[BioC] using genomic DNA as universal reference

[Naomi Altman]

Actually, the requirement for loess normalization is that 
differential expression is not dependent on expression level, and 
that up and down regulation are symmetric.

--Naomi


At 03:27 PM 6/5/2008, Jianping Jin wrote:
>Thanks Sean for your input!
>
>the T-test result was used just for estimation of how many probe 
>expressions were significantly different between RNA/DNA samples. 
>This is also related to my normalization questions. According to my 
>understanding the basic assumption for loess normalization is that 
>most of the probes on the array are not differentially expressed. 
>This is Agilent two-color data. Is loess normalization appropriate 
>for such a different data on each array?
>
>thanks again!
>
>Jianping
>
>--On Thursday, June 05, 2008 1:28 PM -0400 Sean Davis 
><sdavis2 at mail.nih.gov> wrote:
>
>>On Thu, Jun 5, 2008 at 12:31 PM, Jianping Jin <jjin at email.unc.edu> wrote:
>>>Dear list,
>>>
>>>I would like to ask comments and suggestions on how to normalize
>>>microarray data with genomic DNA as reference.
>>>
>>>The experiments were performed with bacterial RNA and genomic DNA
>>>samples. What I noticed was that the data were pretty consistent across
>>>all chips on both channels.  But there exists a huge difference between
>>>the two channels in terms of the distribution of the probe intensities,
>>>although the average intensities were the same for the both channels. T
>>>statistics with non-normalized data showed that there were two thirds
>>>probes with p values <= 0.05 by comparing the hybridization intensities
>>>between red and green channels.
>>>
>>>Regarding to the huge difference described above the normalization
>>>methods people usually use may not be appropriate for the RNA/DNA data
>>>sets. What normalization algorithms would be useful if there is any?
>>>Does anyone have experience with this?
>>
>>While not ideal, this sounds like a common reference design.  You
>>could make use of normal two-channel normalization methods (centering,
>>linear, or loess, etc.), use only single-channel data (and ignore the
>>control), or use some of the single-channel normalization methods for
>>two channel data described in the limma user guide.  I'm not sure that
>>the t-test results are that important in making a decision.  Others
>>might have more insight and (more importantly) more experience in this
>>situation.
>>
>>Sean
>
>
>
>##################################
>Jianping Jin Ph.D.
>Bioinformatics scientist
>Center for Bioinformatics
>Room 3133 Bioinformatics building
>CB# 7104
>University of Chapel Hill
>Chapel Hill, NC 27599
>Phone: (919)843-6105
>FAX:   (919)843-3103
>E-Mail: jjin at email.unc.edu
>
>_______________________________________________
>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

Naomi S. Altman                                814-865-3791 (voice)
Associate Professor
Dept. of Statistics                              814-863-7114 (fax)
Penn State University                         814-865-1348 (Statistics)
University Park, PA 16802-2111