SEARCH

SEARCH BY CITATION

Keywords:

  • dwarfism;
  • Friesian;
  • gene;
  • GWAS;
  • horse;
  • SNP association

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflicts of Interest
  9. Web URLs
  10. References

The recent completion of the horse genome and commercial availability of an equine SNP genotyping array has facilitated the mapping of disease genes. We report putative localization of the gene responsible for dwarfism, a trait in Friesian horses that is thought to have a recessive mode of inheritance, to a 2- MB region of chromosome 14 using just 10 affected animals and 10 controls. We successfully genotyped 34 429 SNPs that were tested for association with dwarfism using chi-square tests. The most significant SNP in our study, BIEC2-239376 (P2df = 4.54 × 10−5, Prec = 7.74 × 10−6), is located close to a gene implicated in human dwarfism. Fine-mapping and resequencing analyses did not aid in further localization of the causative variant, and replication of our findings in independent sample sets will be necessary to confirm these results.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflicts of Interest
  9. Web URLs
  10. References

The recent completion of a draft sequence of the horse genome (Wade et al. 2009), combined with technological and methodological advances in the analysis of complex genetic traits in humans (Steemers & Gunderson 2007), has provided veterinary genetic epidemiologists with an array of tools for the study of diseases with a heritable component. Specifically, the advent of high-density single nucleotide polymorphism (SNP)-based genotyping arrays and the concomitant growth in knowledge of the haplotypic structure of mammalian genomes has led to the adoption of genome-wide association studies (GWAS) for disease gene mapping (Frazer et al. 2007; Orr & Chanock 2008). These studies, which have recently been successful in the identification of large numbers of genetic loci contributing to disease in humans, exploit the high degree of correlation between genetic variants at any particular region of a chromosome to efficiently interrogate the entire genomes of large numbers of samples using a minimally redundant set of the so-called tagging-SNPs (Howie et al. 2006).

The GWAS paradigm in human studies has been to genotype hundreds of thousands of SNPs in many thousands of affected and unaffected individuals. These large numbers are a result of the hypothesis that multiple loci with a range of effect sizes contribute to the aetiology of complex diseases; large numbers of samples are thus required to attain the requisite statistical power for unambiguous detection of such loci. Charlier et al. (2008) and Karlsson et al. (2007) have demonstrated that association-based mapping can be effective in the detection of single locus recessive traits in animals using much smaller numbers of individuals.

Within the Friesian horse breed, congenital dwarfism has been recognized for many years and occurs at a frequency of 0.25% (Osinga 2000; Back et al. 2008). We recently reported a full phenotypic characterization for Friesian dwarfs, and we refer readers to that article for full details of the features of the trait (Back et al. 2008). Briefly, the Friesian dwarf phenotype results from physeal growth retardation in both limbs and ribs, reflected in a characteristic disproportional growth disturbance. The potential for post-natal growth in these animals, albeit at a reduced rate, is responsible for mature dwarfs having a head of the same size as unaffected animals, a broader chest with narrowing at the costochondral junction, a disproportionally long back and abnormally short limbs. Furthermore, radiographs reveal a dysplastic metaphysis of the distal metacarpus and metatarsus. Light microscopy of growth plates at the costochondral junction demonstrates an irregular transition from cartilage to bone, and thickening and disturbed formation of chondrocyte columns, which is similar to findings in osteochondrodysplasia (Back et al. 2008).

The molecular mechanism underlying the growth disturbance in Friesian foals has yet to be determined, though the trait is heritable and appears to follow an autosomal recessive pattern of transmission (Osinga 2000; Back et al. 2008). Here, we report on the mapping of the genetic determinants of dwarfism in the Friesian horse using GWAS methodologies in a small number of Friesian dwarfs and normal controls.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflicts of Interest
  9. Web URLs
  10. References

Horses

Ten Friesian dwarf horses were selected by the investigators over an 8- year period (2001–2008) based on their unique phenotype (Back et al. 2008) and availability of DNA. Ten normal Friesian horses (controls) were selected in 2008 based on a normal phenotype. In both cases and controls, first- and second-degree relatives were excluded from the study. Pedigrees were verified by microchip and/or passport of the Royal Friesian Horse Studbook (KFPS). Ten millilitres of heparinized blood samples were taken and stored at −80 °C. DNA isolation was performed using QIAamp DNA blood maxi kit from Qiagen according to the manufacturer’s instructions. DNA samples were quantified using Quant-iT PicoGreen dsDNA kits (Invitrogen) according to the manufacturer’s instructions, and the DNA concentrations were adjusted to 20 ng/μl.

