James W. MacDonald
jmacdon at med.umich.edu
Fri Apr 14 15:07:10 CEST 2006
Chen, Kegui wrote:
> Hi, All:
> Following the answer from Gordon, I have another question about the
> FDR. Since there two adjust method fro FDR in R.2.2.0, BH and BY,
> which one is the FDR used in SAM? The FDR in SAM is different from
> any above BH or BY? From SMA there is a FDR number (for example
> 5%) with the identified differential gene list. From the AffylmGUI
> there is adjusted P value (FDR) with every gene.
Technically, all three FDR methods are different. BH and BY are
'one-step' methods that are based on the unadjusted p-value (which is
computed using a t-distribution as the Null distribution), whereas SAM
uses permutation to estimate the Null distribution and compute the so
However, the interpretation of an FDR value is the same, regardless of
how it is computed. Simply put, if you have a list of genes and the
_largest_ FDR is 0.05, then you are estimating that 5% of the genes in
that list are not really differentially expressed.
> I think that the FDR 5% in SAM means that there are 50 false
> discovery genes in a of list 10000 genes. So, is there any
> possibility to know the number of false discovery genes in the
> differential gene list (for instance 1000 genes) by BH or BY adjusted
> P value from AffylmGUI. For me, the FDR idea of SMA is more clearly
> stated to make me comfortable.
> I am not a statistician. Probably my questions do not make sense in
> statistics. However, I think it makes sense to biologist.
> K. Chen
> From: Gordon Smyth [mailto:smyth at wehi.edu.au] Sent: Thu 4/13/2006
> 12:25 AM To: Chen, Kegui Cc: BioC Mailing List; Keith Satterley
> Subject: RE: [BioC] AffylmGUI
> Dear K Chen,
> The new option "BH" is the same as the old option "fdr". The new
> names were introduced into the p.adjust() function in R 2.2.0 because
> there are now two FDR control methods (BH and BY). See ?p.adjust
> We removed the Bonferroni, Hochberg and Hommel options because we
> don't think these are ever the best options for microaray data.
> Best wishes Gordon
> At 01:33 PM 13/04/2006, Chen, Kegui wrote:
>> Dear Gordon:
>> I am very sorry for the confusion.
>> The R program I used is Version 2.2.1. The AffylmGUI I used is
>> Version 1.5.4. Limma is Version 2.4.12. The data I used is the
>> example data, Estrogen. My computer is windows PC.
>> I followed the instruction provided in the program step by step.
>> When open the Toptable What I can see in the Toptable options is
>> not the one provided in the instruction. The difference is in the
>> Adjust method. In the instruction there the following options:
>> Bonferroni, Holm, Hochberg, Hommel, FDR and None, total 6 options.
>> What I see in my computer is None, BH, BY, Holm, total 4 options.
>> I' like to know FDR since it is popular now.
>> Please explain it.
>> K. Chen
>> -----Original Message----- From:
>> bioconductor-bounces at stat.math.ethz.ch
>> [mailto:bioconductor-bounces at stat.math.ethz.ch] On Behalf Of Gordon
>> Smyth Sent: Wednesday, April 12, 2006 12:25 AM To:
>> Chen at wehi.edu.au; Kegui at wehi.edu.au.chenkeg.at.missouri.edu Cc:
>> BioC Mailing List Subject: [BioC] AffylmGUI
>>> [BioC] AffylmGUI Chen, Kegui chenkeg at missouri.edu Wed Apr 12
>>> 04:30:28 CEST 2006
>>> Hi, Dear all:
>>> I am new here. I run the AffylmGUI using the example data,
>>> Everything works for me but the FDR. I can not get FDR for the
>>> the manual, it looks that there are 6 choices for the adjust
>>> method. However, I can not get 5 choices without FDR in my
>>> computer? Please
>>> me to solve this issue.
>>> Thanks! K. Chen
>> Dear K Chen,
>> When you ask for help, you need to give other people on the list
>> enough information to be able to help you. It doesn't tell us
>> anything to say that you can't get something. We're not looking
>> over your shoulder while you work at your computer and we can't see
>> what you mean.
>> You need to tell us in detail i) what you did, and ii) what output
>> you got from the software iii) why you think what you got is not
>> Remember, you need to tell us the facts as in (i) and (ii), not an
>> interpretation of them.
>> You also need to tell us exactly what version of R and affylmGUI
>> you are using, and on what sort of computer.
>> Best wishes Gordon
> _______________________________________________ Bioconductor mailing
> list Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor Search the
James W. MacDonald, M.S.
Affymetrix and cDNA Microarray Core
University of Michigan Cancer Center
1500 E. Medical Center Drive
Ann Arbor MI 48109
Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues.
More information about the Bioconductor