METHODS OF USING GENETIC MARKERS AND RELATED EPISTATIC

INTERACTIONS

PRIORITY CLAIM

[0001]This application claims the benefit of U.S. Provisional Application Serial No. 60/971,750 filed September 12, 2007, which is herein incorporated by reference in its entirety.

INCORPORATION OF SEQUENCE LISTING

[0002] A sequence listing containing the file named pa_CandGeneInteractionEffects2_annotated.ST25.txt, which is 84,218 bytes (as measured in Microsoft Windows®) was created on 09/05/08, comprises 175 nucleotide sequences, is submitted herewith, and is herein incorporated by reference in its entirety.

FIELD OF THE INVENTION

[0003] The present invention relates to the enhancement of desirable characteristics in dairy cattle. More specifically, it relates to the use of genes and genetic markers in methods for improving dairy cattle with respect to fitness and/or productivity traits using genetic markers, including simultaneous application of multiple genetic markers and interactions between specific alleles at those markers.

BACKGROUND OF THE INVENTION

[0004] The future viability and competitiveness of the dairy industry depends on continual improvement in milk productivity (e.g. milk, fat, protein yield, fat%, protein % and persistency of lactation), health (e.g. Somatic Cell Count, mastitis incidence), fertility (e.g. pregnancy rate, display of estrus, calving interval and non-return rates in bulls), calving ease (e.g. direct and maternal calving ease), longevity (e.g. productive life), and functional conformation (e.g. udder support, proper foot and leg shape, proper rump angle, etc.). Unfortunately efficiency traits are often unfavorably correlated with fitness traits. Although fitness traits all have some degree of underlying genetic variation in commercial cattle populations, the accuracy of selecting breeding animals with superior genetic merit for many of them is low due to

low heritability or the inability to measure the trait cost effectively on the candidate animal. In addition, many productivity and fitness traits can only be measured on females. Thus, the accuracy of conventional selection for these traits is moderate to low and ability to make genetic change through selection is limited, particularly for fitness traits.

In addition, there are frequently interactions between specific alleles at multiple loci which confound prediction of genetic merit. In other words, the effects of combinations of alleles on traits may not be strictly additive, but rather synergistic (or antagonistic). In the absence of an understanding of these interactions, a priori estimation of genetic merit is obviously more difficult and less accurate.

[0005] Genomics offers the potential for greater improvement in productivity and fitness traits through the discovery of genes, or genetic markers linked to genes, that account for genetic variation and can be used for more direct and accurate selection. Close to 1000 markers with associations with productivity and fitness traits have been reported (see www.bovineqtl.tamu.edu/ for a searchable database of reported QTL), however, the resolution of QTL location is still quite low which makes it difficult to utilize these QTL in marker-assisted selection (MAS) on an industry scale. Only a few QTL have been fully characterized with a strong putative or well-confirmed causal mutation: DGATl on chromosome 14 (Grisard et al., 2002; Winter et al, 2002; Kuhn et al., 2004) GHR on chromosome 20 (Blott et al., 2003), ABCG2 (Cohen- Zinder et al., 2005) or SPPl on chromosome 6 (Schnabel et al., 2005). However, these discoveries are rare and only explain a small portion of the genetic variance for productivity traits and no genes controlling fitness traits have been fully characterized. A more successful strategy employs the use of whole-genome high-density scans of the entire bovine genome in which QTL are mapped with sufficient resolution to explain the majority of genetic variation around productivity and fitness traits.

[0006] Cattle herds used for milk production around the world originate predominantly from the Holstein or Holstein-Friesian breeds which are known for high levels of production. However, the high production levels in Holsteins have also been linked to greater calving difficulty and reduced levels of fertility. It is unclear whether these unfavorable correlations are due to pleiotropic gene effects or simply

due to linked genes. If the latter is true, with marker knowledge, it may be possible to select for favorable recombinants that contain the favorable alleles from several linked genes that are normally at frequencies too low to allow much progress with traditional selection. Since Holstein germplasm has been sold and transported globally for several decades, the Holstein breed has effectively become one large global population held to relatively moderate inbreeding rates. Also, the outbred nature of such a large population selected for several generations has allowed linkage disequilibrium to break down except within relatively short distances (i.e. less than a few centimorgans) (Hayes et al., 2006). In addition, as a result of selection in several countries with different breeding goals, linkage disequilibrium between relatively close loci can be quite variable due to the effects of drift within sub-populations that have become mixed via several generations of global selection and breeding. Given this pattern of linkage disequilibrium, very dense marker coverage is required to refine QTL locations with sufficient precision to find markers that are in very tight linkage disequilibrium with them. Therefore, markers that are in very tight linkage disequilibrium with the QTL are essential for effective population-wide MAS or whole-genome selection (WGS).

[0007] Most productivity and fitness traits are quantitative in nature and hence are governed by hundreds of QTL of small to moderately sized effects. Therefore, to characterize enough QTL to explain a majority of variation for these traits, the whole genome must be scanned with a set of markers mapped to the genome at high resolution (i.e. greater than 1 marker/cM); otherwise known as whole-genome analysis.

[0008] Furthermore, a sufficient number of QTL must be used in MAS in order to accurately predict the breeding value of an animal without phenotyping records on relatives or the animal itself. The application of such a high-density whole-genome marker map to discover and finely-map QTL explaining variation in productivity and fitness traits is described herein. The large number of resulting linked markers can be used in several methods of marker selection or marker-assisted selection, including whole-genome selection (WGS) (Meuwissen et al., Genetics 2001) to improve the genetic merit of the population for these traits and create value in the dairy industry. [0009] Unlike some simple traits which may be fully explained by one causal mutation, many productivity and fitness traits require a large number of markers to

accurately predict the phenotypic performance of the animal. Quantitative phenotypes generally involve multiple genes, multiple pathways, and complex interactions. In some cases, this complexity results in interactions between markers which is exceptionally difficult to predict.

[0010] Few studies have investigated the contribution of the interactions between candidate gene SNPs to quantitative variation in dairy traits. There are several possible reasons for lack of such interaction studies. First, different candidate genes were generally investigated by different groups, and genotypes of different candidate genes were collected on different animals; focuses of most candidate gene studies were to discover/confirm an association between a trait and SNP(s) of their interest; investigation of interaction effects generally needs larger sample sizes. [0011] However, expression of quantitative traits are results of interactions of multiple physiological pathways (for example lipid metabolism, appetite/satiety, etc.), and a large number of genes are generally involved in each physiological pathway. Therefore, it appears to be reasonable to expect some degree of interaction among genes that are involved in the same or different pathways, and to expect a proportion of genetic quantitative variation are due to such interactions.

[0012] The inventors have identified markers associated with novel traits in important genes in dairy cows, as well as numerous interaction effects including epistatic effects between these genes, which can be used to substantially improve the accuracy of genetic evaluations, prediction, and selection.

SUMMARY OF THE INVENTION

[0013] This section provides a non-exhaustive summary of the present invention.

[0014] Various embodiments of the invention provide methods for evaluating an animal's genotype at 1 or more positions in the animal's genome. In various aspects of these embodiments the animal's genotype is evaluated at positions within a segment of DNA (an allele) that contains at least two SNPs selected from the SNPs described in the Tables and Sequence Listing. For each of the SNPs listed in tables 1 and 3, details regarding SNP location, SNP length, and alleles can be found in Table

[0015] Other embodiments of the invention provide methods for allocating animals for use according to their predicted marker breeding value for productivity and/or fitness traits. Various aspects of this embodiment of the invention provide methods that comprise: a) analyzing the animal's genomic sequence at two or more polymorphisms (where the alleles analyzed each comprise at least two SNP) to determine the animal's genotype at each of those polymorphisms; b) analyzing the genotype determined for each polymorphisms to determine which allele of the SNP is present; c) allocating the animal for use based on its genotype at two or more of the polymorphisms analyzed. Various aspects of this embodiment of the invention provide methods for allocating animals for use based on a favorable association between the animal's genotype, at two or more polymorphisms disclosed in the present application, and a desired phenotype. Alternatively, the methods provide for not allocating an animal for a certain use because it has two or more SNP alleles that are either associated with undesirable phenotypes or are not associated with desirable phenotypes.

[0016] Other embodiments of the invention provide methods for selecting animals for use in breeding to produce progeny. Various aspects of these methods comprise: A) determining the genotype of at least two potential parent animals at two or more locus/loci, where at least two of the loci analyzed contains an allele of a SNP selected from the group of SNPs described in Tables 1 and 3. B) Analyzing the determined genotype at two or more positions for at least two animals to determine which of the SNP alleles is present. C) Correlating the analyzed allele(s) with two or more phenotypes. D) Allocating at least two animals for use to produce progeny. Alternative embodiments include analyzing the animal's genotype at two or more loci wherein the analysis comprises evaluating interaction effects.

[0017] Other embodiments of the invention provide methods for producing offspring animals (progeny animals). Aspects of this embodiment of the invention provide methods that comprise: breeding an animal that has been selected for breeding by methods described herein to produce offspring. The offspring may be produced by purely natural methods or through the use of any appropriate technical means, including but not limited to: artificial insemination; embryo transfer (ET), multiple ovulation embryo transfer (MOET), in vitro fertilization (IVF), or any combination thereof.

[0018] Other embodiments of the invention provide for methods of selecting animals for use in breeding to produce progeny wherein interaction effects between multiple markers are applied in the analysis.

DEFINITIONS

[0019] The following definitions are provided to aid those skilled in the art to more readily understand and appreciate the full scope of the present invention. Nevertheless, as indicated in the definitions provided below, the definitions provided are not intended to be exclusive, unless so indicated. Rather, they are preferred definitions, provided to focus the skilled artisan on various illustrative embodiments of the invention.

[0020] As used herein the term "allelic association" preferably means: nonrandom deviation of f(A;B _j ) from the product of f(A;) and f(B _j ), which is specifically defined by r ² >0.2, where r ² is measured from a reasonably large animal sample (e.g., >100) and defined as

r [Equation

1] where Ai represents an allele at one locus, Bi represents an allele at another locus; f(A]Bi) denotes frequency of having both Ai and Bi, f(Ai) is the frequency of Ai, f(Bi) is the frequency of Bi in a population.

[0021] As used herein the terms "allocating animals for use" and "allocation for use" preferably mean deciding how an animal will be used within a herd or that it will be removed from the herd to achieve desired herd management goals. For example, an animal might be allocated for use as a breeding animal or allocated for sale as a non- breeding animal (e.g. allocated to animals intended to be sold for meat). In certain aspects of the invention, animals may be allocated for use in sub-groups within the breeding programs that have very specific goals (e.g. productivity or fitness). Accordingly, even within the group of animals allocated for breeding purposes, there may be more specific allocation for use to achieve more specific and/or specialized breeding goals.

[0022] As used herein the terms "animal" or "animals" preferably refer to dairy cattle.

[0023] As used herein "fitness" preferably refers to traits that include, but are not limited to: pregnancy rate (PR), daughter pregnancy rate (DPR), productive life (PL),somatic cell count (SCC) and somatic cell score (SCS). PR and DPR refer to the percentage of non-pregnant animals that become pregnant during each 21 -day period. PL is calculated as months in milk in each lactation, summed across all lactations until removal of the cow from the herd (by culling or death). SCS = log ₂ (SCC/100,000)+3, where SCC is somatic cells per milliliter of milk.

[0024] As used herein the term "growth" refers to the measurement of various parameters associated with an increase in an animal's size/weight.

[0025] As used herein the term "linkage disequilibrium" preferably means allelic association wherein Ai and Bi (as used in the above definition of allelic association) are present on the same chromosome.

[0026] As used herein the term "marker-assisted selection (MAS) preferably refers to the selection of animals on the basis of marker information in possible combination with pedigree and phenotypic data.

[0027] As used herein the terms "marker breeding value (MBV)" and "predicted marker breeding value (PMBV)" refer to an estimate of an animal's genetic transmitting ability with respect to specific traits and is based on its genotype.

[0028] As used herein the term "natural breeding" preferably refers to mating animals without human intervention in the fertilization process. That is, without the use of mechanical or technical methods such as artificial insemination or embryo transfer. The term does not refer to selection of the parent animals.

[0029] As used herein the term "net merit" preferably refers to a composite index that includes several commonly measured traits weighted according to relative economic value in a typical production setting and expressed as lifetime economic worth per cow relative to an industry base. Examples of a net merit indexes include, but are not limited to, $NM or TPI in the USA, LPI in Canada, etc (formulae for calculating these indices are well known in the art (e.g. $NM can be found on the USDA/AIPL website: www.aipl.arsusda.gov/reference.htm).

[0030] As used herein the term "predicted value" preferably refers to an estimate of an animal's breeding value or transmitting ability based on its genotype and pedigree.

[0031 ] As used herein "productivity" and "production" preferably refers to yield traits that include, but are not limited to: total milk yield, milk fat percentage, milk fat yield, milk protein percentage, milk protein yield, total lifetime production, milking speed and lactation persistency.

[0032] As used herein the term "quantitative trait" is used to denote a trait that is controlled by multiple (two or more, and often many) genes each of which contributes small to moderate effect on the trait. The observations on quantitative traits often follow a normal distribution.

[0033] As used herein the term "quantitative trait locus (QTL)" is used to describe a locus that contains polymorphism that has an effect on a quantitative trait.

[0034] As used herein the term "reproductive material" includes, but is not limited to semen, spermatozoa, ova, and zygote(s).

[0035] As used herein the term "single nucleotide polymorphism" or "SNP" refer to a location in an animal's genome that is polymorphic within the population. That is, within the population some individual animals have one type of base at that position, while others have a different base. For example, a SNP might refer to a location in the genome where some animals have a "G" in their DNA sequence, while others have a "T".

[0036] As used herein the terms "hybridization under stringent conditions" and "stringent hybridization conditions" preferably mean conditions under which a "probe" will hybridize to its target sequence to a detectably greater degree than to other sequences (e.g., at least 5-fold over background). Stringent conditions are target-sequence-dependent and will differ depending on the structure of the polynucleotide. By controlling the stringency of the hybridization and/or washing conditions, target sequences that are 100% complementary to the probe can be identified (homologous probing). Alternatively, stringency conditions can be adjusted to allow some mismatching in sequences so that lower degrees of similarity are detected (heterologous probing).

[0037] Typically, stringent conditions will be those in which the salt concentration is less than about 1.5 M Na ion, typically about 0.01 to 1.0 M Na ion concentration (or

other salts) at pH 7.0 to 8.3 and the temperature is at least about 30 ⁰ C for short probes (e.g., 10 to 50 nucleotides) and at least about 60 ⁰ C for long probes (e.g., greater than 50 nucleotides). Stringency may also be adjusted with the addition of destabilizing agents such as formamide. Exemplary low stringency conditions include hybridization with a buffer solution of 30 to 35% formamide, 1 M NaCl, 1% SDS (sodium dodecyl sulphate) at 37 ⁰ C, and a wash in IX to 2X SSC (2OX SSC=3.0 M NaCl/0.3 M trisodium citrate) at 50 to 55 ⁰ C. Exemplary moderate stringency conditions include hybridization in 40 to 45% formamide, 1 M NaCl, 1% SDS at 37 ⁰ C, and a wash in 0.5X to IX SSC at 55 to 60 ⁰ C. Exemplary high stringency conditions include hybridization in 50% formamide, 1 M NaCl, 1% SDS at 37°C, and a wash in 0.1X SSC at 60 to 65 ⁰ C. The duration of hybridization is generally less than about 24 hours, usually about 4 to about 12 hours.

[0038] Specificity is typically the function of post-hybridization washes, the critical factors being the ionic strength and temperature of the final wash solution. For DNA- DNA hybrids, the thermal melting point (T _m ) can be approximated from the equation of Meinkoth and Wahl (1984) Anal. Biochem. 138:267-284: T _m =81.5°C+16.6 (log M)+0.41 (% GQ-0.61 (% form)-500/L; where M is the molarity of monovalent cations, % GC is the percentage of guanine and cytosine nucleotides in the DNA, % form is the percentage of formamide in the hybridization solution, and L is the length of the hybrid in base pairs. The T _m is the temperature (under defined ionic strength and pH) at which 50% of a complementary target sequence hybridizes to a perfectly matched probe. T _m is reduced by about 1°C for each 1% of mismatching; thus, T _1n , hybridization, and/or wash conditions can be adjusted to hybridize to sequences of the desired identity. For example, if sequences with 90% identity are sought, the T _m can be decreased 10 ⁰ C. Generally, stringent conditions are selected to be about 5°C lower than the T _m for the specific sequence and its complement at a defined ionic strength and pH.

[0039]However, highly stringent conditions can utilize a hybridization and/or wash at 1 , 2, 3, or 4 ⁰ C lower than the thermal melting point (T _1n ); moderately stringent conditions can utilize a hybridization and/or wash at 6, 7, 8, 9, or 10 ⁰ C lower than the thermal melting point (T _1n ); low stringency conditions can utilize a hybridization and/or wash at 11, 12, 13, 14, 15, or 20 ⁰ C lower than the thermal melting point (T _1n ). Using the equation, hybridization and wash compositions, and desired T _1n , those of

ordinary skill will understand that variations in the stringency of hybridization and/or wash solutions are inherently described. If the desired degree of mismatching results in a T _n , of less than 45°C (aqueous solution) or 32°C (formamide solution), it is preferred to increase the SSC concentration so that a higher temperature can be used. An extensive guide to the hybridization of nucleic acids is found in Tijssen (1993) Laboratory Techniques in Biochemistry and Molecular Biology—Hybridization with Nucleic Acid Probes, Part I, Chapter 2 (Elsevier, N.Y.); and Ausubel et al., eds. (1995) Current Protocols in Molecular Biology, Chapter 2 (Greene Publishing and Wiley- Interscience, New York). See also Sambrook et al. (1989) Molecular Cloning: A Laboratory Manual (2d ed., Cold Spring Harbor Laboratory Press, Plainview, N.Y.).

[0040] As used herein the terms "marker breeding value (MBV)" and "predicted marker breeding value (PMBV)" respectively refer to an estimate of an animal's genetic transmitting ability with respect to either productivity traits or fitness traits that is based on its genotype.

[0041] As used herein the term "whole-genome analysis" preferably refers to the process of QTL mapping of the entire genome at high marker density (i.e. approximately one marker per cM) and detection of markers that are in population- wide linkage disequilibrium with QTL.

[0042] As used herein the term "whole-genome selection (WGS)" preferably refers to the process of marker-assisted selection (MAS) on a genome-wide basis in which markers spanning the entire genome at moderate to high density (e.g. approximately one marker per 1 -5 cM), or at moderate to high density in QTL regions, or directly neighboring or flanking QTL that explain a significant portion of the genetic variation controlling two or more traits.

[0043] As used herein, the term "interaction effect" preferably refers to an alteration of the predicted phenotypic effect of a first marker, depending on the allelic state of a second marker. For example, if SNPl has an effect estimate of 10 for a positive allelic association when SNP2 is an A, but SNPl has an effect estimate of 5 for a positive allelic association when SPP2 is a T, the change in effect estimate from 10 to 5 would be considered an interaction effect. Marker-based interaction effects must involve at least two markers.

[0044]As used herein, the term "epistatic interaction" preferably refers to interactions between alleles of genes, for example when the action of one gene is modified by one or several genes that assort independently (but may be linked).

ILLUSTRATIVE EMBODIMENTS OF THE INVENTION

[0045] Various embodiments of the present invention provide methods for evaluating an animal's (especially a dairy animal's) genotype at 1 or more positions in the animal's genome. Aspects of these embodiments of the invention provide methods that comprise determining the animal's genomic sequence at 1 or more locations (loci) that contain single nucleotide polymorphisms (SNPs). Specifically, the invention provides methods for evaluating an animal's genotype by determining which of two or more alleles for the SNP are present for each of 1 or more SNPs selected from the group consisting of the SNPs described in Tables 1 and 3 of the instant application.

[0046] In preferred aspects of these embodiments the animal's genotype is evaluated to determine which allele is present for 10 or more SNPs selected from the group of SNPs described in Tables 1 and 3. More, preferably the animal's genotype is determined for positions corresponding with 2, 10, 100, 200, 500, or 1000, or more of SNPs, at least two of which are described in Tables 1 and 3. In some embodiments of this invention, interactions between two SNPs are used in analysis of the animal's genotype.

[0047] In other aspects of this embodiment, the animal's genotype is analyzed with respect to at least 1 or more SNPs that have been shown to be associated with productivity and/or fitness (see Table 1 for a list of the SNPs associated with these traits). Further, embodiments of the invention provides a method for genotyping 2 or more, 10 or more, 10 or more, 50 or more, 100 or more, 200 or more, or 500 or more, or 1000 or more SNPs, at least one of which has been determined to be significantly associated with a productivity or fitness trait as described in Table 1.

[0048] Aspects of the present invention also provides for both whole-genome analysis and whole genome-selection (WGS) (that is marker-assisted selection (MAS) on a genome-wide basis). Various aspects of this embodiment of the invention provide for either whole-genome analysis or WGS wherein the makers analyzed for an animal span the animal's entire genome at moderate to high density. That is, the animal's

genome is analyzed with markers that on average occur, at least, approximately every 1 to 5 centimorgans in the genome. Moreover the invention provides that of the markers used to carry out the whole-genome analysis or WGS, including 2 or more, 10 or more, 10 or more, 50 or more, 100 or more, 200 or more, 500 or more, or 1000 or more markers, at least one of which are selected from the markers described in Tables 1 and 3. hi preferred aspects of this embodiment the markers may be associated with fitness or productivity traits, or may be associated with both fitness and productivity traits.

[0049] In any embodiment of the invention the genomic sequence at the SNP locus may be determined by any means compatible with the present invention. Suitable means are well known to those skilled in the art and include, but are not limited to direct sequencing, sequencing by synthesis, primer extension, Matrix Assisted Laser Desorption /Ionization-Time Of Flight (MALDI-TOF) mass spectrometry, polymerase chain reaction-restriction fragment length polymorphism, microarray/multiplex array systems (e.g. those available from Affymetrix, Santa Clara, California), and allele-specific hybridization.

[0050] Other embodiments of the invention provide methods for allocating animals for subsequent use (e.g. to be used as sires or dams or to be sold for meat or dairy purposes) according to their predicted value for productivity or fitness. Various aspects of this embodiment of the invention comprise determining at least two animal's genotype for at least two SNPs selected from the group of SNPs described in Tables 1 and 3 (methods for determining animals' genotypes for two or more SNPs are described supra). Thus, the animal's allocation for use may be determined based on its genotype at one or more, 2 or more, 10 or more, 10 or more, 50 or more, 100 or more, 300 or more, or 500 or more, or 1000 or more SNPs. The animal's allocation may further include an analysis of interaction effects between at least two SNPs.

[0051] The instant invention provides embodiments where analysis of the genotypes of the SNPs described in Tables 1 and 3 is the only analysis done. Other embodiments provide methods where analysis of the SNPs disclosed herein is combined with any other desired type of genomic or phenotypic analysis (e.g. analysis of any genetic markers beyond those disclosed in the instant invention). Moreover, the SNPs analyzed may be selected from those SNPs only associated productivity, only associated with fitness, or the analysis may be done for SNPs selected from any

desired combination of fitness and productivity. SNPs associated with various traits are listed in Table 1.

[0052] According to various aspects of these embodiments of the invention, once the animal's genetic sequence for the selected SNP(s) have been determined, this information is evaluated to determine which allele of the SNP is present for at least two of the selected SNPs. Preferably the animal's allelic complement for all of the determined SNPs is evaluated. Finally, the animal is allocated for use based on its genotype for two or more of the SNP positions evaluated. Preferably, the allocation is made taking into account the animal's genotype at each of the SNPs evaluated, but its allocation may be based on any subset or subsets of the SNPs evaluated.

[0053] According to various aspects of embodiments of the invention, once the animal's genetic sequence for the selected SNP(s) have been determined, this information is evaluated to determine which allele of the SNP is present for at least two of the selected SNPs. Preferably the animal's allelic complement for all of the determined SNPs is evaluated. An analysis of the allelic orientations of the SNPs is performed, and preferably, the result of the analysis includes information related to at least one interaction effect. Finally, the animal is allocated for use based on its genotype for two or more of the SNP positions evaluated. Preferably, the allocation is made taking into account the animal's genotype at each of the SNPs evaluated, but its allocation may be based on any subset or subsets of the SNPs evaluated.

[0054] The allocation may be made based on any suitable criteria. For any SNP, a determination may be made as to whether one of the allele(s) is associated/correlated with desirable characteristics or associated with undesirable characteristics. Furthermore, this determination may preferably include information related to interaction effects between multiple makers. This determination will often depend on breeding or herd management goals. Determination of which alleles are associated with desirable phenotypic characteristics can be made by any suitable means. Methods for determining these associations are well known in the art; moreover, aspects of the use of these methods are generally described in the EXAMPLES, below.

[0055] Phenotypic traits that may be associated with the SNPs of the current invention include, but are not limited to; fitness traits and productivity traits. Fitness traits include but are not limited to: pregnancy rate (PR), daughter pregnancy rate (DPR), productive life (PL), somatic cell count (SCC) and somatic cell score (SCS). Productivity traits include but are not limited to: total milk yield, milk fat percentage, milk fat yield, milk protein percentage, milk protein yield, total lifetime production, milking speed and lactation persistency

[0056] According to various aspects of this embodiment of the invention allocation for use of the animal may entail either positive selection for the animals having the desired genotype(s) (e.g. the animals with the desired genotypes are selected for productivity traits), negative selection of animals having undesirable genotypes (e.g. animals with an undesirable genotypes are culled from the herd), or any combination of these methods. According to preferred aspects of this embodiment of the invention animals identified as having SNP alleles associated with desirable phenotypes are allocated for use consistent with that phenotype (e.g. allocated for breeding based on phenotypes positively associated with fitness). Alternatively, animals that do not have SNP genotypes that are positively correlated with the desired phenotype (or possess SNP alleles that are negatively correlated with that phenotype) are not allocated for the same use as those with a positive correlation for the trait.

[0057] Other embodiments of the invention provide methods for selecting potential parent animals (i.e., allocation for breeding) to improve fitness and/or productivity in potential offspring. Various aspects of this embodiment of the invention comprise determining at least two animal's genotype for at least two SNPs selected from the group of SNPs described in Tables 1 and 3. Furthermore, determination of whether and how an animal will be used as a potential parent animal may be based on its genotype at two or more, 2 or more, 10 or more, 50 or more, 100 or more, 300 or more, or 500 or more, including at least one of the SNPs described in Tables 1 and 3. Moreover, as with other types of allocation for use, various aspects of these embodiments of the invention provide methods where the only analysis done is to genotype the animal for two or more of the SNPs described in Tables 1 and 3. Other aspects of these embodiments provide methods where analysis of two or more SNPs disclosed herein is combined with any other desired genomic or phenotypic analysis (e.g. analysis of any genetic markers beyond those disclosed in the instant invention).

Moreover, the SNP(s) analyzed may all be selected from those associated only with fitness traits or only with productivity traits. Conversely, the analysis may be done for SNPs selected from any desired combination of these or other traits.

[0058] According to various aspects of these embodiments of the invention, once the animal's genetic sequence at the site of the selected SNP(s) have been determined, this information is evaluated to determine which allele of the SNP is present for at least two of the selected SNPs. Preferably the animal's allelic complement for all of the sequenced SNPs is evaluated. Additionally, the animal's allelic complement is analyzed and correlated with the probability that the animal's progeny will express two or more phenotypic traits. Finally, the animal is allocated for breeding use based on its genotype for two or more of the SNP positions evaluated and the probability that it will pass the desired genotype(s)/allele(s) to its progeny. Preferably, the breeding allocation is made taking into account the animal's genotype at each of the SNPs evaluated. However, its breeding allocation may be based on any subset or subsets of the SNPs evaluated.

[0059] The breeding allocation may be made based on any suitable criteria. For example, breeding allocation may be made so as to increase the probability of enhancing a single certain desirable characteristic in a population, in preference to other characteristics, (e.g. increased fitness, or even specifically lowering somatic cell score (SCS) as part of fitness); alternatively, the selection may be made so as to generally maximize overall production based on a combination of traits. The allocations chosen are dependent on the breeding goals. Sub-categories falling within fitness, include, inter alia: daughter pregnancy rate (DPR), productive life (PL), and somatic cell score. Sub-categories falling within productivity include, inter alia: milk fat percentage, milk fat yield, total milk yield, milk protein percentage, and total milk protein.

[0060] Other embodiments of the instant invention provide methods for producing progeny animals. According to various aspects of this embodiment of the invention, the animals used to produce the progeny are those that have been allocated for breeding according to any of the embodiments of the current invention. Those using the animals to produce progeny may perform the necessary analysis or, alternatively, those producing the progeny may obtain animals that have been analyzed by another. The progeny may be produced by any appropriate means, including, but not limited to

using: (i) natural breeding, (ii) artificial insemination, (iii) in vitro fertilization (IVF) or (iv) collecting semen/spermatozoa and/or at least two ovum from the animal and contacting it, respectively with ova/ovum or semen/spermatozoa from a second animal to produce a conceptus by any means.

[0061] According to preferred aspects of this embodiment of the invention the progeny are produced by a process comprising natural breeding. In other aspects of this embodiment the progeny are produced through a process comprising the use of standard artificial insemination (AI), in vitro fertilization, multiple ovulation embryo transfer (MOET), or any combination thereof.

[0062] Other embodiments of the invention provide for methods that comprise allocating an animal for breeding purposes and collecting/isolating genetic material from that animal: wherein genetic material includes but is not limited to: semen, spermatozoa, ovum, zygotes, blood, tissue, serum, DNA, and RNA.

[0063] It is understood that most efficient and effective use of the methods and information provided by the instant invention employ computer programs and/or electronically accessible databases that comprise all or a portion of the sequences disclosed in the instant application. Accordingly, the various embodiments of the instant invention provide for databases comprising all or a portion of the sequences corresponding to at least 2 SNPs described in Tables 1 and 3. hi preferred aspect of these embodiments the databases comprise sequences for 1 or more, 5 or more, 10 or more, 20 or more, 50 or more, or substantially all of the SNPs described in Tables 1 and 3.

[0064] It is further understood that efficient analysis and use of the methods and information provided by the instant invention will employ the use of automated genotyping; particularly when large numbers (e.g. 100s) of markers are evaluated. Any suitable method known in the art may be used to perform such genotyping, including, but not limited to the use of micro-arrays.

[0065] Other embodiments of the invention provide methods wherein two or more of the SNP sequence databases described herein are accessed by two or more computer- executable programs. Such methods include, but are not limited to, use of the databases by programs to analyze for an association between the SNP and a phenotypic trait, or other user-defined trait (e.g. traits measured using two or more

metrics such as gene expression levels, protein expression levels, or chemical profiles), and programs used to allocate animals for breeding or market.