Genotyping

Samples were genotyped using EquineSNP50 Genotyping BeadChips (Illumina). This array contains approximately 54 000 SNPs ascertained from the EquCab2 SNP database of the horse genome (http://www.broadinstitute.org/mammals/horse) and has an average spacing of 43.2 kb between adjacent variants. Genotyping was performed in-house on an Illumina Bead Station according to the manufacturer’s recommended protocol. The samples that were genotyped for this study were a subset from 96 equine DNA samples that were genotyped in the same batch, thus ensuring an adequate number of samples for genotype cluster seeding. We included 21 pairs of duplicate samples for QC purposes and observed 100% concordance between genotype calls for 19 duplicate pairs and 99.9% concordance in the remaining two pairs. No samples were excluded on the basis of low assay completion rates. We successfully genotyped 54 633 loci; the overall genotype completion rate was greater than 99.9% (no-call rate = 7.7 × 10−4). We omitted from further analysis 20 204 SNPs that were monomorphic, as well as those that had minor allele frequencies (MAF) less than 5% in our samples, leaving 34 429 SNPs in our working build of the data.

Fine-mapping

We attempted to localize the most significant association by genotyping an additional 319 evenly spaced SNPs from EquCab 2.0 across the region spanning from 3.16 to 5.7 MB on chromosome 14. We included an additional 55 SNPs that were present on the EquineSNP50 array for confirmation of genotype calls from the GWAS. Genotyping for fine-mapping was performed by Sequenom and was conducted using iPlex chemistry on the MassArray platform. Twenty-two SNPs failed to genotype, and the overall call rate of the remaining loci was 94%.

Resequencing

PROP1 was identified in the GWAS as a strong candidate gene for dwarfism. The exons and UTRs were resequenced to search for a possible causative variant. Regions of interest were amplified by polymerase chain reaction using Platinum Taq DNA Polymerase kits (Invitrogen) in the same case–control series as used for the genotyping experiments, according to the manufacturer’s instructions and using M13 sequence tagged target-specific primers (sequences available upon request) designed using Primer3 (Rozen & Skaletsky 2000). DNA sequencing was outsourced to Macrogen Inc and was performed using Applied Biosystems ABI3730XL capillary sequencers. Raw sequence data was processed using PhredPhrap, version 030415 (Ewing et al. 1998) and was screened for polymorphisms using Polyphred, version 6.15 (Nickerson et al. 1997). Traces were visualized and manually inspected using Consed, version 17.0 (Gordon et al. 1998). SNPs were mapped to EquCab 2.0 using blat (Kent 2002) and were classed as novel if they did not appear in the EquCab 2.0 SNP collection.

Statistics

We compared genotype frequencies in cases and controls, testing for disease association using chi-square tests with two degrees of freedom. Where appropriate, we modelled genetic effects by minor allele using one degree of freedom chi-square tests. SNPs in the fine-mapping analysis were analysed using the Cochran–Armitage test to identify trends. As the study was exploratory in nature, we report P-values uncorrected for multiple testing, as is conventional in GWAS. For reference, however, the Bonferroni corrected α-level (corresponding to a study-wide α-level of 5%) for the current study is 1.72 × 10−6. All statistical analyses were conducted using plink, glu and r (Purcell et al. 2007; R Development Core Team 2009). Plots of regional association were generated in r using a modified version of code available at http://www.broadinstitute.org/mpg/snap/index.php.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflicts of Interest
  9. Web URLs
  10. References

The GWAS results, sorted by chromosome, are shown in Fig. 1. We observed a peak of association on chromosome 14, the best SNP being BIEC2-239376 (P = 4.54 × 10−5). We further investigated the association at this locus by fitting specific genetic models. The significance of association at BIEC2-239376 was greatest under a recessive model (P = 7.74 × 10−6). All 10 of the dwarfs in the GWAS were TT homozygotes, while the controls were comprised of four CC homozygotes and six heterozygotes. BIEC2-239376 was the most significant SNP in a cluster of SNPs at the p-arm of chromosome 14. We assessed the independence between BIEC2-239376 and other chromosome 14 loci by calculating the correlation coefficient, r2, between the SNP with the strongest signal and all other SNPs on the chromosome (Fig. 2). We reanalysed each chromosome 14 locus using logistic regression, conditioned on BIEC2-239376, and concluded that linkage disequilibrium with BIEC2-239376 is the main driver of association at the other loci.

