[BioC] Checking output of DESeq package analysis

[Simon Anders]

Hi

On 2012-04-09 22:58, Anupreet kour [guest] wrote:
>
> I run DESeq package on my RNA seq hit count data (without
> replication). The input (hit count values) and output values (base
> means A&  B) in the columns are the same. I want to make sure that
> these results are reliable as with a dataset of 78774 rows all the
> input values are the same as output values. I am getting fold changes
> and p-values which seems to be correct.  The code and outputs are
> given below: (My code is running fine, my concern is output values
> (base mean A and B ) which are the same as WT_hit_counts  and
> Mutant_hit_counts)values. I think for some reason code is returning
> right p values but not the right base means. I will appreciate your
> help.

The baseMeans are simply the genewise means of the normalized counts. If 
you have only one sample per condition, it is hence the normalized count 
itself. In you case, the normalization seems to have had no effect -- 
your sizeFactors seem to be 1. This may happen if you have very low 
count values. In this case, using the shorth (see ?estimateSizeFactors) 
might work better.

In the end, however, with as low count values and no proper replication, 
I am afraid it is unlikely that fixing the normalization will help much 
in getting a signal from your data.

   Simon