A second question is which of the recently reported GWAS findings are true or credible if they were discovered in small cohorts. ENIGMA focuses on meta-analyses, but some have argued that smaller individual studies—when replicated—may discover interesting (and strong) associations. Paus et al. (2012) noted that they had run a successful GWAS in only around 300 participants. Also, some GWAS signals may be important in the context of some individual cohorts, but may be washed out in meta-analyses. A more skeptical line of argument is that if we need 20,000+ samples to detect an effect, most likely what we see has very small effect size. Even so, some recent GWAS analyses appear to have picked up variants with strong effects—e.g., that roughly double disease risk (e.g., in TREM2), and these effects have been confirmed in smaller cohorts by comparing brain images from carriers and non-carriers (Rajagopalan et al. 2013).
