In this study, we observed explained variances between 0.01% and 3.02%. The 25UP and S4S samples were too small to detect significant effects in the individual samples. Virtually all phenotypes reached significance only after adding data from the larger samples (NTR and UKB). The comparison of four cohorts with different sample sizes has shown that in general samples of over 7,000 individuals are needed to find significant effects with these effect sizes (see Figure S1). 45 , 49 This means that for the phenotypes that were available in UK Biobank, the addition of the other samples has not led to a substantial increase in information over and above what we already learned from previous studies. This is the first study to our knowledge, using this method to give a concrete indication of what sample sizes are needed to detect the effect of a single gene. We may conclude that we must be cautious to draw conclusions from individual small samples, but that these smaller samples can be combined in meta‐analyses, especially for (possibly more detailed) phenotypes that are not available in large‐scale data sets.
