Goncalo Abecasis is a Professor of Biostatistics. He received his D.Phil. in Human Genetics from the University of Oxford in 2001 and joined the faculty at the University of Michigan in the same year. Dr. Abecasis' research focuses on the development of statistical tools for the identification and study of genetic variants important in human disease. Software developed by Dr. Abecasis at the University of Michigan is used in several hundred gene-mapping projects around the world. • The focus of my research is the identification and characterization of genes determining human variation and disease. In particular, I have focused on developing analytical methods and statistical tools that will facilitate the mapping of complex traits and allow geneticists to realize the benefits of new high-throughput technologies in the lab. Much of my research has focused on the use of linkage disequilibrium in the mapping of complex disease susceptibility genes. Linkage disequilibrium based mapping strategies search for short segments of ancestral chromosomes shared among present day individuals. High-throughput technologies allow fine-scale characterization of genetic variation and have made these searches possible. • Quantitative Trait Mapping: I have proposed a test of association in general pedigrees that uses all available information on each individual's ancestry to separate population substructure from linkage disequilibrium (Abecasis et al. 2000a; Abecasis et al. 2000b). Using this model, I have investigated the role of various angiotensin-converting enzyme (ACE) gene polymorphisms on circulating ACE levels in blood. The ACE gene, and an Alu insertion-deletion polymorphism it contains, has generated intense interest because of its possible role in hypertension and heart disease. I evaluated the effect of variants in the gene in Jamaican and British samples. The data show that the I/D polymorphism is not the major functional variant in this gene and suggest a number of nearby variants as candidates (McKenzie et al. 2001). • Characterizing Genetic Variation in Humans: The success of mapping strategies based on allelic association and linkage disequilibrium is going to depend, to a large part, on the extent of linkage disequilibrium in the populations being studied. The observation that linkage disequilibrium extends for long distances and that most chromosomes are mosaics of relatively common short haplotypes is a critical underpinning of the NIH haplotype map initiative. I have developed graphical tools that allow scientists to explore and summarize patterns of disequilibrium in regions of interest (Abecasis and Cookson 2000). In addition I have provided some of the first detailed, large-scale descriptions of linkage disequilibrium in the genome (Moffatt et al. 2000; Abecasis et al. 2001d). My observations demonstrated that linkage disequilibrium extends further than theoretical predictions in much of the genome and suggested that it is organized in cluster-like structures. • Computational Tools: A significant challenge in the fine-scale characterization of genomic variation is the sheer amount of data that must be considered. Most current analytical methods focus on the analysis of relatively small numbers of markers (10 to 50) and cannot handle the large numbers of linked variants (1000) that can be assayed using high-throughput technologies. I have described and implemented novel, very efficient algorithms that exploit the structure of genetic data in pedigrees (Abecasis et al. 2002) and populations (Abecasis et al. 2001c) to provide extremely fast solutions to traditional problems. Specifically, I showed that allowing for the small number of recombination events between consecutive markers could greatly simplify likelihood calculations in dense genetic maps. I also described a representation of gene flow within pedigrees that uses sparse binary trees to summarize redundancies in genetic data and further simplifies likelihood calculations (Abecasis et al. 2002). Separately, I showed how population and family data could be combined in haplotype estimation and proposed a divide-and-conquer algorithm for tackling longer haplotypes (Abecasis et al. 2001c). Together, these methods can also handle very large datasets and enable construction of the first chromosome wide linkage disequilibrium map (Dawson et al. 2002). • Practical Challenges: A final aspect of my research concerns the challenges of handling real, sometimes imperfect data. I have investigated the effects of genotyping error on quantitative trait analyses (Abecasis et al. 2001a). In addition, I have proposed strategies for genotype error detection (Abecasis et al. 2002) and identifying misspecified relationships (Abecasis et al. 2001b). The motivation for my research comes from the day-to-day challenges encountered in the gene-mapping projects where I am involved. This research has been motivated by collaboration with researchers seeking to identify genes predisposing to diabetes (Dr. Michael Boehnke, University of Michigan), glaucoma and age-related macular degeneration (Drs. Julia Richard and Anand Swaroop, University of Michigan), schizophrenia (Dr. Maria Karayiorgou, Rockefeller University) and aging related traits (David Schlessinger, National Institutes of Aging). • The Future: I believe that further successes in the area of complex disease gene identification will require even better computational tools and further improved statistical methods that are able to tackle the large single nucleotide polymorphism datasets (SNP) now being collected. Some of the challenges for these tools include haplotype estimation, evaluation of linkage evidence through simulation, and improved detection and modeling of possible genotype and pedigree errors.