Chad P. GarnerAssociate Professor, Epidemiology |
|
|
Research Interests |
Statistical and Population Genetics | |
| URL | Dept. of Epidemiology - School of Medicine | |
| Appointments |
Assoc. Professor, UC Irvine (2008 - present) Asst. Professor, UC Irvine (2002 - 2008) Postdoctoral Fellow, Department of Integrative Biology, UC Berkeley (1999-2002) |
|
|
Research Abstract |
My broad research interest is the statistical genetic analysis of complex human traits. My research is focused in two specific areas. The first is using quantitative genetic analyses to discover the genetics of fetal hemoglobin expression and F cell levels in adults. The second focuses on methodological and applied problems related to population-based studies of clinical disease outcomes. Furthermore, I contribute to a considerable amount of research that falls outside of my specific interests through collaborations with colleagues at my institution and outside institutions. As a scientist in the statistical genetics field, I rely on collaborations with colleagues in the epidemiological, clinical and molecular sciences to provide clinical and trait information, and to generate genetic data so that my research program has access to rich datasets for methodological development and analysis. My research in fetal hemoglobin (HbF) and F cell genetics began in 1998. HbF is largely replaced by adult hemoglobin shortly after birth with small amounts of the fetal molecule continuing to be produced through adulthood (approximately 2% of total hemoglobin on average). The residual expression of the HbF shows considerable variation within populations. HbF expression is restricted to a subset of erythrocytes called F cells and F cell levels are highly correlated with fetal hemoglobin expression and can be measured more reliably, thus F cell levels are the phenotype that is generally studied. In 1996, a major quantitative trait locus (QTL) for adult levels of fetal hemoglobin expression was mapped to chromosome 6q by genome-wide linkage analysis in a large Asian-Indian kindred. I applied a haplotype-based genetic mapping approach to new data in the Indian kindred and localized the QTL to chromosome 6q23 (Garner et al. 1998, AJHG). Identification of the specific gene influencing F cells on chromosome 6q23 has required a large positional cloning effort with the outcome being recently published (Thein et al. 2007. PNAS) In 2000, I utilized data from a large sample of British monozygotic (MZ) and dizygotic (DZ) twins to estimate the heritability of F cells to be 89% (Garner et al. 2000, Blood). The ultimate goal of this research is to identify all the genetic factors contributing to the heritability. The very high heritability of this quantitative trait, combined with access to the large Asian-Indian kindred and U.K. twin samples, make it an ideal model for quantitative genetic studies. The clinical expertise, sample resources and laboratory work provided by Professor Swee Lay Thein (King’s Hospital, London), and the U.K. Twin resource provided by Professor Tim Spector (St. Thomas’ Hospital, London) made this research possible. I am an Honorary Senior Lecturer at King’s College, allowing me to mentor and advise graduate students and postdocs working on this research in Professor Thein’s laboratory. A genetic polymorphism near the fetal hemoglobin gene in the beta-globin complex on chromosome 11p15 (termed the XmnI-G? polymorphism) is known to be associated with HbF and F cell levels. In 2002, I carried out an experiment to determine if unknown QTL were acting on F cell levels through a genetic interaction with the XmnI-G?. Genome-wide analysis of the Asian-Indian kindred identified a region on chromosome 8q that showed strong statistical evidence for a QTL that was influencing F cell levels through an interaction with XmnI-G? (Garner et al. 2002, AJHG). I replicated this finding in the U.K. twin sample (Garner et al. 2004, Blood) and characterized the statistical interaction in an extended U.K. twin sample (Garner et al. 2005, Ann Hum Genet). The chromosome 8q QTL shows strong evidence for statistical interaction with the beta-globin polymorphism but the candidate region for this QTL encompasses approximately 10 Mb of DNA and contains numerous candidate genes. I have recently completed an association study across half of the candidate interval and found evidence for association at the DEPDC2 gene, however 5 Mb of sequence remain unexplored. In late 2006, investigators in London, Oxford and Paris, and I designed a genome-wide association study of F cell levels using the selected genotyping approach. Theoretical work has shown that studying individuals selected from the extreme ends of a quantitative trait distribution can provide increased power for association studies. This study design had not been used for a genome-wide association study. We designed a study that included an initial sample of 179 individuals with roughly equal numbers selected from above and below the 95% and 5% percentiles of the trait distribution. The statistical genetic analysis for this study was done by my group at UC Irvine. The analysis identified the two previously known major QTLs for F cell