companies

will be using in the next few years. Aguilar et al. (2011) showed that single-step

approaches have similar computational requirements to traditional BLUP but the

limiting factor is the construction and inversion of the genomic relationship matrix for a

large number of animals. Major developments in statistical and computational methods

applied to animal breeding will be necessary to overcome the challenge to include a full

13 
 
genotyped pedigree in pig genetic evaluations that may be available in the next few

years.

3. Conclusions

Genomic selection is a reality in pig breeding and could allow new strategies of

selection and utilization of non-additive effects to maximize genetic gain. It may not be

feasible to apply this new technology on every trait and breeding scheme. The efficient

implementation of GS requires a great investment in genotyping, logistic and hardware

that it is not always available. Genotyping a large number of animals every year is an

important part of the process and has only been possible through low cost genotyping

and imputation developments. Many practical and scientific issues must be addressed

before GS becomes common practice in all pig breeding schemes and research on the

genetic background of specific populations are required. Factors that will help the

dissemination of GS in pig breeding are the patterns of linkage disequilibrium in

selected populations, the efficiency of imputation techniques, the pyramidal structure

and the systematic exploitation of crossing and heterosis. However, some particularities

of pig breeding schemes, like reduced generation interval, small value of selection

candidates and changes in linkage disequilibrium after a few generations, play against

the economic efficiency of GS. Finally, it is worthy to remark that the implementation

of GS can be also linked by some economic factors not directly related with the

additional increase of productive performance of pigs, such as the market race between

breeding companies.

Acknowledgements

This work was partially funded by the Ministerio de Economía y

Competitividad, Spain (ProjectRTA2012-00054-C02-01).

14 
 
References

Aguilar, I., Misztal, I., Johnson, D., Legarra, A., Tsuruta, S., Lawlor, T., 2010. A

unified approach to utilize phenotypic, full pedigree, and genomic information for

genetic evaluation of Holstein final score. J. Dairy Sci. 93, 743–752.

Aguilar, I.,Misztal, I.,Legarra, A.,Tsuruta, S., 2011. Efficient computations of the

genomic relationship matrix and other matrices used in single-step evaluation. J. Anim.

Breed. Genet. 128, 422-428.

Bidanel, J.P.,Rothschild, M.F., 2002.Current status of quantitative trait locus mapping

in pigs. Pig News and Information 23(2) 39N-53N.

Brøndum, R.F., Rius-Vilarrasa, E., Strandén, I., Su G., Guldbrandtsen, B., Fikse, W.F.,

Lund, M.S., 2011. Reliabilities of genomic predictions using combined reference data

of the Nordic Red cattle populations. J. Dairy Sci. 94, 4700-4707.

Christensen, O.F., Lund, M.S., 2010. Genomic prediction when some animals are not

genotyped. Genet. Sel. Evol.42:2.

Christensen, O.F., Madsen, P., Nielsen, B., Ostersen, T., Su, G., 2012. Single-step

methodsfor genomic evaluation in pigs.Anim.6, 1565-71.

Cleveland, M.A., Forni, S., Garrick, D.J., Deeb, N., 2010. Prediction of genomic

breeding values in a commercial pig population. Proc. 9th World Congress on Genetics

Applied to Livestock Production. Leipzig, Germany. Aug 2-6, p. 506.

Cleveland, M.A., Hickey, J.M., 2013.Practical implementation of cost-effective

genomic selection in commercial pig breeding using imputation. J. Anim. Sci. 91, 3583-

92.

15 
 
Cleveland, M.A., Hickey, J.M., Forni, S., 2012. A common dataset for genomic analysis

of livestock populations.G3 2, 429-35.

Cultbertson, M.S., Mabry, J. W., Misztal, I., Gengler, N., Bertrand, J. K., Varona, L.,

1998. Estimation of dominance variance in purebred Yorkshire swine. J. Anim. Sci. 76,

448-51.

Deeb, N., Cleveland, M.A., Forni. S., 2010 Linkage disequilibrium decay in

commercial pigs. Plant and Animal Genome XVIII Conference, San Diego, CA

USAP602 (Abstr.).

Dekkers, J.C., 2007. Marker-assisted selection for commercial crossbred performance.

J. Anim. Sci. 85, 2104-14.

Dickerson, G.E., 1952. Inbred lines for heterosis tests?InGowen J.W. (ed.) Heterosis,

Iowa State University,Ames, Iowa (USA), pp 330 – 351.

Dickerson, G.E., 1974. Evaluation and utilization of breed differences. In: Proceedings

of the workingsymposium on breed evaluation and crossing with farm animals. Zeist,

September 1974.

