Development of statistical genetics methodology

点击翻译按钮获取中文摘要

A major project of this section is the development of new statistical genetics methodology as prompted by the needs of our applied studies and the testing and comparison of novel and existing statistical methods. The project to develop propensity scores in linkage analyses as a method for inclusion of covariate effects has been continued in conjunction with Dr. Betty Doan. This method appears promising in that it is generally more powerful than including the covariates directly into the model, and does not have strongly inflated Type I error rates. We are currently using computer simulation studies to examine factors that affect the performance of this method and are applying it to Dr. Bailey-Wilsons lung cancer data. We are completed work on establishing a p-value threshold for genome wide association studies using the number of independent SNPs and blocks within the HapMap database, as well as the Affymetrix and Illumina GWAS panels. Since increased density reduces the number of independent tests, using corrections like Bonferroni are not accurate. Instead, we used HAPMAP data and the linkage disequilibrium structure of the genome to identify the true number of independent SNPs across the genome. This work was published this year (1), giving researchers in the field guidelines for appropriate significance thresholds in several ethnic groups plus algorithms for recomputing these thresholds for other ethnic groups and as newer versions of the HapMap are released. We also explored the utility of various machine learning methods in genome-wide association studies, particularly with respect to power and detection of gene-gene and gene-environment interactions. We used GWAS genotype data from the Framingham Heart Study data repository with computer simulated trait data, thus allowing us to show that these methods may be able to detect interaction effects in suitably-powered studies. A paper presenting these results is in press. We are continuing to pursue the use of machine learning methods in genomics studies. Many of these projects are ongoing.