> In the 3dFDR document, there is an example
>
> p-value q-value z-score
> 0.000100 0.001500 3.174684
>
> Can I understand it as: P(z-score>=3.174684)=0.001500?
No, it means P(|z-score|>=3.174684)=0.0015.
> Then I checked the normal distribution table, the z-score
> corresponding to 0.0015 should be 2.9677 (it's not equal to
> 3.174684). So, I am confused......
That is because P(z-score>=2.9677)=0.0015.
> In addtion, with the FDR z-score bucket of defining overlay, if I
> adjust the p-value (under the threshold bar) to 0.05. Can I say
> E(FDR)<=0.05?
Yes, but be careful with one-tail vs. double-tail as the previous confusion demonstrated.
Gang