The specific aims of the project (Z01 HG000200) are: 1) to develop methods of genetic analysis that address statistical problems involving the detection of trait loci and new marker technologies (e.g., SNPs); 2) to use computer simulation to investigate the statistical properties of existing and newly developed methods for the genetic analysis of quantitative traits, and to apply insights gained from these simulations to ongoing collaborative studies, 3) to identify genetic effects underlying quantitative traits using statistical genetic analysis in collaborative family studies; and 4) to continue the development and dissemination of the Genometric Analysis Simulation Package (G.A.S.P.). This project now includes Z01 HG00096, Z01 HG000110, Z01HG000125 and Z01 GH200320. Theoretical work during the past year focused on linear regression-based tests of associations for quantitative traits and alternatives to theoretically based determination of type I error rates. The linear model for the Regression of Offspring on Mid-Parent (ROMP) method is a hybrid of the traditional regression of offspring on mid-parent model used to estimate heritability and the ANOVA model used to test for association. ROMP was reformulated within the context of collapsibility in linear regression models in order to determine the standard error of the difference between two regressions. In addition, ROMP was extended to include missing values and applications to haplotypes. Simulations with G.A.S.P were used to evaluate the statistical properties of the method. With respect to developing alternatives to theoretically based determination of type I error rates, an empiric method, the Applied Pseudo-Trait (APT) method, was applied to data from the Framingham Heart Study. Rather than relying on a series of assumptions that may or may not be met in the underlying data, APT uses the actual locations, number of alleles and allelic frequencies of the actual marker data, the distribution of recombination, the family structures present in the data, and any genotyping errors that may be present in the data. The method is similar in spirit to the simulation and permutation methods, but it does not require computer simulation or many repetitions of the genomic screen. Computer simulations. In addition to determining statistical properties of the extensions to ROMP, G.A.S.P. was used to evaluate the statistical properties of a newly developed mathematical algorithm for prediction of complex phenotypes. Numerous computer simulations were performed with the algorithm, using different genetic models and various conditions as part of an effort to determine the feasibility of the method. Collaborative projects include: Genetic analysis of diabetes susceptibility. Genetic linkage and association analyses were conducted with qualitative and quantitative phenotypes in a sample of Japanese Americans. Familial idiopathic scoliosis. A large number of SNPs in candidate regions were analyzed for linkage using SIBPAL and for association using ASSOC and FBAT. During the past year genetic analysis has focused on candidate genes in a small group of families with kyphoscoliosis, and for candidate regions on chromosomes 9 and 16 in a larger group of families with idiopathic scoliosis. Genetic analysis of drug treatment response in the Sequenced Treatment Alternatives for Depression (STAR*D). Non-parametric tests for association of remitter status and responder status with each of 768 SNPs were performed on initial and replication samples of available cases. One SNP, significant in both the initial and replication sample, was reexamined in a larger sample including additional cases, resulting in a p-value of approximately .000002. Genetics of type II diabetes in India. The goal of the project is to recruit 30 large multiplex families to study the genetics of type II diabetes in an Indian population. In the past year, DIR funds were used to initiate a contract with collaborators at the Madras Diabetes Research Foundation for family recruitment, and statistical analysis for tests of association of SNPs in candidate genes were performed. Eighteen families have been successfully recruited and two manuscripts have been submitted. Genetics of asthma in an African American population. This is a collaborative project with investigators at Johns Hopkins University to identify asthma susceptibility loci on chromosome 11q. A fine-mapping panel of 609 SNPs in the previously identified linkage region was analyzed using single-SNP and sliding-windows of haplotype tests of association. A software tool, Graphical Assessment of Sliding P-values (GrASP), was developed to simplify the presentation and visualization of results in these types of analyses. Five candidate regions showed evidence for association with asthma.
Showing the most recent 10 out of 23 publications
