The equation used is the following:(4)where is the usual Z statistic and the shrinkage factor r is the ratio of the variance of the prior and total variance (). Assuming a normal distribution, the p-value of each SNP can be converted to standard one-tailed Z-score by using inverse normal cumulative distribution function. So for a SNP, all that it is needed are the p-values from a standard regression output, and , the standard deviation of the normal prior N(0,W) on . The variance of the effect estimate, V, can be approximated using the MAF and sample size. However for imputed data it is preferable to use the variance outputted in standard regression analysis directly in the ABF equation. For the expression dataset used here, the variance and effect estimates from the regression analysis were used for computation of ABFs (see Text S1 for more details).
