Genome-wide association studies (GWAS) have shown great promises in identifying the genetic variants associated with complex diseases. Most commonly used methods include single SNP analysis or haplotype analysis. Due to the fact that many complex diseases are due to multiple genes and their interactions and the fact the SNPs from GWAS are often in linkage disequilibrium, single SNP analysis often leads to loss of power. In this talk, I will present some on-going work on analysis of data from GWAS, including methods for selecting and adjusting for other relevant SNPs in testing a given SNP; a hidden Markov random field approach to account for LD among the SNPs and methods for gene-based analysis of SNP data. I will demonstrate these ideas using a case-control study of neuroblastoma. Extension to incorporate known gene and protein interactions will be briefly discussed.
Return to Biostatistics Working Group