image

Figure 1.  Manhattan plot of P-value in the Friesian horse dwarfism GWAS. Association of 34 429 SNPs with dwarfism represented by −log10P-values from a two degree of freedom chi-square test plotted by chromosome and sorted by chromosomal position. No SNPs in the GWAS remained statistically significant after correction for multiple testing.

Download figure to PowerPoint

image

Figure 2.  Association of chromosome 14 SNPs with Friesian horse dwarfism. Association plot of chromosome 14 SNPs with dwarfism, with chromosomal position on the x-axis and −log10P-value on the y-axis. The most significant chromosome 14 SNP, BIEC2-239376, is indicated by a red diamond. Each square on the plot represents a single SNP; the colour of each square represents the strength of correlation between that SNP and BIEC2-239376, with dark red indicating an r2 of 1 and white representing an r2 of 0.

Download figure to PowerPoint

We genotyped 374 SNPs, of which 319 were not on the GWAS array, in the region of association on chromosome 14. Of these, 184 were either monomorphic or had low MAF in the study population. From the remaining 190 SNPs, BIEC2-250663 (P = 4.94 × 10−5) was more significantly associated with dwarfism than BIEC2-239376 (P = 6.12 × 10−5) (Fig. 3); dwarfs were fully homozygous for both markers, and the homozygous genotype observed in dwarfs was not present in controls. An additional SNP, BIEC2-249929 was in perfect linkage disequilibrium with BIEC2-239376.

image

Figure 3.  Fine-mapping of a 2- MB region of chromosome 14 (3.8–5.4 MB). Association plot of chromosome 14 fine-mapped SNPs with dwarfism, with chromosomal position on the x-axis and −log10P-value on the y-axis. The positions and P-values for association of the best GWAS SNP (BIEC-239376) and a marginally more significant fine-mapped SNP (BIEC2-250663) are indicated. Names and positions of genes retrieved from the Ensemble genes track of Equ-Cab 2.0 are indicated on the lower panel of the figure.

Download figure to PowerPoint

BIEC2-239376 is located 34 kb from the gene encoding the equine homolog of the prophet of PIT1, paired-like homeodomain transcription factor protein, PROP1 (Fig. 3). Inactivating mutations in PROP1 are known to cause dwarfism in humans via combined pituitary hormone deficiency. We therefore selected the PROP1 gene for targeted resequencing of each exon, intron–exon boundary and both UTRs in our sample set. We detected a total of nine SNPs (Table 1), all of which we believe to be novel based on their absence from the EquCab 2.0 SNP database. Each novel SNP was rare, and many were singletons; none were associated with the dwarfism phenotype. Thus, resequencing analysis of the protein-coding components of PROP1 failed to detect a putative functional variant.

Table 1.   Sequences of nine novel SNPs identified by resequencing PROP1.
SNP ID5′ flanking seqAlleles3′ flanking seqSNP Position (bp)LocationProtein
PROP1_1AAGGACACCGC/TCAACCACAAA37737075′ UTR 
PROP1_2CACAAAAAAAA/CCACATCCAAG37736925′ UTR 
PROP1_3AAAGAGGGGAC/TGCTGCCTCCT3773583Exon 1Arg/Cys
PROP1_4CTTGCAGAGCC/GAGAGAGCTGG3773447Intron 1 
PROP1_5CCTGTGTCCCG/TCGCAAGGCCC3725488Exon 2 
PROP1_6GGAGTGCGGGC/TGACCCCTCTG3725604Exon 2Arg/His
PROP1_7GTCCTCCTTGC/TGGGGAAAGCC3725624Exon 2 
PROP1_8CCCTTCCTCCT/CCAGCCCTCCA3724500Exon 4 
PROP1_9CTGTGGCCCTC/TGTGAAACACA37243063′ UTR 

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflicts of Interest
  9. Web URLs
  10. References

High-density SNP-based genotyping arrays have been successfully used to identify genetic variants contributing to both common and rare diseases in humans and animals (Charlier et al. 2008; Manolio et al. 2008). Here, we report the application of SNP genotyping arrays for mapping dwarfism in horses. Using a small panel of affected and unaffected Friesian horses, we have identified a region of association on chromosome 14 that is strongly associated with the dwarf phenotype.

