[BioC] Suggestions on Agilent gene expression data

Francois Pepin fpepin at cs.mcgill.ca
Thu Jun 12 05:54:24 CEST 2008


Hi Allen,

you'll want to use either the log2(ratio) or the log(ratio). As Sean 
said, they're equivalent up to a constant.

Basically all the BioC packages (and most of the other tools) expect 
logged data, especially when it is pre-normalized.

Many people tend to use log2, as they're easier to interpret directly 
(easier to multiply in your head by 2 than by e) and the scanner 
generally gives values that are 1-2^16.

Francois

ss wrote:
> Hi Sean,
> 
> Thanks a lot! The data we downloaded has already been normalized and
> processed.
> So I guess I can go ahead without further quality control. My question then
> is that
> if I want to utilize some other tools or softwares for higher level
> analysis, should I use
> ratio, or log2(ratio) or Log(ratio)? Or it does not matter at all?
> 
> Best,
>      Allen
> 
> 
> 
> On Wed, Jun 11, 2008 at 10:01 PM, Sean Davis <sdavis2 at mail.nih.gov> wrote:
> 
>> On Wed, Jun 11, 2008 at 7:43 PM, ss <affysnp at gmail.com> wrote:
>>> Dear all,
>>>
>>> For whom is familiar with Agilent gene expression data, I would like to
>>> ask help.
>>>
>>> We recently received some Agilent gene expression data from our
>>> collaborators. For individual sample, there are 7 corresponding columns:
>>>
>>>  Unknown:Log(Ratio)  Unknown:Ratio  Unknown:Fold Change
>>  Unknown:Log(Error)
>>> Unknown:P-Value  Unknown:Intensity 1  Unknown:Intensity 2
>>> It seems that Ratio= Intensity1/Intensity2. I wonder whether I should use
>>> log2(Ratio) or Log(Ratio) or just Ratio for further analysis. Besides,
>> why
>>> should Ratio be calculated as Intensity1/Intensity2 instead of
>>> Intensity2/Intensity1?
>> Hi, Allen.  You'll probably need to do some quality control and some
>> normalization.  The choice of ratio is arbitrary; you can always
>> invert it if it is more convenient to do so.  As for log2 or log, they
>> are equivalent to each other with the exception of a constant.  You
>> might look at the limma manual for some guidance on working with
>> two-color arrays.
>>
>> Sean
>>
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> 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



More information about the Bioconductor mailing list