[0066] Other embodiments of the invention provide methods comprising collecting genetic material from an animal that has been allocated for breeding. Wherein the animal has been allocated for breeding by any of the methods disclosed as part of the instant invention.

[0067] Other embodiments of the invention provide for diagnostic kits or other diagnostic devices for determining which allele of a SNP is present in a sample; wherein the SNP(s) are selected from the group of SNPs described in Tables 1 and 3. In various aspects of this embodiment of the invention, the kit or device provides reagents/instruments to facilitate a determination as to whether nucleic acid corresponding to the SNP is present. Such kit/or device may further facilitate a determination as to which allele of the SNP is present. In certain aspects of this embodiment of the invention the kit or device comprises at least two nucleic acid oligonucleotide suitable for DNA amplification (e.g. through polymerase chain reaction). In other aspects of the invention the kit or device comprises a purified nucleic acid fragment capable of specifically hybridizing, under stringent conditions, with at least two allele of at least two SNPs described in Tables 1 and 3.

[0068] In particularly preferred aspects of this embodiment of the invention the kit or device comprises at least two nucleic acid array (e.g. DNA micro-arrays) capable of determining which allele of two or more of the SNPs described in Tables 1 and 3 is present in a sample. Preferred aspects of this embodiment of the invention provide DNA micro-arrays capable of simultaneously determining which allele is present in a sample for 2 or more SNPs. Preferably, the DNA micro-array is capable of determining which SNP allele is present in a sample for 10 or more, 50 or more, 100 or more, 200 or more, 500 or more, or 1000 or more SNPs. Methods for making such arrays are known to those skilled in the art and such arrays are commercially available (e.g. from Affymetrix, Santa Clara, California).

[0069] Genetic markers for fitness and/or productivity that are in allelic association with any of the SNPs described in the Tables may be identified by any suitable means known to those skilled in the art. For example, a genomic library may be screened using a probe specific for any of the sequences of the SNPs described in the Tables.

In this way clones comprising at least a portion of that sequence can be identified and then up to 300 kilobases of 3' and/or 5' flanking chromosomal sequence can be determined. By this means, genetic markers in allelic association with the SNPs described in the Tables will be identified.

[0070] Other embodiments of the present invention provide methods for identifying genes that may be associated with phenotypic variation. According to various aspects of these embodiments, the chromosomal location of a SNP associated with a particular phenotypic variation can be determined, by means well known to those skilled in the art. Once the chromosomal location is determined genes suspected to be involved with determination of the phenotype can be analyzed. Such genes may be identified by sequencing adjacent portions of the chromosome or by comparison with analogous section of the human genetic map (or known genetic maps for other species). An early example of the existence of clusters of conserved genes is reviewed in Womack (1987), where genes mapping to the same chromosome in one species were observed to map to the same chromosome in other, closely related, species. As mapping resolution improved, reports of the conservation of gene structure and order within conserved chromosomal regions were published (for example, Grosz et al, 1992). More recently, large scale radiation hybrid mapping and BAC sequence have yielded chromosome-scale comparative mapping predictions between human and bovine genomes (Everts-van der Wind et al., 2005), between human and porcine genomes (Yasue et al., 2006) and among vertebrate genomes (Demars et al., 2006)

[0071] Other embodiments of the invention provide methods for identifying causal mutations that underlie two or more quantitative trait loci (QTL). Various aspects of this embodiment of the invention provide for the identification QTL that are in allelic association with two or more of the SNPs described in Tables 1 and 3. Once these SNPs are identified, it is within the ability of skilled artisans to identify mutations located proximal to such SNP(s). Further, one skilled in the art can identify genes located proximate to the identified SNP(s) and evaluate these genes to select those likely to contain the causal mutation. Once identified, these genes and the surrounding sequence can be analyzed for the presence of mutations, in order to identify the causal mutation.

[0072] Furthermore, once genes associated with phenotypic variation have been identified, the accuracy of the analysis can be improved by investigating interaction effects. In absence of interaction effects among QTL, one can utilize the marginal effect of individual QTL for faster genetic improvement. In general, one would estimate the breeding value of each allele or genotype and use the estimated breeding values in conjunction with animal's polygenic breeding value to make breeding decisions.

[0073] In presence of interaction effects, the true breeding value of a haplotype consisting of polymorphisms from multiple QTL is different from the summation of the breeding value of individual polymorphisms at each QTL. Therefore, the approach as described above designed for absence of nonallleic interaction is suboptimum. Instead, one should estimate breeding values of haplotypes or genotype configurations for optimization of genetic improvement.

[0074] The population frequencies of haplotypes are used in estimating breeding value of an haplotype. In presence of population- wise linkage disequilibrium, the frequency of a haplotype is different from the product of corresponding allelic frequencies. In this case, it is more appropriate to use haplotype frequencies for breeding value estimation.

[0075] Additional benefits from appropriately using interaction effects in a breeding program come from the difference between the true breeding value of a haplotype and the sum of the breeding values of the corresponding alleles. The sizes of differences are determined by the magnitude of interaction effects and the extent of population- wise linkage disequilibrium between interactive QTL.

[0076] Interaction effects can also be used to produce genetically superior crossbred or hybrid animals for higher efficiency of commercial production. As an example, assume that genotype AiA ₂ BiB ₂ is the best genotype and is better than the summation of the breeding values (and genotypic values) of genotype AiA ₂ and BiB ₂ . One way

to utilize it is to create two lines with genotype AiAiBiBi and A2A2B2B2 respectively. These two lines could be from different breeds to create an ideal crossbred, or from within an existing breed population to create an ideal hybrid. Crossbreds or hybrids created from these two lines will all have genotypes A1A2B1B2, which improves the efficiency of commercial production.

[0077] Interaction effects can also be used within computer mating programs to produce genetically superior offspring for higher efficiency of commercial production. As an example, assume that genotype A1A2B1B2 is the best genotype and is better than the summation of the breeding values (and genotypic values) of genotype Ai A ₂ and B ₁ B ₂ . One way to utilize it is to identify which cows and bulls have genotype AiAiBiBi and A ₂ A ₂ B ₂ B ₂ . When a potential mates are ranked in the mating program, the interaction effects could also be included when calculating the estimated breeding value of potential offspring. For example, for individuals with the A ₁ A ₁ B ₁ B ₁ , potential mates with the A ₂ A ₂ B ₂ B ₂ genotype would have additional mating value from the favorable interaction term. Of course, this idea could be extended for multiple interactions between multiple sets of loci with favorable or unfavorable interactions. Since managing this amount of information would be extremely difficult in the normal application of artificial insemination, computer mating could be used to manage and optimize matings.

[0078] As mentioned above (see paragraph 76), two distinct sub-lines could be created from within an existing breed population in order to create the ideal hybrid via crossing these two sub-lines, such that all commercial offspring have genotypes A ₁ A ₂ B ₁ B ₂ , which improves the efficiency of commercial production. However, the creation of sub-lines optimized for maximizing the interaction terms for multiple interactions could be quite complicated. One solution is the use of computer mating to create ideal sub-lines for eventual use in producing optimized hybrids. Depending on the ultimate commercial value of different traits, several sub-lines could be created to optimize the interaction effects for different breeding goals.

EXAMPLES

[0079] The following examples are included to demonstrate general embodiments of the invention. It should be appreciated by those of skill in the art that the techniques disclosed in the examples which follow represent techniques discovered by the inventors to function well in the practice of the invention, and thus can be considered to constitute preferred modes for its practice. However, those of skill in the art should, in light of the present disclosure, appreciate that many changes can be made in the specific embodiments which are disclosed and still obtain a like or similar result without departing from the invention.

[0080] All of the compositions and methods disclosed and claimed herein can be made and executed without undue experimentation in light of the present disclosure. While the compositions and methods of this invention have been described in terms of preferred embodiments, it will be apparent to those of skill in the art that variations may be applied without departing from the concept and scope of the invention.

Example 1 : Determining Associations between Genetic Markers and Phenotypic Traits

[0081 ] Simultaneous discovery and fine-mapping on a genome- wide basis of genes underlying quantitative traits (Quantitative Trait Loci: QTL) requires genetic markers densely covering the entire genome. As described in this example, a whole-genome, dense-coverage marker map was constructed from microsatellite and single nucleotide polymorphism (SNP) markers with previous estimates of location in the bovine genome, and from SNP markers with putative locations in the bovine genome based on homology with human sequence and the human/cow comparative map. A new linkage-mapping software package was developed, as an extension of the CRIMAP software (Green et al., Washington University School of Medicine, St. Louis, 1990), to allow more efficient mapping of densely-spaced markers genome-wide in a pedigreed livestock population (Liu and Grosz Abstract CO 14; Grapes et al. Abstract W244; 2006 Proceedings of the XIV Plant and Animal Genome Conference, www.intl-pag.org). The new linkage mapping tools build on the basic mapping principles programmed in CRIMAP to improve efficiency through partitioning of large pedigrees, automation of chromosomal assignment and two-point linkage analysis, and merging of sub-maps into complete chromosomes. The resulting whole-

genome discovery map (WGDM) included 6,966 markers and a map length of 3,290 cM for an average map density of 2.18 markers/cM. The average gap between markers was 0.47 cM and the largest gap was 7.8 cM. This map provided the basis for whole-genome analysis and fine-mapping of QTL contributing to variation in productivity and fitness in dairy cattle.

Discovery and Mapping Populations

[0082] Systems for discovery and mapping populations can take many forms. The most effective strategies for determining population-wide marker/QTL associations include a large and genetically diverse sample of individuals with phenotypic measurements of interest collected in a design that allows accounting for non-genetic effects and includes information regarding the pedigree of the individuals measured. In the present example, an outbred population following the grand-daughter design (Weller et al, 1990) was used to discover and map QTL: the population, from the Holstein breed, had 529 sires each with an average of 6.1 genotyped sons, and each son has an average of 4216 daughters with milk data. DNA samples were collected from approximately 3,200 Holstein bulls and about 350 bulls from other dairy breeds; representing multiple sire and grandsire families.