GWAS exploit the underlying correlations between genetic variants that initially arise via mutation on a common genetic background that is gradually eroded by recombination events through successive generations. In humans, complex population structure means that correlation between loci may diminish relatively quickly and thus exist only over short distances. In many breeds of animals, however, selective breeding strategies and small effective population sizes mean that pairwise correlations between loci may be strong over large regions and on less complex backgrounds. Thus, a much smaller number of genetic markers should be required to effectively capture information on untyped variants. The drawback of long range linkage disequilibrium in the context of disease mapping is that localization of causal mutations is difficult, because the associated regions are often very large. In this study, the associated region on chromosome 14 exceeds two megabases in length, peaking with BIEC2-239376 at 3 761 355 bp. Fine-mapping using a denser panel of SNPs failed to resolve the situation; we identified two additional SNPs, BIEC2-249929 and BIEC2-250663, that were at least as statistically significant as BIEC2-239376 and which were located at 3 761 254 and 5 418 619 bp, respectively. No other genotyped SNP in the region was fully homozygous, as would be expected for a recessively inherited trait, in the dwarfs. This study therefore serves as an example of the difficulty in overcoming LD for fine-mapping in inbred populations.

Perturbations in homologous gene families often result in the manifestation of phenotypically similar traits between species. As such, we screened the Online Mendelian Inheritance in Man (OMIM) database (http://www.ncbi.nlm.nih.gov/omim/) to determine whether any of the genes in the chromosome 14 association region were implicated in dwarfism in humans. We identified the equine homolog of PROP1 as being a strong positional candidate gene for dwarfism in Friesians. Inactivating mutations in PROP1 cause disruption of the growth hormone (GH) axis and are manifested as a dwarfism phenotype. Central pituitary underproduction of GH leads typically to proportional dwarfism, as seen in the German Shepherd dog (Andresen & Willeberg 1976; Hanson et al. 2006). In the Friesian dwarf horses, however, a disproportional growth disturbance is seen, which would imply a local defect or disturbance in one of the regulatory systems for growth plate development. Growth plate development is under the control of many autocrine and paracrine factors (Kronenberg 2003). Recently, de Graaf and colleagues investigated the functioning of the hypothalamic-pituitary growth axis in three Friesian dwarfs. No evidence of hypothalamic-pituitary dysfunction or failure of IGF-1 production was found, suggesting that the cause of the congenital growth abnormality was located distal or peripheral to the level of the GH receptor in the liver and may have been a defect in a peripheral IGF-1 or GH receptor or may not involve the GH-IGF-1 axis at all (de Graaf-Roelfsema et al. 2009). Concomitant with this observation, we did not detect coding polymorphisms in PROP1 that were associated with dwarfism in our Friesian horse samples.

When the normal bone remodelling process is disturbed in horses, abnormal defects in the growth plate may result. A local defect or disturbance in one of the regulatory systems for growth plate development potentially can result in a disproportional growth disturbance as is typically seen in Friesian dwarf horses (Vaughan 1976; Jeffcott & Henson 1998; Gee et al. 2005). Further screening of the OMIM database for genes in the chromosome 14 region that could be linked to a disturbed bone remodelling process in abnormal growth plate development identified ZNF346, COL23A1 and B4GALT7 as candidate genes. ZNF346 is proposed to play a role in apoptosis (cell death), a process crucial for the normal transition of cartilage into bone seen during normal physeal growth (Gibson 1998; Ballock & O’Keefe 2003). It could be speculated that disturbed apoptosis plays a role in the physeal growth retardation typically seen in Friesian dwarfism. Both COL23A1 and B4GALT7 are proposed to play a role in collagen network formation. Although a disturbed collagen network is known to effect calcification and subsequent transformation of cartilage into bone (Wassen et al. 2000), the specific role of both genes in collagen formation is highly speculative and largely unclear. B4GALT7 is also proposed to play a role in connective tissue disorders and has been related to disturbed fibril organization and proteoglycan synthesis. Both processes could play a role in the abnormal development of bone and subsequent retardation of growth plate growth that are observed in Friesian dwarfism (Kvist et al. 2006; Burdan et al. 2009).

Finally, based on the equine build from UCSC Broad Institute (UCSC Genome Browser: http://genome.ucsc.edu), both FGFR1 and FGFR2 are included within the critical region of chromosome 14 flanked by BIEC2-249929 and BIEC2-250663. Both fibroblast growth factor receptors play key roles in skeletal development, and mutations have been related to skeletal dysplasia and dwarfing syndromes (Eswarakumar et al. 2002; White et al. 2005). Profound effects on bone elongation have been shown through supposed suppression of chondrocyte and osteoblast function. However, in contrast to the relative normal head proportions seen in the Friesian dwarf syndrome, FGFR1 and FGFR2 mutations seem to have a significant effect on flat bone growth and skull formation.

Using an agnostic genome-wide approach, we have identified a putative region that may harbour a gene for dwarfism in Friesian horses. Validation of this finding in a larger group of animals and segregation analysis in known pedigrees is warranted before further localization of the causative mutation is conducted. This may prove extremely challenging given the strong linkage disequilibrium observed in the region. This study suggests that, with the advent of new genomic tools, studies of equine diseases may yield important new insight into pathogenesis and may be translatable to orthologous human traits.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflicts of Interest
  9. Web URLs
  10. References

We would like to thank the equine practitioners for submitting the dwarf cases and the studbook (http://www.fps-studbook.com) for their technical assistance. This material is based upon works supported by the Science Foundation Ireland under the Research Frontiers Programme (Foundation Grant No. R10526).

