[BioC] limma and Rank Products: comparison of the number of results

[Thomas Hampton]

I have used rank products ad limma head to head many times, so I think  
I understand your question.

I can imagine scenarios where limma would provide a longer list, but I  
have not observed it myself.

It is in many ways amazing that the two tests generate similar lists  
at all, given the enormous
difference in the way they go about selecting genes. Limma "cares"  
quite a lot about within
group variation, and can select genes with very small between group  
differences as significant
as long as within group variation is sufficiently small.  Genes with  
smaller p values appear at the
head of the list in limma, where genes with the largest fold  
differences appear at the head of
a RankProd list. The net of all this is that I believe you will find  
that the two lists are not only different
in length, but quite different in order. So whether "false prediction  
rates" and "false discovery rates" are
arguably similar or not, the two tests make rather different  
assumptions which will ultimately drive the two
estimates apart. I generally feel that "longer is better" when it  
comes to gene lists, especially for pathway
analysis, but I generally require a difference of 1.4 fold between  
conditions before I get enthusiastic.

Best,


Tom



T
On Feb 17, 2010, at 9:10 AM, Juan Carlos Oliveros wrote:

> Dear all
>
> When working with comparative experiment based on Affymetrix gene  
> expression arrays I usually apply one of the following combination  
> of methods:
>
> RMA + limma + FDR
>
> or
>
> RMA+ Rank Products
> (rank products "Percentage of false prediction" values are supposed  
> to be equivalent to FDR)
>
> Usually we obtain much more differentially expressed genes when  
> using Rank Products than when using limma at the same FDR threshold.
>
> I wonder if in your case is the same. Do you obtain many more  
> results with Rank Products than with limma at the same FDR cutoff?
>
> In a recent experiment we obtained the opposite (more results with  
> limma) and I'd like to know your experience when using both methods  
> regarding the number of results.
>
> best,
>
> Juan Carlos Oliveros, Ph.D.
> CNB-CSIC, Madrid, Spain
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor