at multiple nearby SNPs. In any single sample, many of the clusters of positive findings at nearby SNPs could be due to stochastic differences in haplotype frequencies between cases and controls that are not related to phenotype. As the same chromosomal regions are identified by more and more independent samples, however, the likelihood that such identification is due to stochastic differences in haplotype frequencies that are unrelated to phenotype declines sharply. Assessment in multiple, independent samples, as we perform here, provides the best control for stochastic differences in haplotype frequencies that might be expected, by chance, between any single case and control samples. 7a) “Template” GWA analyses that focus on single SNPs may provide modest biases toward identification of large genes that contain many SNPs; 7b) The current analyses require nominally significant associations for multiple SNPs that lie within narrow chromosomal regions. A number of the smallest genes cannot be identified by this approach [25], in ways that might lead to an even more prominent bias toward identification of larger genes in this way; 8a) “Template” GWA approaches focus on metaanalyses as means to evaluate convergence of data from single SNPs across many independent samples; 8b) As more such data
