GCRMA Normalization Differences between Windows and MAC
3
0
Entering edit mode
@martin-martinez-3985
Last seen 9.6 years ago
I am performing a gene expression analysis and I founded important differences between the results of the GCRMA normalization obtained in a Windows machine (gcrma ver.2.16) and a MAC one (gcrma ver.2.18.1). It consists in a cut off, so all the lower values dissapear. The code is next: library(affy) library(gcrma) affy.data= ReadAffy() eset.gcrma=justGCRMA() (the same as eset.gcrmax=gcrma(affy.data)) my question is about default values. Does them variate between versions? The difference is very significant and I tried changing all the arguments, but not satisfactory results are obtained. Regards, Martín M. [[alternative HTML version deleted]]
Normalization gcrma Normalization gcrma • 962 views
ADD COMMENT
0
Entering edit mode
@sean-davis-490
Last seen 12 weeks ago
United States
On Tue, Mar 23, 2010 at 8:46 AM, martin martinez <tininmartinez at="" gmail.com=""> wrote: > I am performing a gene expression analysis and I founded important > differences between the results of the GCRMA normalization obtained in a > Windows machine (gcrma ver.2.16) and a MAC one (gcrma ver.2.18.1). It > consists in a cut off, so all the lower values dissapear. The code is next: > > library(affy) > library(gcrma) > affy.data= ReadAffy() > eset.gcrma=justGCRMA() (the same as eset.gcrmax=gcrma(affy.data)) > > my question is about default values. Does them variate between versions? The > difference is very significant and I tried changing all the arguments, but > not satisfactory results are obtained. Hi, Martin. It is quite possible that there are changes between versions of code that might be expected to change results. You'll want to use the same versions in order to make a valid comparison. If you do find differences when using the same versions, you'll probably need to provide a reproducible example and sessionInfo() for both platforms. Sean
ADD COMMENT
0
Entering edit mode
I am not following gcrma very closely, but this exact question seems like a question that has been discussed in at least 5 lengthy threads over the last year or so. So I would strongly advise spending time looking at the email archives. Kasper On Tue, Mar 23, 2010 at 8:55 AM, Sean Davis <seandavi at="" gmail.com=""> wrote: > On Tue, Mar 23, 2010 at 8:46 AM, martin martinez > <tininmartinez at="" gmail.com=""> wrote: >> I am performing a gene expression analysis and I founded important >> differences between the results of the GCRMA normalization obtained in a >> Windows machine (gcrma ver.2.16) and a MAC one (gcrma ver.2.18.1). It >> consists in a cut off, so all the lower values dissapear. The code is next: >> >> library(affy) >> library(gcrma) >> affy.data= ReadAffy() >> eset.gcrma=justGCRMA() (the same as eset.gcrmax=gcrma(affy.data)) >> >> my question is about default values. Does them variate between versions? The >> difference is very significant and I tried changing all the arguments, but >> not satisfactory results are obtained. > > Hi, Martin. > > It is quite possible that there are changes between versions of code > that might be expected to change results. ?You'll want to use the same > versions in order to make a valid comparison. ?If you do find > differences when using the same versions, you'll probably need to > provide a reproducible example and sessionInfo() for both platforms. > > Sean > > _______________________________________________ > 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 >
ADD REPLY
0
Entering edit mode
@wolfgang-huber-3550
Last seen 3 months ago
EMBL European Molecular Biology Laborat…
Dear Martin it would help if you were a bit more precise about how the results differ. Could you provide us a reproducible example (e.g. using the data in the ALLMLL package, http://www.bioconductor.org/packages/release/data/experiment instead of your own private files), code for a plot or other kind of comparison that shows what you perceive as an 'important difference', and also the output of sessionInfo()? I doubt that the difference is related to "Windows" vs "Mac", but rather to changes in the "gcrma" package, and the package maintainer might have more insights on changes between the two versions (2.16 and 2.18.1), and where they are documented. The svn log does report: ---------------------------------------------------------------------- -- r43600 | jmacdon | 2009-12-17 16:59:57 +0100 (gio, 17 dic 2009) | 1 line Modifications to justGCRMA to keep consistent with gcrma. Because gcrma has become more complex (e.g., being able to use either the internal NSB estimates, MM probes, or NCprobes), it is no longer feasible to keep justGCRMA completely consistent. Instead, justGCRMA will only use the internal NSB estimates (which is the default for gcrma anyway). Best wishes Wolfgang martin martinez ha scritto: > I am performing a gene expression analysis and I founded important > differences between the results of the GCRMA normalization obtained in a > Windows machine (gcrma ver.2.16) and a MAC one (gcrma ver.2.18.1). It > consists in a cut off, so all the lower values dissapear. The code is next: > > library(affy) > library(gcrma) > affy.data= ReadAffy() > eset.gcrma=justGCRMA() (the same as eset.gcrmax=gcrma(affy.data)) > > my question is about default values. Does them variate between versions? The > difference is very significant and I tried changing all the arguments, but > not satisfactory results are obtained. > > Regards, > > Mart?n M. > > [[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 -- Best wishes Wolfgang -- Wolfgang Huber EMBL http://www.embl.de/research/units/genome_biology/huber/contact
ADD COMMENT
0
Entering edit mode
Zhijin Wu ▴ 260
@zhijin-wu-2378
Last seen 9.6 years ago
We did make changes of the package over the years. One of the major changes is in the step of "gene specific binding (GSB) adjustment". This adjustment is done because probes have different efficiency in hybridization. At the background adjustment step, we do not really give probes 0 even if they appear to be mostly background, because of the trouble for log transformation. So those probes were shrunk to a lower bound parameter instead of 0. The original GSB adjustment was done on all probes. later we have decided to leave those probes at the lower bound un-adjusted because these were most likely background only (Thanks to Lim et al, who discovered the problem and suggested the fix. see http://bioinformatics.oxfordjournals.org/cgi/content/abstract/23/13/i2 82 .) This change would mostly affect genes that are probably not expressed, but would also affect some genes that have a few very weak probes. best, Jean martin martinez wrote: > I am performing a gene expression analysis and I founded important > differences between the results of the GCRMA normalization obtained in a > Windows machine (gcrma ver.2.16) and a MAC one (gcrma ver.2.18.1). It > consists in a cut off, so all the lower values dissapear. The code is next: > > library(affy) > library(gcrma) > affy.data= ReadAffy() > eset.gcrma=justGCRMA() (the same as eset.gcrmax=gcrma(affy.data)) > > my question is about default values. Does them variate between versions? The > difference is very significant and I tried changing all the arguments, but > not satisfactory results are obtained. > > Regards, > > Mart?n M. > > [[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 -- ------------------------------------ Zhijin (Jean) Wu Assistant Professor of Biostatistics Brown University, Box G-S121 Providence, RI 02912 Tel: 401 863 1230 Fax: 401 863 9182 http://www.stat.brown.edu/zwu
ADD COMMENT

Login before adding your answer.

Traffic: 706 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6