Biostatistics & Computational Biology Branch
Examples of ongoing projects include:
Design of genome-wide studies based on rankings of associations: The huge numbers of single nucleotide polymorphisms (SNPs) studied in genome-wide association studies (GWAS) create a widely appreciated potential for false positives, and findings need to be replicated in follow-up studies. The design of such studies requires selecting a certain number of top hits (e.g., based on ranking p values) to be identified as good candidates for further study. The rank-based approaches developed by members of the Branch can adjust for population stratification, account for an estimated distribution of effect sizes, and determine the number of SNPs to carry forward to a replication stage.
Shared controls in genome wide association studies: An appealing genome-wide association study design compares one large control group against several disease samples. While reusing a control sample provides effective utilization of data, it also creates correlation between association statistics across diseases. Accounting for the correlation is particularly important when screening for SNPs that might be involved in a set of diseases with overlapping etiology. Members of the Branch have created association methods that account for dependency due to shared controls.
Multilocus association methods: Approaches are needed for efficiently capturing joint effects of multiple SNPs and SNPs interacting with multiple environmental factors, including methods that are robust with respect to the underlying genetic model of association, and with respect to the model that specifies the interactive effects. The methods under development should prove useful for discovery of interactions in case-control association studies and family studies, and should be informative in pathway analysis applications.