The inheritance of many traits of evolutionary, medical, and agricultural importance is polygenic in nature. The observation that complex traits have measurable genetic components, despite apparent stabilizing selection which should erode this variation over time, is a paradox in evolutionary biology. A number of theoretical models address the question of what maintains quantitative genetic variation, yet all models rely on assumptions regarding underlying genetic parameters. This is problematic, as despite great recent advances in our understanding of 'Mendelian' traits, we still know little about the molecular genetic basis of complex traits. My research uses the powerful genetic and molecular tools of model systems to elucidate the molecular genetic basis of polygenic characters. Approaches range from the use of cutting edge modern genomic tools, to the use of computer simulations based on population genetics theory, to purely statistical approaches that make maximal use of experimental data. The information obtained from these model systems can be used to address the basis of polygenic variation in humans. Current projects in the lab include:

		Using linkage disequilibrium to dissect complex traits in Drosophila
		Theoretical models for dissecting complex traits
		Using high density arrays to study adaptation
		The statistics of high density arrays
		BioInformatics
		The genetic basis of wing pattern variation in butterflies

   Using linkage disequilibrium to dissect complex traits in Drosophila   

By sequencing a number of alleles of neurogenic candidate genes, we have identified every common Single Nucleotide Polymorphism (SNP) in these regions that could potentially contribute to standing variation in bristle number. We are now using identified SNPs to carry out large scale association studies in large samples of wild caught flies (about 1,000 wild caught males and females are being genotyped). These studies will be important as a validation of the association study methods that human geneticists plan to carry out (Long et al. 1997, Long et al. 2000). In this regard, we have already typed a number of SNPs in our large wild sample and found patterns of association between SNPs and bristle number to be different from those seen in previous laboratory studies (Macdonald and Long 2004). It is also becoming apparent that SNPs in cis-regulatory regions may be more likely to contribute to variation in complex traits than SNPs that encode amino acid changes (Genissel et al. 2004). In addition to the important scientific question this work will address, we are using Drosophila melanogaster bristle number as a system for assessing the utility of various high throughput SNP typing technologies. Stuart Macdonald is working on these issues.

   Theoretical models for dissecting complex traits   

There is currently a great deal of interest in estimating the power of association studies in humans using rigorous population genetics models of evolution (i.e., the gene coalescent). We have carried out large scale simulations based on the gene coalescent theory of population genetics to assess the power of association studies to detect gene regions contributing to complex traits (Long and Langley 1999). Current studies are focused on extending the simulations to examine: additional statistical tests, more realistic models of human population structure and demography, and the case control design. In theory the coalescent machinery can deal with the complexities of human history and demography, but as of yet the coalescent machinery has not been able to generate human alleles that completely "match" samples seen in surveys of human variation. The ability to generate such samples using a mechanistic based simulation machine will be important if we wish to assess the utility of various methods to identify disease genes using linkage disequilibrium.

   Using high density arrays to study adaptation   

We have begun to use high density arrays to examine the genetic architecture of thermal adaptation in experimentally evolved populations of E. coli (Riehle et al. 2001). We have found that adaptation is quite replicable at the molecular level and involves a region that is tandemly duplicated in 3 out of 6 experimental replicates. The duplicated regions contain candidate genes important in adaptation, that were not a priori strong candidates for adaptation to high temperature. Nancy Aguilar-Roca is currently working on other lab-evolved E. coli lines. The combination of experimentally evolved lines and modern genomic tools are a very potent combination and is likely to result in great insights into the molecular nature of adaptation. We are using similar approaches to examine expression changes in butterfly wings in order to identify genes expressed in eyespots but not the surrounding non-eyespot tissue. We hope to be able to identify a set of genes important in adaptation that were not strong a priori candidates. In both projects, the discovery of genes and pathways important in adaptation raise the exciting possibility of interactions between physiologists, molecular biologists, and experimental evolutionists.

   The statistics of high-density arrays   

High density arrays represent a powerful new tool for examining gene expression changes at a genome wide scale. The problem is that the technology has far outpaced the availability of tools for the proper identification of genes showing changes in expression. The biggest problem with arrays is that a large number of genes are often examined for changes in expression, but the differences in expression for any given gene are obtained from experiments with little replication. We developed statistical methods that employ the t-test, but incorporate a Bayesian prior into estimates of the within-treatment variance. This approach results in improved statistical inference with minimal levels of replication (Arfin et al. 2000, Baldi and Long 2001, Long et al. 2001). We implemented this approach into a set of web-based statistical tools (Cyber-T) that currently get over 2000 "hits" per month from all over the world. The approaches we describe can be extended to other experimental situations such as SNP typing on microarrays and more complex expression microarray experimental designs.

   BioInformatics   

Our lab is part of the NIH-funded UCI Biomedical Informatics Training Program (BIT) which provides in-depth training in bioinformatics for graduate students and postdoctoral fellows of computational and life sciences. BIT graduate students and post-docs receive training in machine learning, data mining, and bioinformatics approaches to problems in molecular structure prediction and determination, computational chemistry, comparative and functional genomics, and dynamic modeling of biological systems. The Long-lab is mostly interested in genomics questions and we do a great deal of programing in Perl, MySQL, C, and R (a personal favorite). Nancy Aguilar-Roca (post-doc), and Jonathan Gruber (graduate student) are both part of the BIT.

   The genetic architecture of wing pattern variation in butterflies   

We have cloned a gene (Distal-less) believed to be important in butterfly eyespot formation, and asked if naturally occurring variation in that gene contributes to standing variation in eyespot size in Bicyclus anynana butterflies. We found that Distal-less does contribute to inter-individual variation in this trait (Beldade et al. 2002). This observation has widespread implications for understanding adaptation, speciation, and how organisms interact with their environment. We are continuing by building a linkage map in Bicyclus anynana composed of genes expressed in the wing imaginal discs. We have already identified approximately 4,500 wing UniGenes (collaboration with Stephen Rudd in Finland), and plan to begin adding them to a linkage map in 2005. We hope to be able to use the genome sequence of another Lepidopteran, the silkmoth Bombyx mori, to identify genomic blocks shared between the two species. Patricia Beldade is working on this project.