Falconer, D.S., 1989.Introduction to Quantitative Genetics. Longmans, London.

Fernando, R.L., Garrick, D., Dekkers, J.C.M., 2013. Bayesian regression method for

genomic analysis with incomplete genotype data.Proc. 64th Annual meeting of the

European Association of Animal Production, Nantes, France. p. 225.

Forni, S., Aguilar, I., Misztal, I., 2011. Different genomic relationship matrices for

single-step analysis using phenotypic, pedigree and genomic information. Genet. Sel.

Evol. 43(1): 1.

16 
 
Fujii, J., Otsu, K., Zorzato, F., De leon, S., Khanna, V.K., Weiler, J.E., O’Brien, P.J.,

MacLennan, D.H., 1991. Identification of a mutation in porcine ryanodine receptor

associated with malignant hyperthermia. Science 25, 448-451.

Gianola, D., van Kaam, B.C.H.M., 2008. Reproducing kernel Hilbert spaces

regressionmethodsforgenomicprediction of quantitativetraits.Genetics 178, 2289-2303.

Gianola, D., 2013. Priors in whole-genome regression: the Bayesian alphabet returns.

Genetics 194, 573-96.

Goddard, M.E., 2009. Genomic selection: prediction of accuracy and maximisation of

long term response. Genetica 136, 245–257.

Groβe-Brinkhaus, C., Jonas, E., Buschbell, H., Phatsara, C., Tesfaye, D., Jüngst, H.,

Looft C., Schellander, K.,Tholen, E., 2010. Epistatic QTL pairs associated with meat

quality and carcass composition traits in a porcine Duroc x Pietrain population. Genet.

Sel. Evol. 42:39.

Hayes, B.J., Bowman, P.J., Chamberlain, A.C., Verbyla, K., Goddard, M.E., 2009b.

Accuracy of genomic breeding values in multi-breed dairy cattle population. Genet. Sel.

Evol.41:51.

Hayes, B.J., Bowman, P.J., Chamberlain, A.J., Goddard, M.E., 2009a. Invited review:

Genomic selection in dairy cattle: progress and challenges. J. DairySci. 92, 433-43.

Hazel, L.N., 1943.The genetic bases for constructing selection indexes. Genetics

28,476-490.

Henderson, C.R., 1984.Applications of Linear Models in Animal Breeding. Guelph,

Ont: University of Guelph.

17 
 
Hu, Z., Li Y., Song, X., Han, Y., Cai, X., Xu, S, Li. W., 2011.Genomic value prediction

for quantitative traits under the epistatic model. BMC Genetics 12:15.

Huang, Y., Hickey, J.M., Cleveland, M.A., Maltecca, C., 2012.Assessment of

alternative genotyping strategies to maximize imputation accuracy at minimal cost.

Genet. Sel. Evol. 44:25.

Ibañez-Escriche, N, Gonzalez-Recio, O., 2011. Review. Promises, pitfalls and

challenges of genomic selection in breeding programs. Span. J. Agric. Res. 9, 404–413.

Ibañez-Escriche, N., Blasco, A., 2011. Modifying growth curve parameters by multitrait

genomics. J.Anim.Sci. 89, 661-669.

Ibañez-Escriche, N., Fernando, R., Toosi, A., Dekkers, J., 2009. Genomic selection of

purebreds for crossbred performance. Genet. Sel. Evol.41: 12.

Kinghorn, B.P., Hickey J.M., Van der Werf, J.H.J., 2010. Reciprocal Recurrent

Genomic Selection for Total Genetic Merit in Crossbred individuals. Paper 0036. In

Proceedings of the 9th world congress on genetics applied to livestock production, 1–6

August 2010, Leipzig, Germany.

Kizilkaya, K., Fernando, R.L., Garrick, D.J., 2010. Genomic prediction of simulated

multibreed and purebred performance using observed fifty thousand single nucleotide

polymorphism genotypes. J. Anim. Sci. 88, 544-551.

Legarra, A., Aguilar, I., Misztal, I., 2009. A relationship matrix including full pedigree

and genomic information. J. Dairy Sci. 92, 4656–4663.

Lillehammer M., Meuwissen T.H., Sonesson A.K., 2011. Genomic selection for

maternal traits in pigs. J. Anim. Sci. 89, 3908-16.

18 
 
Lillehammer, M., Meuwissen, T.H., Sonesson, A.K., 2013 Genomic selection for two

traits in a maternal pig breeding scheme. J. Anim. Sci. 91, 3079-87.

