Bioinformatics 2005 21(24):4435; doi:10.1093/bioinformatics/bti777
Erratum
Bias in the estimation of false discovery rate in microarray studiesYudi Pawitan, Karuturi R Krishna Murthy, Stefan Michiels and Alexander Ploner
- In Section 2.2, in the distribution of P-value, the central and non-central distributions F0 and Gi were in the wrong order. It should be
and the approximate non-central t distributions are
This means all the formulas associated with the densities are as reported, except we need to change the sign of
. (The previous formulas worked because the DE genes were symmetric.)
- The analysis of BRCA data (Section 3.3) was performed on the original scale. If one uses the logged scale, the results differ slightly. A 3-nonzero-component model (q = 3) still minimizes the AIC, but we would recommend a more stable 2-nonzero-component model.
The author wishes to apologise for this error.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||