Phenotypic Analyses

[0083] Dairy traits under evaluation include traditional traits such as milk yield ("MILK") (pounds), fat yield ("FAT") (pounds), fat percentage ("FATPCT") (percent), productive life ("PL") (months), somatic cell score ("SCS") (Log), daughter pregnancy rate ("DPR") (percent), protein yield ("PROT") (pounds), protein percentage ("PROTPCT") (percent), and net merit ("NM") (dollar). These traits are sex-limited, as no individual phenotypes can be measured on male animals. Instead, genetic merits of these traits defined as PTA (predicted transmitting ability) were estimated using phenotypes of all relatives. Most dairy bulls were progeny tested with a reasonably larger number of daughters (e.g., >50), and their PTA estimation is generally more or considerably more accurate than individual cow phenotype data. The genetic evaluation for traditional dairy traits of the US Holstein population is performed quarterly by USDA. Detailed descriptions of traits, genetic evaluation procedures, and genetic parameters used in the evaluation can be found at the USDA AIPL web site (www.aipl.arsusda.gov). It is meaningful to note that the dairy traits

evaluated in this example are not independent: FAT and PROT are composite traits of MILK and FATPCT, and MILK and PROTPCT, respectively. NM is an index trait calculated based on protein yield, fat yield, production life, somatic cell score, daughter pregnancy, calving difficulty, and several type traits. Protein yield and fat yield together account for >50% of NM, and the value of milk yield, fat content, and protein content is accounted for via protein yield and fat yield.

[0084] PTA data of all bulls with progeny testing data were downloaded from the USDA evaluation published at the AIPL site in November 2005. The PTA data were analyzed using the following two models: y _jj = S ₁ + PTAd y [Equation

2] yi = μ + P ₁ (SPTA) ₁ + PTAd ₁ [Equation

3] where y _s (vy) is the PTA of the i ^th bull (PTA of the j ^th son of the i ^th sire); s; is the effect of the i ^th sire; (SPTA); is the sire's PTA of the i ^th bull of the whole sample; μ is the population mean; PTAd, (PTAd _y ) is the residual bull PTA.

[0085] Equation 2 is referred to as the sire model, in which sires were fitted as fixed factors. Among all USA Holstein progeny tested bulls, a considerably large number of sires only have a very small number of progeny tested sons (e.g., some have one son), and it is clearly undesirable to fit sires as fixed factors in these cases. It is well known the USA Holstein herds have been making steady and rapid genetic progress in traditional dairy traits in the last several decades, implying that the sire's effect can be partially accounted for by fitting the birth year of a bull. For sires with <10 progeny tested sons, sires were replaced with son's birth year in Equation 2. Equation 3 is referred to as the SPTA model, in which sire's PTA are fitted as a covariate. Residual PTA (PT Adj or PTAdy) were estimated using linear regression.

SNP-trait Association Analyses

[0086] In the present example, linkage disequilibrium (LD) mapping was performed in the aforementioned discovery population using statistical analyses based on probabilities of individual ordered genotypes estimated conditional on observed marker genotypes. The first step was to estimate sire's ordered genotype probabilities at all linked markers conditional on grandsire's and offspring marker genotype data.

The exact calculation quickly becomes computationally infeasible as the size and complexity of the pedigree and number of linked markers increases. For example, there are, in total 2 ^k ordered genotypes for all linked loci when a sire has k linked heterozygous loci. A stepwise procedure developed based on a likelihood ratio test was used for estimating probabilities of sire's ordered genotypes at all linked markers.

[0087]The probabilities of ordered genotypes at loci of interest were estimated conditional on flanking informative markers as follows:

P(H _slk H _dlk | M) = ∑ ∑ P(H _sa H _db I M) * P{H _sιk H _dlk | H _sa H _db ,M) [Equation a b

4]

Where P(H _sa H _db | M) is the probability of sire having a pair of haplotypes (or order genotype) H _sa H _db at all linked loci conditional on the observed genotype data M, and P{H _sιk H _dlk I H _sa H _db , M) is the probability of a son having ordered genotype H _sιk H _dlk at loci of interest conditional on sire's ordered genotype H _sa H _db at all linked loci and the observed genotype data M.

[0088] To determine associations between haplotypes probabilities and trait phenotypes, haplotypes of neighboring (and/or non-neighboring) markers across each chromosome were defined by setting the maximum length of a chromosomal interval and minimum and maximum number of markers to be included. Clearly, one needs to set similar parameters to form or define groups of marker loci for haplotype evaluation. The association between pre-adjusted trait phenotypes and haplotype (or pair of haplotype that is alternatively termed as ordered genotypes) was evaluated via a regression approach with the following models:

PTAd _k = ∑ β _s ,P(H _slk ) + e _k [Equation 5 ]

PTAd _k = ∑ β _dl P(H _dlk ) + e _k [Equation

6]

PTAd _k = ∑ β _sι [P(H _sιk ) + P(H _dlk )] + e _k [ Equation

7 ]

PTAd _k = ∑ β _il [P(H _ilk H _djk ) + P(H _≠ H _dlk )] + e _k [Equation

8] where PTAd _k is the preadjusted PTA of the k ^λ bull as defined in Equation 3 under the sire model and can be replaced with PT Adi as defined in Equation 3 under the SPTA model, and β _k is the residual; P(H _s ,k) and P(Hdik) are the probability of paternal and maternal haplotype of individual k being haplotype i; P(H _S ikH _d ,k) is the probability of individual k has paternal haplotype i and maternal haplotype j that can be estimated using Equation 4; all β are corresponding regression coefficients. Equations 5, 6, 7, and 8 are designed to model paternal haplotype, maternal haplotype, additive haplotype, and genotype effects, respectively.

[0089] Least-squares methods were used to estimate the effect of a haplotype or haplotype pair on a phenotypic trait and the regular F-test used to test the significance of the effect. Permutation tests were performed based on phenotype permutation (20,000) within each paternal half-sib family to estimate Type I error rate (p value).

Example 2: Analyzing for Interaction Effects between Multiple Genetic Markers

[00901 Clustering of SNPs from a candidate gene. Mainly due to small effective population sizes and strong selection, alleles from tightly-linked SNPs are generally associated in animal populations (e.g., Farnir et al., 2000; Du et al., 2007). Clearly, if two SNPs are in perfect LD, their association with traits of interest and interaction with other SNPs on traits of interest will be similar, which doesn't provide much additional statistical evidence. It is, therefore, helpful to cluster SNPs from the same candidate gene when multiple SNPs at a single gene are genotyped.

[0091] Trait phenotype preadjustment. This study focuses on traditional dairy traits, including milk yield ("MILK") (pounds), fat yield ("FAT") (pounds), fat percentage ("FATPCT") (percent), productive life ("PL") (months), somatic cell score ("SCS") (Log), daughter pregnancy rate ("DPR") (percent), protein yield ("PROT") (pounds), protein percentage ("PROTPCT") (percent), and net merit ("NM") (dollar). These traits are sex-limited, as no individual phenotypes can be measured on male animals. Instead, genetic merits of these traits defined as PTA (predicted transmitting ability) were estimated using phenotypes of all relatives. Most dairy bulls were progeny

tested with a reasonably larger number of daughters (e.g., >50), and their PTA estimation is generally more or considerably more accurate than individual cow phenotype data. The genetic evaluation of traditional dairy traits of US Holstein population was performed quarterly by USDA. Detailed description of traits, genetic evaluation procedures, and genetic parameters used in the evaluation can be found at the USDA AIPL web site (http://aipl.arsusda.gov). It is meaningful to note that the dairy traits evaluated in this study are not independent: FAT and PROT are composite traits of MILK and FATPCT, and MILK and PROTPCT, respectively. NM is an index trait calculated based on protein yield, fat yield, production life, somatic cell score, daughter pregnancy, calving difficulty, and several type traits.

[0092] PTA data of all bulls with progeny testing data were downloaded from the USDA February 2007 genetic evaluation published at the AIPL site. The PTA data were analyzed using following two models: y.. = S _j + PTAd ₍ J [Equation

9] y _j = μ + P ₁ (SPTA) ₁ + PTAd ₁ [Equation

10] where y, (y _υ ) is the PTA of the i ^th bull (PTA of the j ^th son of the i ^th sire); s, is the effect of the i ^th sire; (SPTA) ₁ is the sire's PTA of the i ^th bull of the whole sample; μ is the population mean; PTAd, (PTAd _y ) is the residual bull PTA.

[0093] Equation 9 is referred to as the sire model, in which sires were fitted as fixed factors. Among all USA Holstein progeny tested bulls, a considerably large number of sires only have a very small number of progeny tested sons (e.g., some have one son), and it is clearly undesirable to fit sires as fixed factors in these cases. It is well known the USA Holstein herds have been making steady and rapid genetic progress in traditional dairy traits in the last several decades, implying that the sire's effect can be partially accounted for by fitting the birth year of a bull. For sires with <10 progeny tested sons, sires were replaced with son's birth year in Equation 9. Equation 10 is referred to as the SPTA model, in which sire's PTA are fitted as a covariate. Residual PTA (PTAd ₁ or PTAd ₁ ,) were estimated using SAS PROC GLM procedure and used for further candidate gene analysis in this study.

[0094] Candidate sene interaction analysis. The association between SNP and residual PTA of each dairy trait was analyzed using the following linear models:

PTAd, = + e, [Equation

[0095] where PTAd ₁ is the preadjusted PTA of the i ^th bull as defined in Equation 10 under the sire model and can be replaced with PTAd ₁ as defined in Equation 9 under the SPTA model; n _a is the number of unordered genotypes at SNP j (j = 1 , 2); e, is the residual effect; β _k is the effect of genotype indicator l _n γ, and δ _kh is the interaction effect between genotype indicator I,i _k at the 1 ^st SNP and genotype indicator I _12I1 at the 2 ^nd SNP; and genotype indicator I^ is defined as 1 if genotype being k at the jth SNP . [Equation

0 otherwise

12]

[0096] Overall analyses consist of two steps. Original PTA data was first preadjusted using all bulls evaluated by USDA (Equation 9 and 10), and the preadjusted PTA was analyzed using Equation 1 1 for statistical associations between SNP and trait. The combination of Equations 9 and 11 , and 10 and 11 was referred as to the sire model and the SPTA model, respectively.

[0097] Results of this analysis are shown in Table 1.

Example 3: Use of Single Nucleotide Polymorphisms (SNPs) to improve offspring traits

[0098] To improve the average genetic merit of a population for a chosen trait, two or more of the markers with significant association to that trait can be used in selection of breeding animals. In the case of each discovered locus, use of animals possessing a marker allele (or a haplotype of multiple marker alleles) in population- wide LD with a favorable QTL allele will increase the breeding value of animals used in breeding, increase the frequency of that QTL allele in the population over time and thereby increase the average genetic merit of the population for that trait. This increased genetic merit can be disseminated to commercial populations for full realization of value.