Lund, M.S., de Roos, A.P.W., de Vries, A.G., Druet, T., Ducrocq, V., Fritz, S.,

Guillaume, F., Guldbrandtsen, B., Liu, Z., Reents, R., Chrooten, C., Seefried, F., Su, G.,

2011. A common reference population from four European Holstein populations

increases reliability of genomic predictions. Genet. Sel. Evol. 43:43.

McLaren, D.G., Cleveland, M.A., Deeb, N., Forni, S., Mileham, A.J., Newman, S.,

Southwood, O. I., Wang, L., 2013. 9th International Conference on Pig Reproduction,

Olsztyn, Poland.

Meuwissen, T.H., 2009. Accuracy of breeding values of ‘unrelated’ individuals

predicted by dense SNP genotyping. Genet.Sel.Evol. 41:35.

Meuwissen, T.H., Hayes, B.J., Goddard, M.E., 2001. Prediction of total genetic value

using genome-wide dense marker maps.Genetics 157, 1819-1829.

Meuwissen, T. H. E., 1997. Maximizing the response of selection with a predefined rate

of inbreeding. J. Anim. Sci. 75, 934-940.

Misztal, I., Legarra, A., Aguilar, I., 2009. Computing procedures for genetic evaluation

including phenotypic, full pedigree, and genomic information. J.DairySci. 92, 4648–

4655.

Noguera, J.L., Rodríguez, C, Varona, L., Tomás, A., Muñoz, G., Ramírez, O., Barragán,

C, Arqué, M., Bidanel, J.P., Amills, M., Oviló, C., Sánchez, A., 2009. A bi-dimensional

genome scan for prolificacy traits in pigs shows the existence of multiple epistatic QTL.

BMC Genomics. 10:636.

19 
 
Norris, D., Varona, L., Ngambi, J.W., Visser, D.P. Mbajiorgu, C. A., Vooderwing, S.

F., 2010. Estimation of the additive and dominance variance in SA Duroc pigs. Lives.

Sci. 131,144-147.

Ostersen, T., Christensen, O.F., Henryon, M., Nielsen, B., Su, G., Madsen, P., 2011.

Deregressed EBV as the response variable yield more reliable genomic predictions than

traditional EBV in pure-bred pigs.Genet. Sel.Evol. 43:38.

Ovilo, C., Fernández, A., Noguera, J.L., Barragán, C., Letón, R., Rodríguez C.,

Mercadé, A., Alves, E., Folch, J.M., Varona, L., Toro, M., 2005. Fine mapping of

porcine chromosome 6 QTL and LEPR effects in body composition in multiple

generations of an Iberian by Landrace intercross. Genet. Res. 85, 57-67.

Ramos, A.M., Crooijmans, R.P.M.A., Affara, N.A., Amaral, A.J., Archibald, A.L., et

al., 2009. Design of a high density SNP genotyping assay in the pig using SNPs

identified and characterized by next generation sequencing technology. PLoS

ONE 4: e6524.

Resende, M.F. Jr Muñoz, P., Resende, M.D., Garrick, D.J., Fernando, R.L., Davis

J.M., Jokela, E.J., Martin, T.A., Peter, G.F., Kirst, M., 2012. Accuracy of genomic

selection methods in a standard data set of loblolly pine (Pinustaeda L.). Genetics 190,

1503-10.

Rothschild, M., Jacobson, C., Vaske, D., Tuggle,, C., Wang, L., Short, T., Eckardt, G.,

Sasaki, S., Vincent,, A., McLaren, D., Southwood, O., van der Steen, H., Mileham, A.,

Plastow, G., 1996. The estrogen receptor locus is associated with a major gene

influencing litter size in pigs. Proc. Natl. Acad. U.S.A. 93:201-5.

20 
 
Simianer, H., 2009. The potential of genomic selection to improve litter size in pig

breeding programs. Proc. 60th Annual meeting of the European Association of Animal

Production, Barcelona, Spain. p. 210.

Smith, C., Smith, D.J., 1993. The need for close linkages in markers-assisted selection

for economic merit in livestock. Anim. Breed. Abst. 61, 197-204.

Sonesson, A.K., Woolliams, J.A., Meuwissen, T.H.E., 2012. Genomic selection requires

genomic control of inbreeding. Genet. Sel.Evol.44:27.

Su, G., Christensen, O.F., Ostersen, T., Henryon, M, Lund, M.S., 2012. Estimating

additive and non-additive genetic variances and predicting genetic merits using genome-

wide dense single nucleotide polymorphism markers. Plos One 7(9):e45295.

Sun, C.,VanRaden, P.M., O’Connell, J.R., Weigel, K.A., Gianola, D., 2013. Mating

programs including genomic relationships and dominance effects. J. Dairy Sci. 96,1-10.

