[BioC] sam analysis question

[James W. MacDonald]

lettieri at igb.cnr.it wrote:
> 
>  I'm running SAM in R package, I am analysing two affy array's groups data (two
> conditions in 3 separate pairs). I am using SAM starting from a filtered list
> containing 3000 genes but I had some problems. Infact when I analyze the delta
> table I have the number of the false positives that drastically goes down.For
> example, I have:
>  
>    Delta    p0  False Called   FDR
> 1    0.1 0.169 242.10   1259 0.033
> 2    0.2 0.169  31.10    272 0.019
> 3    0.4 0.169   2.50     23 0.018
> 4    0.5 0.169   1.10     13 0.014
> 5    0.7 0.169   0.15      3 0.008
> 6    0.8 0.169   0.15      3 0.008
> 7    1.0 0.169   0.10      2 0.008
> 8    1.1 0.169   0.05      1 0.008
> 9    1.3 0.169   0.05      1 0.008
> 10   1.4 0.169   0.05      1 0.008
> 
> anyone else has run across this and can tell me what should be done to solve it?

There are two problems here. If you are doing a paired analysis, there 
are only 2^3 permutations, so your null distribution will be very 
coarse. You would probably be better off using a t-distribution as your 
null rather than trying to permute (see the limma package). In addition, 
filtering your data down to 3000 genes that (I assume) are more likely 
to be differentially expressed is probably not a good idea. For a 
permutation method I would tend to use all genes and filter based on 
p-value and possibly fold change afterwards.

Jim


> 
> Thanks
> 
> mirella
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor


-- 
James W. MacDonald
Affymetrix and cDNA Microarray Core
University of Michigan Cancer Center
1500 E. Medical Center Drive
7410 CCGC
Ann Arbor MI 48109
734-647-5623