In the context of this study, the two parameters of the beta distribution, k and n, can be interpreted as the rank of the associated variant and the effective number of independent tests performed in cis, respectively. We looked at the ML estimate distributions of these parameters across all genes in the GEUV_EUR dataset and find first that parameter k values tend to center around 1.0, in line with what is expected for the top variant (Fig. 1a). Second, we find that the parameter n values show high dispersion (Fig. 1b) and are consistently smaller than the actual number of variants being tested in cis (Fig. 1c); both suggesting that the beta distribution captures well the redundancies between variants, a consequence of LD. This also highlights the importance of performing permutations instead of using a Bonferroni correction based on the number of variants, which would result in a substantial proportion of false negative results.
