The analysis of complex genetic traits, including mapping and identification of causative genes, has long been an enigma of genetic biology, whether in the animal sciences or in medical sciences. Traits of agricultural interest and traits of medical interest are often under the influence of both environmental factors and multiple genes, each with modest contributions to the total variance in the trait. Although the number of known mutations underlying complex traits is still relatively small, advances in genomics have greatly enhanced traditional pathways to their analysis and gene mining. The candidate gene approach, linkage analysis, and association studies are all significantly more powerful with recent advances in genome mapping, sequencing, and analysis of individual variation. Avenues to gene discovery are discussed with emphasis on genome wide association studies (GWAS) and the use of single nucleotide polymorphisms (SNPs) as revealed by increasingly powerful commercially available microarrays.

In addition to providing a complete human genome sequence, along with cataloging of a wealth of naturally occurring genomic variation, the Human Genome Project spawned genome projects for a large number and wide variety of other organisms, including most of the common domestic animal species. Single nucleotide polymorphisms (SNPs) are the most common form of genetic variation in most species, and animal genomes appear to be replete with these, small insertions and deletions (indels), and copy number variants (CNVs). sbSNP, a database for short genetic variations, lists more than 60 million entries for humans, 9.5 million for cattle, more than 3 million for dogs and chickens, 1 million for horses, and 0.5 million for pigs as of May 16, 2012 (http://www.ncbi.nlm.nih.gov/projects/SNP/). Among the many valuable scientific resources to come from these genome projects are microarrays which permit the screening of individuals for hundreds of thousands of genomic variants in a single assay. These so-called SNP chips and CNV arrays are commercially available for humans and the common domestic species. One chip for human variants, for example, contains markers for more than 900,000 SNPs and 900,000 CNVs. A high-density cattle SNP chip contains almost 800,000 markers.

The most intuitive approach to identification of specific genes and mutations responsible for phenotypes is the candidate gene approach. Prior knowledge of the function of a gene, coupled with some understanding of the physiological basis of the trait, has been a historical first approach to the discovery of gene/trait relationships. This logic has been exploited effectively for a number of monogenic human disease genes as well as some important phenotypes in domestic animals. As pointed out by Georges,³ however, the candidate gene literature is replete with nonreproducible results. Georges³ makes a rational plea for higher standards in candidate gene analysis, including (a) more rigid significance thresholds, (b) thorough characterization of the haplotype diversity and structure in the study population, (c) replication of positive associations in an independent cohort, and (d) resequencing individuals from functionally distinct haplotypes to find causative SNPs. As is usually the case, Michel Georges’ advice should not go unheeded.

As discussed above and demonstrated in Table 1, a large number of QTLs have been mapped in domestic animal species, especially those of agronomic significance but only a few QTMs have been discovered. The positional candidate approach to gene discovery employs linkage mapping to map a trait, and then searching for most likely candidates among the genes known to lie in this region. While this has been effective in both human and animal genetics for monogenic traits, it has had limited success for QTLs. One reason is that the function of most genes is still unknown. Another is that mapping resolution is usually severely limited. A typical confidence interval in a QTL mapping study might be 20 or more centimorgans, corresponding to roughly 20 million base pairs and potentially 200 or more genes. As discussed by Georges,³ several factors limit mapping resolution. These are (a) marker density (which is no longer a problem in most species with the availability of SNP technology), (b) crossover density because only recombinants provide map information, (c) QTL detectance, the accuracy of inferring the genotype/phenotype relationship at loci with small contributions to the total phenotype, and (d) the molecular architecture of the QTL, that is, linked QTMs. A classic example of success with linkage analysis is the discovery by Van Laere et al. of the causative mutation for muscle growth and fat deposition in pigs.⁴ Identical by Descent (IBD) mapping revealed a noncoding QTN in an intron of the IGF2 gene that abates binding of a repressor and results in higher IGF2 expression in skeletal and heart muscle. This was the first noncoding QTN reported in any species. Most successful mapping studies have enhanced crossover density, often by utilizing historical recombinants and the knowledge of linkage disequilibrium (LD) and haplotype structure in the targeted species and study populations.

Complex traits remain an enigma in both human and animal genetics. While large numbers of QTLs underlying complex traits are known, the causative mutations remain elusive. However, the recent advances in genome sequencing projects, particularly the discovery of vast numbers of SNPs and their availability in affordable arrays, have opened the door to a new era in gene discovery. These tools, especially when used in GWAS, offer more than a faint hope of discovery of large numbers of genes contributing to complex traits in the very near future. The promise is especially important to studies in host resistance to infectious diseases, an important new frontier in animal and human health.

The authors declare no conflicts of interest.