However, there are also disadvantages associated with the meta-analysis approach. For instance, it is more challenging to perform meta-analyses for more complex and potentially more informative analyses such as (1) polygenic scoring, which determines how much of the phenotypic variance can be explained by common SNPs in aggregate (Purcell et al. 2009), (2) structural equation modeling to demonstrate evidence for the endophenotype concept (Kendler and Neale 2010), and (3) stepwise linear or ridge regression, to identify the causal variant at an associated locus (McCarthy et al. 2008; McCarthy and Hirschhorn 2008). Not being able to perform interaction analyses (e.g., disease by SNP) is also a limitation, although this could be overcome as ENIGMA’s disease related working groups expand.