Toro, M.A., Varona, L., 2010. A note on mate allocation for dominance handling in

genomic selection. Genet. Sel. Evol. 42:33.

Tribout, T., Larzul, C., Phocas, F., 2012.Efficiency of genomic selection in a purebred

pig male line. J. Anim. Sci. 90, 4164–4176.

Tribout, T., Larzul, C., Phocas, F., 2013.Economic aspects of implementing genomic

evaluations in a pig sire line breeding scheme. Genet. Sel. Evol.45:40.

Tusell, L., Perez, P., Forni, S., Wu, X.L., Gianola, D., 2012.Genome-enabled methods

for predicting litter size in pigs: a comparison. Anim. 11, 1739-1749.

21 
 
Van Eenennaam, A.L., Weigel K.A., . Young, A.E., Cleveland, M.A., Dekkers, J.CM.,

2014. Applied Animal Genomics: Results from the field. Ann. Rev. Anim. Bio. 2, 105-

139.

Van Laere, A.S., Nguyen, M.,Braunschweig, M., Nezer, C., Collete, C., Moreau, L.,

Archibald, A.L., Haley, C.S., Buys, N., Tally, M., Andersson, G., Georges, M.,

Andersson, L., 2003. A regulatory mutation in IGF2 causes a major QTL effect on

muscle growth in the pig. Nature 425, 832-836.

VanRaden, P.M., 2008. Efficient methods to compute genomic predictions. J. Dairy Sci.

91, 4414-4423.

Veroneze, R., Lopes, P.S., Guimarães, S.E.F., Silva, F.F., Lopes, M.S., Harlizius, B.,

Knol, E.F., 2013. Linkage disequilibrium and haplotype block structure in six

commercial pig lines. J. Anim.Sci. 91, 3493-3501.

Visscher, P., Pong-Wong, R., Whittemore, C., Haley, C., 2000.Impact of biotechnology

on (cross) breeding programmes in pigs.Livest. Prod. Sci. 65, 57–70.

Vitezica, Z., Varona, L., Legarra, A., 2013. On the additive and dominant variance and

covariance of individuals within the genomic selection scope. Genetics 195, 1223-30.

Watson, J.D., Crick, F.H.C., 1953. A Structure of Deoxyribose Nucleic Acid.Nature

171:737-8.

Wellman, R., Bennewitz, J., 2012. Bayesian models with dominance effects for

genomic evaluation of quantitative traits. Genet. Res. 94, 21-37.

22 
 
Zeng, J., Toosi, A., Fernando, R.L., Dekkers, J.C., Garrick, D.J., 2013. Genomic

selection of purebred animals for crossbred performance in the presence of dominant

gene action. Genet. Sel. Evol. 45:11.

Table1. Results provided by Genus company (Selma Forni, personal

communication). Average accuracy obtained with the prediction error variance of
breeding values with pedigree relationships (EBV) and with combined pedigree
and genomic relationships (GEBV) for young pigs at the moment of selection.

Accuracy

Trait EBV GEBV

Total number of piglets born 0.25 0.42

Growth rate in commercial environment

0.29 0.50
(gr/day)

Mortality from birth to weaning in

0.02 0.10
commercial environment (dead/alive)

Feed intake (kg/day) 0.27 0.50

Loin depth (mm) 0.28 0.51

23 
 
Highlights: 

We present a review of the application of the genomic information in pig breeding in 
which is described: 

ƒ Current issues of Genomic Selection in pig breeding 
ƒ New Scenarios  of pig breeding Genomic Selection 
ƒ New application possibilities of Genomic selection in pig breeding 
ƒ Future challenges of Pig breeding Genomic selection 
 

24 
 
Conflict of interest

‐ We wish to confirm that there are no known conflicts of interest associated with this  
 
publication and there has been no significant financial support for this work that could have 
influenced its outcome.  
 
‐ We confirm that the manuscript has been read and approved by all named authors and that 
there are no other persons who satisfied the criteria for authorship but are not listed. We 
further confirm that the order of authors listed in the manuscript has been approved by all of 
us.  
 
‐ We confirm that we have given due consideration to the protection of intellectual property 
associated with this work and that there are no impediments to publication, including the 
timing of publication, with respect to intellectual property. In so doing we confirm that we 
have followed the regulations of our institutions concerning intellectual property.  
 
‐ We understand that the Corresponding Author is the sole contact for the Editorial process 
(including Editorial Manager and direct communications with the office). She is responsible for 
communicating with the other authors about progress, submissions of  
 
revisions and final approval of proofs. We confirm that we have provided a current, correct 
email address which is accessible by the Corresponding Author and which has been configured 
to accept email from (noelia.ibanez@irta.es). 

25