[BioC] Normalized microarray data and meta-analysis
kwyatt.mcmahon at ttuhsc.edu
Thu Dec 18 00:48:33 CET 2008
Excellent, Claus! I appreciate your input! That was my idea as well -
if you are trying to do one big experiment, then you'd definitely need
to adjust for any "study effects," but just trying to combine p-values
(I'm using an average effect size, which converts directly to p-values)
then it's less important.
Are there any other ideas on this subject?
Thanks to everyone so far,
K. Wyatt McMahon, Ph.D.
Texas Tech University Health Sciences Center
Department of Internal Medicine
3601 4th St.
Lubbock, TX - 79430
"It's been a good year in the lab when three things work. . . and one of
those is the lights." - Tom Maniatis
> -----Original Message-----
> From: Mayer, Claus-Dieter [mailto:c.mayer at abdn.ac.uk]
> Sent: Wednesday, December 17, 2008 5:43 PM
> To: Mcmahon, Kevin; bioconductor at stat.math.ethz.ch
> Subject: RE: Normalized microarray data and meta-analysis
> Dear Kevin,
> that is a difficult question indeed. I am not sure what type of
> microarrays we are talking about here, but if it were Affy arrays then
> normalisation methods like RMA or GCRMA perform an "across array"
> normalisation step, i.e. the normalised data from the same study will
> be more similar to each other than the ones from different studies. So
> for a better comparibility across studies it seems better to normalise
> the raw arrays from all studies together.
> Having said that, even if you are able to do this you will typically
> find that the data from the different studies cluster together, i.e.
> the normalisation is not able to remove all the differences between
> studies. So any proper meta analysis must somehow take into account
> this study effect (and there is a growing amount of literature how to
> do that).The importance of having the raw data depends on which
> approach you take; if you use a p-value comination approach like
> Stouffers method for example it shouldn't matter much for example, but
> if you try to put all data into one big analysis it might very well
> Best Wishes
> Hello Bioconductor-inos,
> I have more of a statistical/philosophical question regarding using
> vs. normalized data in a microarray meta-analysis. I've looked
> the bioconductor archives and have found some addressing of this
> but not exactly what I'm concerned with. I don't mean to waste
> time, but I was hoping I could get some help here.
> I've performed a meta-analysis using the downloaded data from 3
> different GEO data sets (GDS). It is my understanding that these are
> normalized data from the various microarray experiments. Seems to me
> that the data from those normalized results are normally distributed,
> those three experiments are perfectly comparable (if you think the
> author's respective normalization approaches were reasonable). All
> need to do is calculate some sort of effect size/determine a
> p-value/etc. for all genes in the experimental conditions of interest
> and then combine these statistics across the different experiments.
> However, I consistently read things like "raw data are required for a
> microarray meta-analysis." Does this mean that normalized data are
> directly comparable with eachother? If so, then why does GEO even
> such data?
> Any help would be wonderful!
> K. Wyatt McMahon, Ph.D.
> Texas Tech University Health Sciences Center
> Department of Internal Medicine
> 3601 4th St.
> Lubbock, TX - 79430
> "It's been a good year in the lab when three things work. . . and one
> those is the lights." - Tom Maniatis
> [[alternative HTML version deleted]]
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> Search the archives:
> The University of Aberdeen is a charity registered in Scotland, No
More information about the Bioconductor