[BioC] DESeq Question

[Simon Anders]

Hi

On 03/21/2012 12:26 AM, Omar Darwish wrote:
>   I used the following commands to find the differentially expressed genes.
> I have two questions,
>
>     1. Are the following commands correct ? or am i missing something?

Yes, except for two oddities:

 > libsizes<- clibsizes<- c(X.1=18143070, X.2=13544150, Y.1=17853990,
 > Y.2=15038501)
 > sizeFactors(cds)<- libsizes[-1]
 > cds<- estimateSizeFactors( cds )

Why do you set the size factors manually, and the use 
'estimateSizefactors' to overwrite this information? The first two 
commands are gratuitous (and don't even work becaus of the "[-1]"). Omit 
them.

 > resSig<- res[ res$padj<  .001, ]

Do your really want an FDR of only 0.1%? This is extremely stringent. Do 
you have any reason for this unusual choice?

>     2. How can i determine which genes are differentially expressed?
>     induced? suppressed? unchanged?

Those with and adjusted p value (column "padj") below your chosen FDR 
threshold are significant, and the column "log2FoldChange" tells you the 
log2 fold change and hence whether they are up- or downregulated.

   Simon