AvgCorr

Threshold Z score above which the p-value likely corresponds to a true positive.

Force correlation estimate to fall in the interval (a,1-a);

Estimates the correlation between correlated p-values on the Z-score scale. The
p-values are taken as having arisen from different tests of the same hypothesis
using the same data. Since an estimate of correlation under the null is of interest,
pairs where at least one of the Z scores exceeds the threshold \(\tau\) are excluded.

Implementation of genetic association tests for continuous outcomes utilizing the rank-based inverse normal transformation (INT). For outcomes whose residual distribution is heavily skewed or enriched for outliers, INT-based tests provided valid inference and improved power. The primary contribution is a rank normal omnibus test (RNOmni), which synthesizes two complementary INT-based approaches. In simulations against non-normal phenotypes, the omnibus test controlled the type I error in the absence of genetic associations, and improved power in the presence of genetic associations. Under the same settings, standard linear regression variously failed to control the type I error in the absence of associations, and was underpowered in the presence of associations.

Zachary McCaw

AvgCorr: Average Correlation Estimate.

Description

Usage

Arguments

Value