Question

Question about the usage of Package, qvalue

1

Entering edit mode

manozawa ▴ 10

@manozawa-22650

Last seen 4.9 years ago

Hi,

I am a very beginner of R and Bioconductor. I would like to calculate q-values from a list of p-values.

So, following the instruction on https://www.bioconductor.org/packages/release/bioc/html/qvalue.html, I installed the package 'qvalue' to RStudio Version 1.1.463 as follows.

if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager")

BiocManager::install("qvalue")

Then, using the following commands, I tried to obtain a list of q-values.

data<-read.csv("test.csv", header=TRUE) test2 <- data$test

tmp <- qvalue::qvalue(test2)

Then, the results somehow gives a lot of q-values that are smaller than the corresponding p-values. In addition, many q-values show exactly the same q-value, 4.670181e-02.

You can download the input list of p-values (removed). Sorry for the very long one (4571 p-values).

I would greatly appreciate if you have any suggestions to solve this issue.

Best regards, Fumi Nozawa

statistical significance p-value q-value FDR qvalue • 1.8k views

ADD COMMENT • link 4.9 years ago manozawa ▴ 10

1

Entering edit mode

This is a frequently asked question with answer here:

https://github.com/StoreyLab/qvalue/blob/master/README.md#frequently-asked-questions

ADD REPLY • link 4.9 years ago Storey, John D. ▴ 100

0

Entering edit mode

You could consider the more conservative Benjamini-Hochberg method of FDR adjustment. This will never produce smaller adjusted p-values than the input. (It will still sometimes produce identical adjusted p-values even if all the input values are distinct, this is expected behaviour.)

p.adjust(test2, method="BH")

ADD REPLY • link 4.9 years ago Paul Harrison ▴ 100

score 2 · Answer 1 · 2020-01-07

2

Entering edit mode

Gordon Smyth 51k

@gordon-smyth

Last seen 14 minutes ago

WEHI, Melbourne, Australia

It is known that qvalue can give q-values smaller than the corresponding p-values and there have been several posts about this topic over the years, for example:

The following might also be of interest: https://support.bioconductor.org/p/49864/

ADD COMMENT • link 4.9 years ago Gordon Smyth 51k

0

Entering edit mode

Thank you very much for your information! That helps me a lot to understand the concept of p-value and q-value. I may ask further questions in the near future.

Thanks again for your kind advise.

ADD REPLY • link 4.9 years ago manozawa ▴ 10