Skip to main content

Table 3 Empirical null estimates from correlated test statistics yield proper control of the false positives rate without any reduction in power

From: Controlling bias and inflation in epigenome- and transcriptome-wide association studies using the empirical null distribution

Method

False positive rate

Power

 

mean (stdev)

mean (stdev)

 

Uncorrelated

No correction

0.050 (0.003)

0.770 (0.020)

Genomic control

0.028 (0.003)

0.710 (0.020)

Bayesian control

0.052 (0.003)

0.770 (0.020)

 

Correlated

No correction

0.040 (0.030)

0.770 (0.020)

Genomic control

0.023 (0.006)

0.730 (0.090)

Bayesian control

0.054 (0.020)

0.800 (0.060)

  1. Mean and standard deviation of the number of false positives and true positives (power) for a simulation study repeated 100×. Correlated test statistics were generated according to the simulation setup of Efron [51]. The table summarizes the results for uncorrelated test statistics and correlated test statistics, without any correction for inflation or bias, using genomic control and using our Bayesian method
Back to article page

Genome Biology

ISSN: 1474-760X

Contact us