To some researchers, these preliminary observations suggest an expected pattern of effect sizes, whereby GWAS for cognitive traits may have top hits with smaller effect sizes than those for imaging traits, which in turn may tend to have smaller effect sizes than expression traits. Also, now that consistent hits have been identified in ENIGMA1 and ENIGMA2, it should be possible to estimate effect sizes for the same hits in new replication samples, to understand what sample sizes are sufficient to detect them. Conversely, some have argued that the available data on genetic effect sizes at different levels of neuroscience are currently too limited to draw conclusions from only the three data points cited here. Clearly, we do not yet have sufficient information to say whether imaging genetics effects sizes will always be larger than effect sizes for cognitive/behavioral phenotypes. One way to begin to address this question might be to quantify effects of some specific functional SNP at three different related levels, such as eQTL data from the hippocampus, hippocampal volume on MRI, and a behavior that is tightly dependent on hippocampal function; this may show the appropriate differential effect sizes in one framework.
