SNPs occur about every 700 base pairs in Bos taurus and every 300 base pairs in Bos indicus cattle (The Bovine HapMap Consortium, 2009), which means there is more genetic variation in Bos indicus cattle. Thus, there are around 4 million SNPs in the Bos taurus genome. If time and money were not limiting, one could sequence the genome of every animal and thus know their exact genetic makeup. This would be less useful for selecting animals for breeding purposes than one might surmise, mostly because we simply do not know what most genes do, nor do we know most of the genes that affect a given trait such as milk production, growth rate, feed efficiency, disease resistance, etc., or how alleles of genes differ in their effects on phenotypes. The relationship between alleles and phenotype, however, can be determined in a number of indirect ways, and animal breeders have exploited such relationships for centuries. For example, large animals tend to have large offspring and thus usually have the alleles for large size.
These approaches have become quite sophisticated with EPDs (expected progeny differences) in beef cattle and PTAs (predicted transmitting abilities) of dairy cattle. With rare exceptions, these very effective tools for animal breeding do not use any DNA sequence information, only pedigrees and phenotypes. Applying these techniques does, however, alter the allelic structure of populations. Obvious examples are the different breeds of cattle and changes within breeds such as size changes in beef breeds and increased milk production in dairy breeds. It turns out that genomic selection using SNPs is just another method of matching unknown alleles with phenotypes, but now DNA sequence information is used in the process.
Even though the vast majority are unknown, hundreds of specific genes and their alleles that affect phenotypes are known. Most of these known genes are in two categories: all or none dominance effects and additive gene action effects. Familiar examples of the former are coat color, polled/horned, and certain genetic abnormalities such as mulefoot, bovine leukocyte adhesion deficiency (BLAD), and curly calf syndrome (arthrogryposis multiplex). These dominant effect genes have the same phenotype if the dominant allele is inherited from both parents or only one parent. For example, one or two copies of the polled allele give polled offspring, and one copy of black and one copy of the recessive red allele or two copies of black give black offspring; two copies of the recessive red allele result in red offspring. With genetic diseases, one similarly needs two copies (one from each parent) of the defective (recessive) allele to get the disease. Note that the exact DNA sequence is known for most of these genes/alleles, and that information is the basis of the available genetic tests.
The second category for which alleles are known and can be selected for directly is illustrated by meat tenderness genes such as calpain. In this case, one copy of the desirable allele (heterozygous) is intermediate in tenderness to two copies (homozygous). One submits blood or other body tissues to companies that analyze the DNA and report on the alleles for tenderness. In addition to those that are known and selected for, there are hundreds of genes that affect meat tenderness. Although presence of the tenderness alleles just described does not ensure tender meat, it increases the chances that meat will be tender. Clearly, environment greatly affects meat tenderness; for example, how animals are fed, and how the meat is aged and cooked. However, some genetically inferior meat will not be tender even if the environment is optimal. These same principles apply to most economically important traits, such as protein composition of milk.
As just mentioned, many genes/alleles affect most production traits, but exactly which genes/alleles are responsible is largely unknown. However, there are animals in the population with different phenotypes, such as producing milk with high or low percent butterfat. DNA from such cows can be evaluated to determine how it differs. There are a variety of ways of doing this. In most cases, the DNA differences found that correlate to the different phenotypes are not the different alleles themselves (see Fig. 1). These are called markers, and for that animal and its close relatives, for breeding purposes, knowing the marker is just as useful as knowing the allele; if the correct marker is selected for, the desirable allele (e.g., top versus bottom in Fig. 1) is also selected. SNPs are examples of such markers, and in sufficient, appropriately spaced numbers, can serve as markers for essentially all alleles of all genes in an animal.