[0099] For example, a progeny-testing scheme could greatly improve its rate of genetic progress or graduation success rate via the use of markers for screening juvenile bulls. Typically, a progeny testing program would use pedigree information and performance of relatives to select juvenile bulls as candidates for entry into the program with an accuracy of approx 0.5. However, by adding marker information, young bulls could be screened and selected with much higher accuracy. In this example, DNA samples from potential bull mothers and their male offspring could be screened with a genome- wide set of markers in linkage disequilibrium with QTL, and the bull-mother candidates with the best marker profile could be contracted for matings to specific bulls. If superovulation and embryo transfer (ET) is employed, a set of 5-10 offspring could be produced per bull mother per flush procedure. Then the marker set could again be used to select the best male offspring as a candidate for the progeny test program. If genome-wide markers are used, it was estimated that accuracies of marker selection could reach as high as 0.85 (Meuwissen et al., 2001). This additional accuracy could be used to greatly improve the genetic merit of candidates entering the progeny test program and thereby increasing the probability of successfully graduating a marketable progeny-tested bulls. This information could also be used to reduce program costs by decreasing the number of juvenile bull candidates tested while maintaining the same number of successful graduates. In the extreme, very accurate marker breeding values (MBV) could be used to directly market semen from juvenile sires without the need of progeny-testing at all. Due to the fact that juveniles could now be marketed starting at puberty instead of 4.5 to 5 years, generation interval could be reduced by more than half and rates of gain could increase as much as 68.3% (Schrooten et al., 2004). With the elimination of the need for progeny testing, the cost of genetic improvement for the artificial insemination industry would be vastly improved (Schaeffer, 2006).

[0100] In an alternate example, a centralized or dispersed genetic nucleus (GN) population of cattle could be maintained to produce juvenile bulls for use in progeny testing or direct sale on the basis of MBVs. A GN herd of 1000 cows could be expected to produce roughly 3000 offspring per year, assuming the top 10-15% of females were used as ET donors in a multiple-ovulation and embryo-transfer (MOET) scheme. However, markers could change the effectiveness MOET schemes and in vitro embryo production. Previously, MOET nucleus schemes have proven to be

promising from the standpoint of extra genetic gain, but the costs of operating a nucleus herd together with the limited information on juvenile animals has limited widespread adoption. However, with marker information, juveniles can be selected much more accurately than before resulting in greatly reduced generation intervals and boosted rates of genetic response. This is especially true in MOET nucleus herd schemes because, previously, breeding values of full-sibs would be identical, but with marker information the best full-sib can be identified early in life. The marker information would also help limit inbreeding because less selection pressure would be placed on pedigree information and more on individual marker information. An early study (Meuwissen and van Arendonk, 1992) found advantages of up to 26% additional genetic gain when markers were employed in nucleus herd scenarios; whereas, the benefit in regular progeny testing was much less.

[0101] Together with MAS, female selection could also become an important source of genetic improvement particularly if markers explain substantial amounts of genetic variation. Further efficiencies could be gained by marker testing of embryos prior to implantation (Bredbacka, 2001). This would allow considerable selection to occur on embryos such that embryos with inferior marker profiles could be discarded prior to implantation and recipient costs. This would again increase the cost effectiveness of nucleus herds because embryo pre-selection would allow equal progress to be made with a smaller nucleus herd. Alternatively, this presents further opportunities for preselection prior to bulls entering progeny test and rates of genetic response predicted to be up to 31% faster than conventional progeny testing (Schrooten et al., 2004).

[0102] The first step in using a SNP for estimation of breeding value and selection in the GN is collection of DNA from all offspring that will be candidates for selection as breeders in the GN or as breeders in other commercial populations (in the present example, the 3,000 offspring produced in the GN each year). One method is to capture shortly after birth a small bit of ear tissue, hair sample, or blood from each calf into a labeled (bar-coded) tube. The DNA extracted from this tissue can be used to assay an essentially unlimited number of SNP markers and the results can be included in selection decisions before the animal reaches breeding age.

[0103] One method for incorporating into selection decisions the markers (or marker haplotypes) determined to be in population-wide LD with valuable QTL alleles (see

Example 1) is based on classical quantitative genetics and selection index theory (Falconer and Mackay, 1996; Dekkers and Chakraborty, 2001). To estimate the effect of the marker in the population targeted for selection, a random sample of animals with phenotypic measurements for the trait of interest can be analyzed with a mixed animal model with the marker fitted as a fixed effect or as a covariate (regression of phenotype on number of allele copies). Results from either method of fitting marker effects can be used to derive the allele substitution effects, and in turn the breeding value of the marker:

(Xi = q[a + d(q - p)] [Equation 13]

(X ₂ = -p[a + d(q - p)] [Equation 14]

(X = a + d(q - p) [Equation 15]

g _A- Ai = 2((Xi) [Equation 16]

g _A iA2 = (CLx) + (CL ₂ ) [Equation 17]

gA2A2 = 2(CL ₂ ) [Equation 18]

where (Xi and (X ₂ are the average effects of alleles 1 and 2, respectively; (X is the average effect of allele substitution; p and q are the frequencies in the population of alleles 1 and 2, respectively; a and d are additive and dominance effects, respectively; g _A i _A i, gAiA2 and gA2A2 are the (marker) breeding values for animals with marker genotypes AlAl, A1A2 and A2A2, respectively. The total trait breeding value for an animal is the sum of breeding values for each marker (or haplotype) considered and the residual polygenic breeding value:

EBVj _j = σ gj + Oj [Equation 19]

where EBVy is the Estimated Trait Breeding Value for the i ^th animal, σ g j is the marker breeding value summed from j = 1 to n where n is the total number of markers (haplotypes) under consideration, and U ₁ is the polygenic breeding value for the i ^th animal after fitting the marker genotype(s).

[0104] These methods can readily be extended to estimate breeding values for selection candidates for multiple traits, the breeding value for each trait including information from multiple markers (haplotypes), all within the context of selection index theory and specific breeding objectives that set the relative importance of each trait. Other methods also exist for optimizing marker information in estimation of breeding values for multiple traits, including random models that account for recombination between markers and QTL (e.g., Fernando and Grossman, 1989), and the potential inclusion of all discovered marker information in whole-genome selection (Meuwissen et al., Genetics 2001). Through any of these methods, the markers reported herein that have been determined to be in population-wide LD with valuable QTL alleles may be used to provide greater accuracy of selection, greater rate of genetic improvement, and greater value accumulation in the dairy industry.

Example 4: Use of multiple SNPs with interaction effects to improve offspring traits

[0105] To illustrate the use of interaction effects in a breeding program, consider two causal mutations at two biallelic QTLs (denoted by A and B). Let Ai and A ₂ , and Bi and B ₂ be the two alleles of QTL A and B, respectively. One way to model both interaction and main effects is to fit the effects of all genotype configurations: y, = ∑/?(A _t A _{j ;} B _s B _k )I(A _t A _{j ;} B _s B _k ) + a, + ε _t [ Equation

20]

[0106] Where a,- denotes to a polygenic random effect; (A _t A _j ; B ₅ B _k ) denotes to a genotype configuration consisting of genotypes at A and B; β(A _t A _j ; B _s B _k ) is the regression coefficient for genotype configuration (A _t A _j ; B ₅ B _k ); I(A _t A _j ; B ₅ B _k ) is an index function defined as: 1 if genotype is (A ₁ A ₁ ; B, B _L ) y ^V ' ^{J s k} [Equation

0 otherwise ^M

21]

[0107] Equation [20] can be used for both detection and utilization of interaction effects. The effect of genotype configuration (A ₁ A _j -; B ₅ B _k ) in Equation 20 can be fitted as fixed effects or a random effect.

[0108] The breeding value of a haplotype consisting of one allele from each QTL can be calculated using: α(A,B _j ) = ^(A ₁ A, ;B _J B ₁ )f(A ₁ B _I ) + ^(A,A _I ;B _J B ₂ )f(A ₁ B ₂ ) + /?(A ₁ A ₂ ;B _J B ₁ )f(A ₂ B ₁ ) + _/? (A,A ₂ ;B _J B ₂ )f(A ₂ B ₂ ) [Equation

22] where f( AkB ₈ ) (k, s =1, 2) represents the frequency of haplotype A _k B _s . It should be noted that f( A _k B ₅ ) is not equal to the product of the corresponding allele frequency in presence of population-wise linkage disequilibrium.

[0109] The breeding value of an animal with genotype configuration (A _J AJ; B _k B ₅ ) can be calculated as:

BV(A ₁ A _j JB _k B ₅ ) = 2[P(A ₁ BJa(A ₁ BJ + P(A ₁ BJa(A ₁ B ₈ )

[Equation + P(A _j BJa(A _j BJ + _P (A _j B _s )α(A _j B _s )]

23] where p(AjB _j ) is the probability of a gamete produced by this animal having gamete haplotype AjBj. It should be noted that the sum of probabilities of all possible haplotypes is equal to 1 and that the value of p( AjB _j ) is a function of the recombination fraction between QTL A and B in case of a genotype configuration being heterozygous at both loci. To explain the linkage effect further, consider an animal with genotype A ₁ B ₁ /A ₂ B ₂ (i.e. consisting of haplotypes AiBi and A ₂ B ₂ ). The probabilities of four different haplotypes for this animal can be calculated as

P(A ₁ B ₁ ) = P(A ₂ BJ = 0.5(1 - 0 _AB ) [Equation

24] and

P(A ₁ B ₂ ) = P(A ₂ B ₁ ) = 0.50 _AB [Equation

25] where θ _AB represents the recombination fraction between locus A and B.

[0110] The breeding value of a genotype configuration can be used for genetic improvement purpose in the same manner as the conventional polygenic breeding value.

[0111] It should be noted that the interaction effects can be estimated using various statistical models. It should also be noted that the above procedure can be easily extended for cases with multiple alleles and/or multiple loci (e.g., by including all possible genotype configurations in Equation 20).

