[BioC] converting intensity values to log ratios?

[James W. MacDonald]

Hi Mark,

On 1/31/2012 4:16 PM, Mark Baumeister wrote:
> Hi all,
>
> I am trying to convert RMA normalized intensity values in my matrix
> (generated from Affymetirx raw .CEL data composed of
> ~30 tumor samples and 7 normal samples) to log ratios in which the
> intensity values from each tumor sample are compared with a reference
> sample (in this case I chose one of my normals).  I was told that I could
> use a function like
>
>> log(eset/esetRef,2)
> to generate the new matrix containing log ratios.

Wow. That was some horrible advice. Please stop relying on whoever told 
you that.

First, please note that RMA normalized intensity values are already on 
the log_2 scale. So taking logs again isn't necessary, nor advisable.

Second, why are you converting to log ratios? You are losing information 
for no apparent gain. Your best bet is to simply make comparisons 
between groups using something like limma or siggenes or multtest and 
proceed from there. Also note that the numerator of your statistic 
(which limma returns by default with the topTable() function) is the log 
ratio that you are seeking, so you get that for free.

Best,

Jim


>
> I have reduced my data using a general filter to remove low signal and low
> variance:
>
>> dim(esetSubNat)
> [1] 1470   36
>> dim(esetSubRef)
> [1] 1470    1
>
> When I try this log function I get the following error:
>
>> esetSubLog<- log(esetSubNat/esetSubRef,2)
> Error in esetSubNat/esetSubRef : non-conformable arrays
> Can anyone give me any idea where I am going wrong or how else I might try
> to generate the log ratios?
>
> Thank you,
> Mark
>
>
>
>
>
> http://sites.google.com/site/lfmmab/
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor

-- 
James W. MacDonald, M.S.
Biostatistician
Douglas Lab
University of Michigan
Department of Human Genetics
5912 Buhl
1241 E. Catherine St.
Ann Arbor MI 48109-5618
734-615-7826

**********************************************************
Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues