Genetic associations with diseases are identified by studies that link variations in the genetic code, called single nucleotide polymorphisms (SNPs), to disease-carrying individuals. Findings from genetic association studies can be used to understand how genetics may play a role in the disease of interest. The statistical analyses of data obtained from genetic association studies typically consist of single SNP analysis or haplotype (a sequence of SNPs in a chromosome) analysis, both of which have limitations. VIDI assistant member Dr. James Dai and colleagues wanted to improve upon existing analytical methods by developing a new algorithm to search for the most informative subset of SNPs in candidate genetic regions by scanning models with different sets of SNPs in a step-wise fashion and comparing the prediction errors by cross-validation. In this work, the new algorithm, named SNP-Haplotype Adaptive REgression (SHARE), was validated using simulations and actual datasets to show that SHARE has an increased power over other algorithms, having the potential to significantly improve genetic association studies.
SHARE: an adaptive algorithm to select the most informative set of SNPs for candidate genetic association. Dai JY, Leblanc M, Smith NL, Psaty B, Kooperberg C. Biostatistics. 2009 July 15