p.adjust function fdr and BH
1
0
Entering edit mode
@zhangjianhai-12955
Last seen 23 months ago

Hello,

In the p.adjust R function, there are multiple adjust methods, as shown here: https://www.rdocumentation.org/packages/stats/versions/3.5.0/topics/p.adjust

I read many posts, it seems the "BH" is used for "FDR", but why in the function "BH" and "fdr" are two independent methods?

Another question is  can I understand "holm", "hochberg", "hommel", "bonferroni", "BH", "BY", are used for adjusted p value, and "fdr" is used for false discovery rate? What is the difference between adjusted p value and fdr?

Regards,

Jianhai

fdr BH • 2.1k views
ADD COMMENT
3
Entering edit mode
@gordon-smyth
Last seen 16 minutes ago
WEHI, Melbourne, Australia

"BH" and "fdr" are the same method. The p.adjust help page (that you've linked to) says that "fdr" is merely an alias for "BH".

"BH", "BY" and "fdr" all produce false discovery rates, again as explained by the help page.

A little history

Originally, p.adjust() didn't have any FDR methods. Then Benjamini and Hochberg's method was added. At that time, Benjamini and Hochberg's method was the only FDR method, so it was just called "fdr". Later on, we added the Benjamini & Yekutieli's method as well, so we needed new names. We called the new method "BY" and renamed the old method to "BH". The old name "fdr" was also kept as well, just for backward compatibility, which is how we ended up with two names for the same thing.

ADD COMMENT
0
Entering edit mode

Hello Gordon,

Now I understand. Thanks.

Regards

ADD REPLY

Login before adding your answer.

Traffic: 405 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6