Example 5: Identification of SNPs

[01 12] A nucleic acid sequence contains a SNP of the present invention if it comprises at least 20 consecutive nucleotides that include and/or are adjacent to a polymorphism described in Tables 1 and 3 and the Sequence Listing. Alternatively, a SNP of the present invention may be identified by a shorter stretch of consecutive nucleotides which include or are adjacent to a polymorphism which is described in Tables 1 and 3 and the Sequence Listing in instances where the shorter sequence of consecutive nucleotides is unique in the bovine genome. A SNP site is usually characterized by the consensus sequence in which the polymorphic site is contained, the position of the polymorphic site, and the various alleles at the polymorphic site. "Consensus sequence" means DNA sequence constructed as the consensus at each nucleotide position of a cluster of aligned sequences. Such clusters are often used to identify SNP and Indel (insertion/deletion) polymorphisms in alleles at a locus. Consensus sequence can be based on either strand of DNA at the locus, and states the nucleotide base of either one of each SNP allele in the locus and the nucleotide bases of all Indels in the locus, or both SNP alleles using degenerate code (IUPAC code: M for A or C; R for A or G; W for A or T; S for C or G; Y for C or T; K for G or T; V for A or C or G; H for A or C or T; D for A or G or T; B for C or G or T; N for A or C or G or T; Additional code that we use include I for "-"or A; O for "-" or C; E for "-" or G; L for "-" or T; where "-" means a deletion). Thus, although a consensus sequence may not be a copy of an actual DNA sequence, a consensus sequence is useful for precisely designing primers and probes for actual polymorphisms in the locus.

[0113] Such SNP have a nucleic acid sequence having at least 90% sequence identity, more preferably at least 95% or even more preferably for some alleles at least 98% and in many cases at least 99% sequence identity, to the sequence of the same number of nucleotides in either strand of a segment of animal DNA which includes or is adjacent to the polymorphism. The nucleotide sequence of one strand of such a segment of animal DNA may be found in a sequence in the group consisting of SEQ ID NO:1 through SEQ ID NO: 175. It is understood by the very nature of polymorphisms that for at least some alleles there will be no identity at the polymorphic site itself. Thus, sequence identity can be determined for sequence that is exclusive of the polymorphism sequence. The polymorphisms in each locus are described in Tables 1 and 3.

[0114] Shown below are examples of public bovine SNPs that match each other:

SNP ss38333809 was determined to be the same as ss38333810 because 41 bases (with the polymorphic site at the middle) from each sequence match one another perfectly (match length=41, identity=100%).

ss38333809 : tcttacacatcaggagatagytccgaggtggatttctacaa

I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I Il I I I I I I I I ss38333810 : tcttacacatcaggagatagytccgaggtggatttctacaa ss38333809 is SEQ ID NO:172 and ss38333810 is SEQ ID NO:173

[0115] SNP ss38333809 was determined to be the same as ss38334335 because 41 bases (with the polymorphic site at the middle) from each sequence match one another at all bases except for one base (match length=41, identity=97%).

ss38333809 : tcttacacatcaggagatagytccgaggtggatttctacaa

Il I I I I I I I I I I I I I I I I I Il I I I I I I I I I I I I I I I I I M ss38334335 : tcttacacatcaggagatggytccgaggtggatttctacaa ss38333809 is SEQ ID NO: 174 and SS38334335 is SEQ ID NO: 175

Example 6: Quantification of and genetic evaluation for production traits

[01 16] Quantifying production traits can be accomplished by measuring milk of a cow and milk composition at each milking, or in certain time intervals only. In the USDA yield evaluation the milk production data are collected by Dairy Herd Improvement Associations (DHIA) using ICAR approved methods. Genetic evaluation includes all cows with the known sire and the first calving in 1960 and later and pedigree from birth year 1950 on. Lactations shorter than 305 days are extended to 305 days. All records are preadjusted for effects of age at calving, month of calving, times milked per day, previous days open, and heterogeneous variance. Genetic evaluation is conducted using the single-trait BLUP repeatability model. The model includes fixed effects of management group (herd x year x season plus register status), parity x age, and inbreeding, and random effects of permanent environment and herd by sire interaction. PTAs are estimated and published four times a year (February, May, August, and November). PTAs are calculated relative to a five year stepwise base i.e., as a difference from the average of all cows born in 2000. Bull PTAs are published estimating daughter performance for bulls having at least 10 daughters with valid lactation records.

Example 7: Quantification of reproductive traits in daughters (cows) and sires' PTAs

[0117] Quantification of and genetic evaluation of the reproductive capability such as calving ease (CE), occurrence of stillbirths (SB) and daughter pregnancy rate (DPR). Calving ease measures the ability of a particular cow (daughter) to calve easily. CE is scored by the owner on a scale of 1 to 5, 1 meaning no problems encountered or unobserved birth and 5 meaning extreme difficulty. The CE PTAs for sires are expressed as percent difficult births in primiparous daughter heifers (%DBH), where

difficult births are those scored as requiring considerable force or being extremely difficult (4 or 5 on a five point scale). SB is scored by the owner on a scale of 1 to 3, 1 meaning the calf was born alive and was alive 48 h postpartum, 2 meaning the calf was born dead, and 3 indicating the calf was born alive but died within 48 h postpartum. SB scores of 2 and 3 are combined into a single category for evaluation. The SB PTAs for sires are expressed as percent stillbirths in daughter heifers (%SBH), where stillborn calves are those scored as dead at birth or born alive but died within 48 h of birth (2 or 3 on a three point scale). Pregnancy rate is a function of the number of days open, which is the number of days between calving and a successful breeding. DPR is defined as the percentage of nonpregnant cows (daughters) that become pregnant during each 21 -day period. A DPR PTA of "1" implies that daughters from this bull are 1% more likely to become pregnant during that estrus cycle than a bull with a DPR PTA of zero.

Example 8: Quantification of and genetic evaluation for productive life (PL)

[0118] Productive life (PL) is defined as the length of time a cow remains in a milking herd before removal by voluntary or involuntary culling (due to health or fertility problems), or death. PL is usually measured as the number of days, months, or days in milk (DIM) from the first calving to the day the cow exits the herd (due to death, culling, or selling to non-dairy purposes). Because some cows are still alive at the time of data collection, their records are projected (VanRaden, P.M. and EJ. H. Klaaskate. 1993) or treated as censored (Ducrocq, 1987). The USDA genetic evaluation for PL includes all cows with first calving in 1960 and later (born in 1950 and later for the pedigree). Cows born at least 3 years prior to evaluation, with a valid sire ID and first lactation records are considered. PL is considered to be completed at 7 years of age. Records are extended for cows that have not had the opportunity to reach 7 years of age because they are still alive, were sold for dairy purposes, or the herd discontinued testing. Cows sold for dairy purposes or in herds that discontinued testing receive extended records if they had opportunity to reach 3 years of age; otherwise their records are discarded. The method of genetic evaluation is a single trait BLUP animal model. The statistical model includes effects of management group (based on herd of first lactation and birth date) and sire by herd interaction.. Sires' PTAs for PL are calculated relative to a five year stepwise base i.e., as a difference from the average PL of all cows born in 2000.

Example 9: Quantification of somatic cell score in daughters (cows) and sires' PTAs

[0119] Quantifying somatic cell score (SCS) is accomplished by calculating log ₂ (SCC/100,000)+3, where SCC is number of somatic cells per milliliter of milk from a cow (daughter). The SCS PTAs for sires are expressed as a deviation from a SCS PTA of zero.

Example 10: Discovery of Novel Associations of SNPs within Candidate Genes:

\Q\2O\Animal sample and senotvpins. A total of 3145 Holstein bulls with a NAAB code were downloaded from USDA AIPL web site (http://aipl.arsusda.gov) to form a resource population for this study. A total of 22 SNPs (single nucleotide polymorphisms) from 10 candidate genes (leptin, poulFl, kappa casein, osteopontin, beta2-adrenergic receptor, growth hormone receptor, proteinase inhibitor, breast cancer resistance protein, diacylglycerol acyltransferase) were genotyped internally using the ABI Taqman platform or externally (Genaissance Pharmaceuticals, Inc., New Haven, CT) using various chemistries.

[012I]AIl SNPs used in this study have two alleles, resulting in a total of three unordered genotypes for each SNP (two homozygotes and one heterozygote). If <300 bulls are homozygous for the minor allele, the minor allele homozygote class can be merged with the heterozygote to form a composite genotype (genotype iiij is denoted to both genotype ii and ij) or excluded from analyses. Consequently, analyses can be performed using original genotypes, composite genotypes, and data that excludes the least frequent genotype when the number of bulls with least frequent genotype is smaller than 300.

\0l22)Trait phenotyye preadiustment. Analyzed traits include milk yield ("MILK") (pounds), fat yield ("FAT") (pounds), fat percentage ("FATPCT") (percent), productive life ("PL") (months), somatic cell score ("SCS") (Log), daughter pregnancy rate ("DPR") (percent), protein yield ("PROT") (pounds), protein percentage ("PROTPCT") (percent), and net merit ("NM") (dollar). These traits are sex-limited, so genetic merits of these traits are defined as PTA (predicted

transmitting ability) and were estimated using phenotypes of all relatives. Detailed description of traits, genetic evaluation procedures, and genetic parameters used in the evaluation can be found at the USDA AIPL web site (http://aipl.arsusda.gov). It is meaningful to note that the dairy traits evaluated in this study are not independent.

[0123]PTA data of all bulls with progeny testing data were downloaded from the USDA evaluation published at the AIPL site in November, 2005. The PTA data were analyzed using the following two models: y.. = S ₁ -I- PTAd ₁ J [Equation

26] y _; = μ + P ₁ (SPTA) ₁ + PTAd ₁ [Equation

27] where yj (yy) is the PTA of the i ^th bull (PTA of the j ^th son of the i ^th sire); s; is the effect of the i ^th sire; (SPT A)j is the sire's PTA of the i ^th bull of the whole sample; μ is the population mean; PTAd; (PT Ad _y ) is the residual bull PTA.

[0124] Equation 26 is referred to as the sire model, in which sires were fitted as fixed factors. For sires with <10 progeny tested sons, sires were replaced with son's birth year in Equation 26. Equation 27 is referred to as the SPTA model, in which sire's PTA are fitted as a covariate. Residual PTA (PTAd; or PTAd _y ) were estimated using SAS PROC GLM procedure and used for further candidate gene analysis in this study.

\0\25 ^~ \Candidate sene analysis. The association between SNP and residual PTA of each dairy trait was analyzed using the following linear models:

PTAd ₁ = μ + P ₁ X ₁ -H e ₁ [Equation

28] i, PTAd ₁ = £ l, _k β _k + e, [Equation k=l

29] where PTAd; is the preadjusted PTA of the i ^th bull as defined in Equation 26 under the sire model and can be replaced with PTAd j as defined in Equation 27 under the SPTA model; Xj is the number of copies of a specific SNP allele that the i ^th bull has, and β ₂ is

the regression coefficient for XJ; n _g is the number of unordered genotypes; e _\ is the residual effect; and β _k is the effect of genotype indicator Ijk that is defined as 1 if genotype being k [Equation 0 otherwise

30]

[0126]Overall analyses consist of two steps. Original PTA data was first preadjusted using all bulls evaluated by USDA (Equations 26 and 27), and the preadjusted PTA was analyzed using Equations 28 and 29 for statistical associations between SNP and trait. The combination of Equations 26 and 28, 26 and 29, 27 and 28, and 27 and 29 was referred as to the sire_allele, sire_genotype, SPTA_allele, SPTA_genotype model, respectively.

[0127]The effect of a SNP on a trait was described by additive (= (G _U - G _jj )/2), dominance (= G _y - (G _U - G _jj )/2), or difference between two genotype (G _y - G _jj ), where i, and j denote the two alleles of the SNP, and G _y represents the mean of genotype ij.

[0128]Results of this analysis are shown in Table 1 and the Sequence Listing. Abbreviations for traits include the following: Fitness traits including pregnancy rate (PR), daughter pregnancy rate (DPR), productive life (PL),somatic cell count (SCC) and somatic cell score (SCS); and productivity traits including total milk yield (MY), milk fat percentage (FP), milk fat yield (FY), milk protein percentage (PP), milk protein yield (PY), total lifetime production (PL); and Net Merit (NM).

Table 1: The following table describes genes, markers, trait associations, and interactions effects resulting from the experiments described herein.

^♦ Details for each polymorphism including location, length, SEQ ID number, and alleles, are located in Table 4 and the sequence listing.

Example 11 : Discovery of New Markers in the CATSPER, CD14, IGF2R, LIF, OSM, RCN3, RIM2, and TLE4 Genes and Association with Dairy Productivity Traits

[0129]A whole-genome scan was conducted using 3000 Holstein bulls to identify quantitative trait loci (QTL) for dairy productivity traits on all bovine chromosomes. This invention concerns QTL (and selected candidate genes) on chromosomes BTA07 (CD14), BTA08 (TLE4) BTA09 (IGF2R), BTA14 (RIM2), BTAl 7 (LIF, OSM), BTAl 8 (RCN3), and BTA29 (CATSPER). Flanking sequences of the SNPs used in the whole-genome scan that were found to be associated with dairy productivity traits were used to BLAST against the public bovine genome sequence assembly. Genes were identified proximal and distal (within ~5 cM) to the QTL SNP location and researched to determine putative function. For selected QTL on chromosomes BTA07, BTA08, BTA09, BTA14, BTA17, BTA18, and BTA29 candidate genes, CD14, TLE4, IGF2R, RIM2, LIF and OSM, RCN3, and CATSPER, respectively, were chosen for novel marker discovery. Gene and NCBI GeneID numbers can are shown in Table 2 below (www.ncbi.nlm.nih.gov/sites/entrez?db=Gene).

Table 2: The following table describes genes correlated NCBI GeneID numbers.

[0130] A total of 23 Holstein bulls, selected from the 3000 used for the whole- genome scan, were used as a discovery panel to identify novel genetic markers (SNPs and insertion-deletions, or INDELs) by sequencing the candidate genes and comparing forward and reverse strand sequences between all 23 samples. All Holstein DNA was extracted from semen using standard protocols. Standard laboratory PCR was used to amplify DNA fragments containing the coding region and regulatory regions of the genes for sequencing. Standard direct PCR product sequencing was conducted and resolved on an ABI 3730x1 Automated Sequencer (Applied Biosystems, Foster City, CA)..

[0131] To perform association analysis, genetic markers discovered in candidate genes using the panel of 23 Holsteins were genotyped by sequencing on a panel of 108 additional Holsteins (with 88 selected from the 3000 used in the whole-genome scan and 20 unique to the 108 animal panel). Genotypes of the 108 animal panel were combined with the genotypes from the 23 animal discovery panel for a total of 131 genotypes per genetic marker. Association analysis was carried as described above.

[0132] This experiment resulted in number confirmed associations in and around CD 14, TLE4, IGF2R, RIM2, LIF and OSM, RCN3, and CATSPER as well as the identification of a large number of SNPs. Results of the association study are further described in Tables 1 and the Sequence Listing, and novel polymorphisms are identified in Tables 3 and the Sequence listing. In each case, details regarding the location, length, and alleles for each polymorphism are described in Table 4.

Table 3: The following table includes a list of novel markers, gene names, and SEQ ID numbers resulting from the experiment described above.

*Details for each polymorphism including location, SEQ ID number, and alleles, are located in Table 4 and the sequence listing.

Table 4. The following table describes the polymorphisms listed in Tables 1 and 3 in more detail, including the SEQ ID number, polymorphism Position, and Alleles.

REFERENCES

[0133] The references cited in this application, both above and below, are specifically incorporated herein by reference.

Non-Patent Literature

Abdel-Azim, G and Freeman, AE, (2002) J. Dairy Sci. 85:1869-1880.

Blott, S., Kim, J.J., Moisio, S., et al. (2003). Genetics 163: 253-266.

Ciobanu, DC, Bastiaansen, JWM, Longergan, SM, Thomsen, H, Dekkers, JCM, Plastow, GS, and Rothschild, MF, (2004) J. Anim. Sci. 82:2829-39.

Cohen-Zinder, M. et al. (2005) Genome Res. 15:936-44. Davis, GP and DeNise, SK, (1998) J. Anim. Sci. 76:2331-39. Dekkers, JCM, and Chakraborty, R, (2001) J. Anim. Sci. 79:2975-90.

Demars J, Riquet J, Feve K, Gautier M, Morisson M, Demeure O, Renard C, Chardon P, Milan D. (2006), BMC Genomics, 24:7-13

Du and Hoeschele, (2000) Genetics 156:2051-62.

Ducrocq, V. 1987. An analysis of length of productivelife in dairy cattle. Ph.D. Diss.. Cornell Univ., Ithaca, NY; Univ. Microfilms Int., Ann Arbor, MI.

Everts-van der Wind A, Larkin DM, Green CA, Elliott JS, Olmstead CA, Chiu R, Schein JE, Marra MA, Womack JE, Lewin HA. (2005) Proc Natl Acad Sci U S A, 20; 102(51): 18526-31.

Falconer, DS, and Mackay, TFC, (1996) Introduction to Quantitative Genetics. Harlow, UK: Longman.

Fernando, R, and Grossman, M, (1989) Marker assisted selection using best linear unbiased prediction. Genetics Selection Evolution 21 :467-77 '.

Franco, MM, Antunes, RC, Silva, HD, and Goulart, LR (2005) J. Appl. Genet. 46(2): 195-200.

Grisart, B. et al. (2002) Genome Res. 12:222-231

Grosz, MD, Womack, JE, and Skow, LC (1992) Genomics, 14(4):863-868. Hayes, B, and Goddard, ME, (2001) Genet. SeI. Evol. 33:209-229.

Hayes, B, Chamberlain, A.J., Goddard, M.E. (2006) Proc. 8 ^th WCGALP 22:(16).

Kaminski, S, Ahman, A, Ruse, A, Wojcik, E, and Malewski, T (2005) J. Appl. Genet. 46(l):45-58.

Kaminski S, Brym P, Ruse A, and Wojcik E, (2006) Associations between Milk Performance Traits in Holstein Cows and 16 Candidate SNPs Identified by Arrayed Primer Extension (APEX) Microarray, Animal Biotechnology, 1 Vill i, 2006.

Kuhn, C. et al. (2004). Genetics 167:1873-81.

Kwok PY, Methods for genotyping single nucleotide polymorphisms, (2001), Annu. Rev. Genomics Hum. Genet., 2:235-258.

Meuwissen, THE, and Van Arendonk, JAM, (1992) J. Dairy ScL 75:1651-1659. Meuwissen, THE, Hayes, BJ, and Goddard, ME, (2001) Genetics. 157:1819-29. Rothschild and Plastow, (1999), AgBioTechNet 10:1-8. Schaeffer, LR (2006) J. Anim. Breed. Genet. 123:218-223. Schnabel, R. et al. (2005) PNAS 102:6896-6901.

Schrooten, C, Bovenhuis, H, van Arendonk, JAM, and Bijma, P (2005) J. Dairy ScL 88:1569-1581.

Sharma, BS, Jansen, GB, Karrow, NA, Kelton, D, and Jiang, Z, (2006) J. Dairy ScL 89:3653-3663.

Short, TH, et al. (1997) J. Anim. ScL 75:3138-3142.

Spelman, RJ and Bovenhuis, H, (1998) Animal Genetics, 29:77-84.

Spelman, RJ and Garrick, DJ, (1998) J. Dairy Sci, 81 :2942-2950.

Stearns, TM, Beever, JE, Southey, BR, Ellis, M, McKeith, FK and Rodriguez-Zas, SL, (2005) J. Anim. ScL 83:1481-93.

Syvanen AC, Accessing genetic variation: genotyping single nucleotide polymorphisms, (2001) Nat. Rev. Genet. 2:930-942.

VanRaden, P.M. and E.J.H. Klaaskate. 1993. J. Dairy ScL 76:2758-2764. Verrier, E, (2001) Genet. SeI. Evol. 33:17-38.

Villanueva, B, Pong- Wong, R, Fernandez, J, and Toro, MA (2005) J. Anim. ScL 83:1747-52.

Weller JI, Kashi Y, Soller M. (1990) J. Dairy ScL 73:2525-37 Williams, JL, (2005), Rev. ScL Tech. Off. Int. Epiz. 24(1):379-391. Windig, JJ, and Meuwissen, THE, (2004) J. Anim. Breed. Genet. 121 :26-39. Winter, A. et al. (2002). PNAS, 99:9300-9305. Womack, J, (1987), Dev. Genet. 8(4):281-293.

Yasue H, Kiuchi S, Hiraiwa H, Ozawa A, Hayashi T, (2006), Cytogenet. Genome Res., 112(1-2):121-125.

Youngerman, SM, Saxton, AM, Oliver, SP, and Pighetti, GM, (2004) J. Dairy ScL 87:2442-2448. ^'

Patent Literature (Dairy)