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflicts of Interest
  9. Web URLs
  10. References
  • Andresen E. & Willeberg P. (1976) Pituitary dwarfism in German shepherd dogs: additional evidence of simple, autosomal recessive inheritance. Nordisk Veterinaer Medicin 28, 4816.
  • Back W., Van Der Lugt J.J., Nikkels P.G., Van Den Belt A.J., Van Der Kolk J.H. & Stout T.A. (2008) Phenotypic diagnosis of dwarfism in six Friesian horses. Equine Veterinary Journal 40, 2827.
  • Ballock R.T. & O’Keefe R.J. (2003) Physiology and pathophysiology of the growth plate. Birth Defects Research. Part C, Embryo Today: Reviews 69, 12343.
  • Burdan F., Szumilo J., Korobowicz A. et al. (2009) Morphology and physiology of the epiphyseal growth plate. Folia Histochemica et Cytobiologica 47, 516.
  • Charlier C., Coppieters W., Rollin F. et al. (2008) Highly effective SNP-based association mapping and management of recessive defects in livestock. Nature Genetics 40, 44954.
  • Eswarakumar V.P., Monsonego-Ornan E., Pines M., Antonopoulou I., Morriss-Kay G.M. & Lonai P. (2002) The IIIc alternative of Fgfr2 is a positive regulator of bone formation. Development 129, 378393.
  • Ewing B., Hillier L., Wendl M.C. & Green P. (1998) Base-calling of automated sequencer traces using phred I. Accuracy assessment. Genome Research 8, 17585.
  • Frazer K.A., Ballinger D.G., Cox D.R. et al. , Adebamowo C.A., Ajayi I., Aniagwu T., Marshall P.A., Nkwodimmah C., Royal C.D., Leppert M.F., Dixon M., Peiffer A., Qiu R., Kent A., Kato K., Niikawa N., Adewole I.F., Knoppers B.M., Foster M.W., Clayton E.W., Watkin J., Muzny D., Nazareth L., Sodergren E., Weinstock G.M., Yakub I., Birren B.W., Wilson R.K., Fulton L.L., Rogers J., Burton J., Carter N.P., Clee C.M., Griffiths M., Jones M.C., McLay K., Plumb R.W., Ross M.T., Sims S.K., Willey D.L., Chen Z., Han H., Kang L., Godbout M., Wallenburg J.C., L’Archeveque P., Bellemare G., Saeki K., An D., Fu H., Li Q., Wang Z., Wang R., Holden A.L., Brooks L.D., McEwen J.E., Guyer M.S., Wang V.O., Peterson J.L., Shi M., Spiegel J., Sung L.M., Zacharia L.F., Collins F.S., Kennedy K., Jamieson R. & Stewart J. (2007) A second generation human haplotype map of over 3.1 million SNPs. Nature 449, 85161.
  • Gee E.K., Firth E.C., Morel P.C., Fennessy P.F., Grace N.D. & Mogg T.D. (2005) Enlargements of the distal third metacarpus and metatarsus in Thoroughbred foals at pasture from birth to 160 days of age. New Zealand Veterinary Journal 53, 43847.
  • Gibson G. (1998) Active role of chondrocyte apoptosis in endochondral ossification. Microscopy Research and Technique 43, 191204.
  • Gordon D., Abajian C. & Green P. (1998) Consed: a graphical tool for sequence finishing. Genome Research 8, 195202.
  • De Graaf-Roelfsema E., Back W., Keizer H.A., Stout T.A.E. & Van Der Kolk J.H. (2009) Normal function of the hypothalamic-pituitary growth axis in three dwarf Friesian foals. The Veterinary Record 165, 3736.
  • Hanson J.M., Mol J.A., Leegwater P.A., Kooistra H.S. & Meij B.P. (2006) The leukemia inhibitory factor receptor gene is not involved in the etiology of pituitary dwarfism in German shepherd dogs. Research in Veterinary Science 81, 31620.
  • Howie B.N., Carlson C.S., Rieder M.J. & Nickerson D.A. (2006) Efficient selection of tagging single-nucleotide polymorphisms in multiple populations. Human Genetics 120, 5868.
  • Jeffcott L.B. & Henson F.M. (1998) Studies on growth cartilage in the horse and their application to aetiopathogenesis of dyschondroplasia (osteochondrosis). Veterinary Journal 156, 17792.
  • Karlsson E.K., Baranowska I., Wade C.M. et al. (2007) Efficient mapping of mendelian traits in dogs through genome-wide association. Nature Genetics 39, 13218.
  • Kent W.J. (2002) BLAT – the BLAST-like alignment tool. Genome Research 12, 65664.
  • Kronenberg H.M. (2003) Developmental regulation of the growth plate. Nature 423, 3326.
  • Kvist A.J., Johnson A.E., Morgelin M. et al. (2006) Chondroitin sulfate perlecan enhances collagen fibril formation Implications for perlecan chondrodysplasias. Journal of Biological Chemistry 281, 3312739.
  • Manolio T.A., Brooks L.D. & Collins F.S. (2008) A HapMap harvest of insights into the genetics of common disease. Journal of Clinical Investigation 118, 1590605.
  • Nickerson D.A., Tobe V.O. & Taylor S.L. (1997) PolyPhred: automating the detection and genotyping of single nucleotide substitutions using fluorescence-based resequencing. Nucleic Acids Research 25, 274551.
  • Orr N. & Chanock S. (2008) Common genetic variation and human disease. Advances in Genetics 62, 132.
  • Osinga A. (2000) Het Fokken van het Friese Paard. Schaafsma & Brouwer, Dokkum.
  • Purcell S., Neale B., Todd-Brown K. et al. (2007) PLINK: a tool set for whole-genome association and population-based linkage analyses. American Journal of Human Genetics 81, 55975.
  • R Development Core Team (2009) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna.
  • Rozen S. & Skaletsky H. (2000) Primer3 on the WWW for general users and for biologist programmers. Methods in Molecular Biology 132, 36586.
  • Steemers F.J. & Gunderson K.L. (2007) Whole genome genotyping technologies on the BeadArray platform. Biotechnology Journal 2, 419.
  • Vaughan L.C. (1976) Growth plate defects in foals. The Veterinary Record 98, 1658.
  • Wade C.M., Giulotto E., Sigurdsson S. et al. (2009) Genome sequence, comparative analysis, and population genetics of the domestic horse. Science 326, 8657.
  • Wassen M.H., Lammens J., Tekoppele J.M., Sakkers R.J., Liu Z., Verbout A.J. & Bank R.A. (2000) Collagen structure regulates fibril mineralization in osteogenesis as revealed by cross-link patterns in calcifying callus. Journal of Bone and Mineral Research 15, 177685.
  • White K.E., Cabral J.M., Davis S.I. et al. (2005) Mutations that cause osteoglophonic dysplasia define novel roles for FGFR1 in bone elongation. American Journal of Human Genetics 76, 3617.