Convert Fdr Control Q-Value To P-Value

Question

[J.F.Jiang]

Hi everyone,

I have downloaded a set of processed data of gene expression relationship, while they offered a result of q value. Since I want to use the p value to adjust the next step, how can I convert this Q value to p value?

Thanks!

[Ryan Thompson]

Is it possible that your dataset was analyzed by a method that does not produce p-values? For example, baySeq, a tool for testing differential expression in RNA-Seq data, produces posterior probabilities from a Bayesian analysis and calculates FDR directly from them. The nature of the calculations is such that no p-value is involved.

[ff.cc.cc]

How did you get your Q values ?

Answer 1

[David Quigley]

I think the short answer is that you can't. One way to calculate a Q value estimate is from a distribution of P values (e.g. the way Storey's qvalue works), but I don't think there is any way to reconstruct the P value distribution from a set of estimated Q values. You should calculating the P values directly if that's what you want. The manual to the qvalue package has more information about what a Q value is and how it can be calculated, which you might want to read.

[J.F.Jiang]

YES, I have read the wiki of Q value and methods of original author, I can not convert this Q, but I can still estimate the p-value