[BioC] sampleinfo and lumiN of lumi package

[Pan Du]

Hi Michal,

The lumiR parses the column headers to retrieve the sample information. It
follows the convention defined in our Northwestern University Microarray
core and most other Illumina data we have seen. The purpose of this is just
for the convenience for the subsequence analysis like differentially
expressed gene identification. If you find the retrieved information is
incorrect, you can reset them.

As for the loess normalization method in lumiN, it basically calls
normalize.loess() function. As for whether it will do log transform again if
you set log.it=TRUE, it depends how normalize.loess() function implements
it. I just have a try. It seems the function can automatically check the
data and determine the operation. So you don't need to worry about it.


Pan


On 6/23/07 5:00 AM, "bioconductor-request at stat.math.ethz.ch"
<bioconductor-request at stat.math.ethz.ch> wrote:

> 
> Message: 11
> Date: Fri, 22 Jun 2007 12:22:13 -0400
> From: Michal Blazejczyk <michal.blazejczyk at mail.mcgill.ca>
> Subject: [BioC] Another lumi question (sample info)
> To: bioconductor at stat.math.ethz.ch
> Message-ID: <15310669673.20070622122213 at mail.mcgill.ca>
> Content-Type: text/plain; charset=us-ascii
> 
> Hi,
> 
> I have another question regarding lumi.  Function lumiR() parses
> column headers in the file to retrieve sample and label information.
> the result is stored in an AnnotatedDataFrame object.  My question
> is the following: is this information used in calculations in lumi?
> That is, will I get different results depending on how my samples
> are organized, and in what circumstances?
> 
> Best,
> Michal Blazejczyk
> 
> 
> 
> ------------------------------
> 
> Message: 12
> Date: Fri, 22 Jun 2007 13:05:40 -0400
> From: Michal Blazejczyk <michal.blazejczyk at mail.mcgill.ca>
> Subject: [BioC] Double-logging in lumi + loess?
> To: bioconductor at stat.math.ethz.ch
> Message-ID: <843884624.20070622130540 at mail.mcgill.ca>
> Content-Type: text/plain; charset=us-ascii
> 
> Hi again,
> 
> I've been reading the code of lumiN() function...  If I call it with
> method = 'loess' then lumiN() will call normalize.loess() to normalize
> the data.  However, normalize.loess() has a parameter called log.it
> which defaults to TRUE.  Doesn't it mean that if I stabilized my
> variance prior to calling lumiN(), and then call lumiN() with method =
> 'loess', my data will be double-logged?  I know that I could simply
> provide log.it = FALSE when I call lumiN() but this looks like a
> potential problem for users who won't attentively read the whole
> documentation.
> 
> Best,
> Michal Blazejczyk
> 
> 
> 
> ---