levels on chromosome 6q23 and 11p15, as well as a new QTL in the BCL11A gene on chromosome 2p15. Associations with all three loci were replicated and the three combined were estimated to account for 45% of the variance in F cell levels. The results of this study are in press (Menzel et al. 2007, Nat Genet). With 89% heritability, approximately half of the genetic variation in the European population remains unexplained. Although the F cell QTL on chromosome 8q was detected using the full U.K. twin sample, neither a main or interaction effect of the locus was identified in the genome-wide association study. The lack of a statistically significant association could be due to the small sample size used in the genome-wide study (i.e., lack of statistical power) or because the 8q QTL does not have any detectable effect on extreme high or low trait values. A genome-wide set of single nucleotide polymorphisms (SNPs) has recently been genotyped on a large unselected set of the U.K. dizygotic twin samples with measured F cell levels. These data should become available to my group within the next several months and will be essential for localizing the chromosome 8q QTL as well as for identifying the remaining unknown QTL. Our work on HbF and F cells has strong clinical implications because of the ameliorating effect of high HbF expression on ?-thalassemia and sickle cell disease. An understanding of the genetic factors and mechanisms underlying adult expression of fetal hemoglobin could inform the development of pharmaceuticals to reactivate its expression. Clearly, an important next step will be to study individuals of African descent. Over the course of my career in human genetics, there has been a shift from family-based genetic linkage studies to population-based association studies and my research has followed this movement. In 2002, I developed a statistical method for studying linkage disequilibrium and the age of disease mutations (Garner and Slatkin, 2002, Theor Pop Biol). The method was used in a study of the effects of natural selection due to malaria on the G6PD gene (Saunders et al., 2005, Genetics). Although the method works well for estimating the age or location of highly penetrant disease mutations, it is not appropriate for the study of common, low penetrant genetic variants that are believed to play a large role in complex diseases. I turned my attention to research questions that were more appropriate to the problems surrounding the study of complex diseases. In 2003, large numbers of single nucleotide polymorphisms (SNPs) were being discovered and described in public databases. The question of which SNPs would be best for genetic association studies was a topic of great interest. Based on extensive population genetic simulations resulting in distributions of linkage disequilibrium, I described the conditions under which SNPs would be most informative for association studies (Garner and Slatkin, 2003, Genet Epidemiol). As more SNPs were discovered, the problem of finding informative SNPs was replaced by the need for data reduction, or selecting a subset of SNPs from a much larger set of SNPs in order to tag the common genetic variation in a specific region and reduce redundancy (and genotyping costs). In an NIH/NCI funded project called Hereditary Breast Cancer: Genetics and Molecular Studies, I selected 1536 such tagging SNPs in 150 candidate breast cancer genes that were then genotyped in a large population-based case-control sample. The primary analysis of the SNPs has been completed and a manuscript describing the results is currently under review. As my research became focused on case-control association studies of complex disease, I became interested in the often overlooked aspects of the study design. The most obvious way in which an investigator can maximize their chances of detecting an association is by studying the largest possible samples of cases and controls. While it was known that cases must be accurately diagnosed and that controls must be matched to cases on population of origin, there has been a range of opinions regarding the extent to which controls need to be characterized and matched to cases with regards to the outcome and known risk factors. The manner in which controls are ascertained can have a profound effect on the cost of a study, but little was known about the impact on statistical power. In order to address this problem, I assessed the effect of random control selection on statistical power and showed that the cost was minimal (less than 5%) for diseases having a prevalence of less than 5% (Garner, 2006, Human Hered). I have used these findings in a recently submitted grant proposal for a genome-wide association study of celiac disease that will partially rely on controls ascertained for other studies. Indeed, the Wellcome Trust Case Control Consortium utilized a common, population-matched set of 3,000 controls for genome-wide association studies of seven diseases published in 2007. Genome-wide association (GWA) studies present many new challenges to statistical genetics. Much research has focused on the interpretation of statistical significance in the context of the hundreds of thousands of tests involved in GWA studies. In addition to a p-value, many of the statistical tests of association also provide an estimate of the effect size, e.g., an odds ratio for case-control studies. These estimates can be highly biased as a result of the large number of tests carried out in a genome-wide study and overestimate the strength of the association considerably. Although this bias had been formally described for genome-wide linkage studies, I provided the first such formal description of the bias for genome-wide association studies (Garner, 2007, Genet Epidemiol). It is important for investigators to be aware of the bias not only in the interpretation of their results but also in the design of their replication studies. In addition to the genome-wide association studies of F cells described above, my group is carrying out statistical genetic analysis for a genome-wide association study of iron deficiency. I have just submitted a proposal to NIH/NIDDK to do a genome-wide association study of celiac disease. This study will present new statistical genetic challenges because the sample is a collection of both simple and extended family structures. Methods for powerful meta- or combined analysis of the existing celiac genome-wide case-control association study with the results of the family-based study will need to be developed. I am also planning a study with a design similar to that used for F cells to search for genes controlling platelet and white blood cell counts in collaboration with a large blood bank. This GWA study will complement a genome-wide linkage analysis of platelets published in 2006 (Garner et al., 2006, Euro J Hum Genet) Recently I have been working on the analytical problems and possibilities that will arise when DNA sequencing becomes the standard approach to assay genetic variation. The influence of rare variation on complex disease has largely been overlooked, due to the inability to assess the relationship under the current genotyping and analytical paradigm. The few studies that have studied rare variation have sequenced a small number of candidate genes and utilized standard statistical approaches for assessing differences in cumulative allele counts between cases and controls. I am working on statistical methods to study rare variation that use the case-control study design. I recently proposed using a classical population genetics theory called the Ewens sampling formula to identify regions showing patterns of rare variation that were consistent with disease involvement. The method performs better than the approach based on cumulative allele counts under certain conditions (this work is currently under review for publication). My future research in this area anticipates the time when whole-genome sequencing of case and control samples will be a realistic study design. I have began experimenting with hidden Markov models for scanning the genome for regions that show an increased concentration of rare genetic variants among disease cases relative to controls. A major focus of my research over the next year will be to develop and evaluate methods of genome-wide analysis of rare variation, and to secure funding for this line of research. |
|
| Publications | Menzel S, Garner C, Gut I, Matsuda F, Yamaguchi M, Heath S, Foglio M, Zelenika D, Boland A, Rooks H,Spector TD, Farrall M, Lathrop M and Thein SL (to appear Oct. 2007) Genome-wide association maps a QTL influencing F cell production to a Zn-finger protein on chromosome 2p15. Nature Genetics | |
| Garner C, Ding YC, Steele L, Book L, Carmi R, Leiferman K, Zone JJ and S Neuhausen (2007) Genome-wide linkage analysis of 160 families with celiac disease. Genes and Immunity 8:108-114 | ||
| Garner C (2007) Upward bias in odds ratio estimates from genome-wide association studies. Genetic Epidemiology 31:288-295 | ||
| Thein SL, Menzel S, Peng X, Best S, Jiang J, Close J, Silver N, Gerovasilli A, Ping C, Yamaguchi M, Wahlberg K, Ulug P, Spector TD, Garner C, Matsuda F, Farrall M and Lathrop M (2007) Intergenic variants of HBS1L-MYB are responsible for a major quantitative trait locus on chromosome 6q23 influencing HbF levels in adults. Proceedings of the National Academy of Sciences 104:11346-11351 | ||
| Garner C, Best S, Menzel S, Rooks H, Spector TD and SL Thein (2006) Two candidate genes for low platelet count identified by genome-wide linkage analysis in an Asian-Indian kindred: glycoprotein IX and thrombopoietin. European Journal of Human Genetics 14:101-108 | ||
| Garner C (2006) The use of random controls in genetic association studies. Human Heredity 61:22-26 | ||
| Lai MI, Jiang J, Silver N, Best S, Menzel S, Mijovic A, Garner C, Weiss MJ and SL Thein (2006) ASHP is a quantitative trait gene that modifies the phenotype of ? thalassemia. British Journal of Haematology 133:675-682 | ||
|
Professional Societies |
American Society of Human Genetics American Association for the Advancement of Science |
|
| Graduate Programs |
Epidemiology |
|
| Research Center | Genetic Epidemiology Research Institute - Dept. of Epidemiology | |
| Link to this profile | http://www.faculty.uci.edu/profile.cfm?faculty_id=5191 | |
| Last updated | 11/05/2008 | |