More than a century after the rediscovery of Mendel, the genetic basis of complex and quantitative traits resists generalization. Basic questions remain unanswered, including the number of loci that underlie variation in heritable phenotypes, the distribution of their effect sizes, their molecular natures and mechanisms of action and interaction, and their dependence on environmental variables. These questions are at the centre of pressing issues in medical and agricultural genetics, as well as in basic evolutionary biology, in which the outstanding unresolved question concerns the forces that create, maintain and sort heritable phenotypic variation. Now, an emerging approach, genetic mapping of genome-wide gene expression (BOX 1), is beginning to provide the req- uisite empirical data to address these questions. Since the first empirical linkage study of global transcript levels was published in 2002 (REF. 1), many general principles have been established and represent solid ground on which further work can build. Although small-scale studies of the genetics of gene expression have a long and rich history (BOX 2), modern large-scale studies owe their existence to the develop- ment of microarray technology in the mid-1990s. Microarrays were first applied to the study of genetic variation in 2000. They revealed that gene expression differs between strains in both yeast and mice2,3 and that such differences segregate in crosses4,5. Subsequent studies documented abundant heritable variation in gene expression in Drosophila melanogaster 6 and kil- lifish7. By the time Jansen and Nap8 proposed genetic mapping of genome-wide gene expression, such work was well underway in several research groups, and the first empirical study mapping global gene expression in a yeast cross appeared early the following year1. Since then, further studies have documented heritable variation in genome-wide gene expression in more than a dozen species and have mapped the loci for many expression traits in yeast, mice, maize, humans, rats, Eucalyptus and Arabidopsis thaliana9���21. This diversity of model systems promises to reveal important connections between genome-wide gene expression and features of population biology ��� population sizes, breeding systems, demographic histories and patterns of natural selection. Differences can already be seen among spe- cies: most species show ubiquitous heritable variation in expression, whereas the malarial parasite Plasmodium falciparum shows remarkably little22. The abundance of a transcript is a quantitative trait and, like all such traits, its inheritance can be described using the classical methods of biometrical genetics and its genetic basis can be discovered using linkage and asso- ciation mapping. However, transcript abundance is in many ways an extraordinary phenotype, with special attributes that confer particular importance on an understanding of its genetics. The primary transforma- tive potential of genome-wide gene expression genetics is the sheer number of traits ��� thousands ��� that can be assayed simultaneously. Whereas studies of one or a few traits offer only anecdotal examples of the underly- ing genetic architectures, studying thousands of traits allows a detailed description of the distribution over the landscape of all possible architectures. Individual traits are typically preselected on the basis of their phenotypic divergence or biological interest, whereas genome-wide expression studies provide data on a large and unbiased set of traits. The radical increase in the number of traits accessible to study has raised new challenges to analysis and interpretation, and genome-wide genetic mapping of gene expression has consequently become a central proving ground for new statistical genetics techniques23. Another special feature of transcript abundance as a phenotype is that it represents the phenotype most immediately connected to DNA sequence variation Lewis-Sigler Institute for Integrative Genomics and Department of Ecology and Evolutionary Biology, Princeton University, Princeton, New Jersey 08544, USA. Correspondence to L.K. e-mail: leonid@genomics. princeton.edu doi:10.1038/nrg1964 Complex and quantitative traits Phenotypes that are shaped by multiple and possibly interacting genetic and environmental factors. Quantitative traits (as distinguished from discrete traits) are measured on continuous scales. Effect size The magnitude of contribution of a locus to variation in a phenotype. Genetics of global gene expression Matthew V. Rockman and Leonid Kruglyak Abstract | A new field of genetic analysis of global gene expression has emerged in recent years, driven by the realization that traditional techniques of linkage and association analysis can be applied to thousands of transcript levels measured by microarrays. Genetic dissection of transcript abundance has shed light on the architecture of quantitative traits, provided a new approach for connecting DNA sequence variation with phenotypic variation, and improved our understanding of transcriptional regulation and regulatory variation. REVIEWS 862 | NOVEMBER 2006 | VOLUME 7 www.nature.com/reviews/genetics �� 2006 Nature Publishing Group

��� Correlate genotype with transcript abundance Fold change (log 2 ) Daughter-cell specificity Genome location Chr 2 locus: inherit BY Laboratory (BY) Wine (RM)