vsn2 bug?
1
0
Entering edit mode
Tobias Straub ▴ 420
@tobias-straub-2182
Last seen 7.1 years ago
Dear all, with the upgrade to bioc 1.5.0 I am facing vsn2. as already discussed in an older thread there are calculation differences. however, in my dataset(s) I get a massive outlier comparing the array means after normalization with old vsn and vsn2. scatterplot here: http://server7.pictiger.com/img/198354/other/vsn- bioc.png my data are six replicates of dual color tiling microarray (193290 spots), i.e. i pass a 193290 x 12 matrix to vsn. i guess that's a bug, is it? Tobias ====================================================================== Dr. Tobias Straub Adolf-Butenandt-Institute, Molecular Biology tel: +49-89-2180 75 439 Schillerstr. 44, 80336 Munich, Germany
Microarray Normalization vsn Microarray Normalization vsn • 466 views
ADD COMMENT
0
Entering edit mode
@wolfgang-huber-3550
Last seen 3 days ago
EMBL European Molecular Biology Laborat…
Dear Tobias, it looks odd indeed, - can you provide me a reproducible script to reproduce your problem? As Gordon pointed out earlier, I also need the necessary input data (which does not need to be your data, but any sort of data that reproduces your problem). 1. How does the "meanSdPlot" look for "old" and "new" ? 2. How the "pairs" plot of the 12 matrix columns against each other (see man page of geneplotter::smoothScatter for an example how to make it)? 3. You can always use the old "vsn" function, in spite of the deprecation warning it is still there as always and will remain so for a while. Of course I hope this kind of problems can be sorted out with the new implementation; working hard on it. Thank you & Best wishes Wolfgang ------------------------------------------------------------------ Wolfgang Huber EBI/EMBL Cambridge UK http://www.ebi.ac.uk/huber Straub ha scritto: > Dear all, > > with the upgrade to bioc 1.5.0 I am facing vsn2. > as already discussed in an older thread there are calculation > differences. however, in my dataset(s) I get a massive outlier > comparing the array means after normalization with old vsn and vsn2. > > scatterplot here: http://server7.pictiger.com/img/198354/other/vsn- > bioc.png > > my data are six replicates of dual color tiling microarray (193290 > spots), i.e. i pass a 193290 x 12 matrix to vsn. > > i guess that's a bug, is it? > > Tobias > > > ====================================================================== > Dr. Tobias Straub Adolf-Butenandt-Institute, Molecular Biology > tel: +49-89-2180 75 439 Schillerstr. 44, 80336 Munich, Germany >
ADD COMMENT
0
Entering edit mode
Dear Wolfgang, this is how I run vsn: ##################################### library(vsn) #get the matrix of raw intensities d.raw <- as.matrix(til[,5:16]) d.norm.old <- vsn(d.raw, subsample=20000) d.norm.new <- vsn2(d.raw, subsample=20000) ##################################### I just discovered, that if I don't use subsampling the poblem gets even worse: http://www.badongo.com/pic/689985 the meanSdPlot (with subsampling) looks like this: http://www.badongo.com/pic/690005 the pairs of my matrix (pre-vsn) looks like this: http://www.badongo.com/pic/690007 If you want I can provide you the complete dataset via a local download link. best wishes Tobias On May 30, 2007, at 6:36 PM, Wolfgang Huber wrote: > > Dear Tobias, > > it looks odd indeed, - can you provide me a reproducible script to > reproduce your problem? As Gordon pointed out earlier, I also need the > necessary input data (which does not need to be your data, but any > sort > of data that reproduces your problem). > > 1. How does the "meanSdPlot" look for "old" and "new" ? > > 2. How the "pairs" plot of the 12 matrix columns against each other > (see > man page of geneplotter::smoothScatter for an example how to make it)? > > 3. You can always use the old "vsn" function, in spite of the > deprecation warning it is still there as always and will remain so > for a > while. Of course I hope this kind of problems can be sorted out > with the > new implementation; working hard on it. > > Thank you & Best wishes > Wolfgang > > ------------------------------------------------------------------ > Wolfgang Huber EBI/EMBL Cambridge UK http://www.ebi.ac.uk/huber > > > > > Straub ha scritto: >> Dear all, >> >> with the upgrade to bioc 1.5.0 I am facing vsn2. >> as already discussed in an older thread there are calculation >> differences. however, in my dataset(s) I get a massive outlier >> comparing the array means after normalization with old vsn and vsn2. >> >> scatterplot here: http://server7.pictiger.com/img/198354/other/vsn- >> bioc.png >> >> my data are six replicates of dual color tiling microarray (193290 >> spots), i.e. i pass a 193290 x 12 matrix to vsn. >> >> i guess that's a bug, is it? >> >> Tobias >> >> >> ===================================================================== >> = >> Dr. Tobias Straub Adolf-Butenandt-Institute, Molecular >> Biology >> tel: +49-89-2180 75 439 Schillerstr. 44, 80336 Munich, >> Germany >> ====================================================================== Dr. Tobias Straub Adolf-Butenandt-Institute, Molecular Biology tel: +49-89-2180 75 439 Schillerstr. 44, 80336 Munich, Germany
ADD REPLY
0
Entering edit mode
Dear Tobias, thank you. It would be helpful if you could provide me the file resulting from save(til, file="til.rda") via a download link. I will treat this confidentially if necessary. Best wishes Wolfgang ------------------------------------------------------------------ Wolfgang Huber EBI/EMBL Cambridge UK http://www.ebi.ac.uk/huber Straub ha scritto: > Dear Wolfgang, > > this is how I run vsn: > > ##################################### > library(vsn) > > #get the matrix of raw intensities > d.raw <- as.matrix(til[,5:16]) > > d.norm.old <- vsn(d.raw, subsample=20000) > d.norm.new <- vsn2(d.raw, subsample=20000) > > ##################################### > > I just discovered, that if I don't use subsampling the poblem gets even > worse: > http://www.badongo.com/pic/689985 > > the meanSdPlot (with subsampling) looks like this: > http://www.badongo.com/pic/690005 > > the pairs of my matrix (pre-vsn) looks like this: > http://www.badongo.com/pic/690007 > > If you want I can provide you the complete dataset via a local download > link. > > best wishes > Tobias > > On May 30, 2007, at 6:36 PM, Wolfgang Huber wrote: > >> >> Dear Tobias, >> >> it looks odd indeed, - can you provide me a reproducible script to >> reproduce your problem? As Gordon pointed out earlier, I also need the >> necessary input data (which does not need to be your data, but any sort >> of data that reproduces your problem). >> >> 1. How does the "meanSdPlot" look for "old" and "new" ? >> >> 2. How the "pairs" plot of the 12 matrix columns against each other (see >> man page of geneplotter::smoothScatter for an example how to make it)? >> >> 3. You can always use the old "vsn" function, in spite of the >> deprecation warning it is still there as always and will remain so for a >> while. Of course I hope this kind of problems can be sorted out with the >> new implementation; working hard on it. >> >> Thank you & Best wishes >> Wolfgang >> >> ------------------------------------------------------------------ >> Wolfgang Huber EBI/EMBL Cambridge UK http://www.ebi.ac.uk/huber >> >> >> >> >> Straub ha scritto: >>> Dear all, >>> >>> with the upgrade to bioc 1.5.0 I am facing vsn2. >>> as already discussed in an older thread there are calculation >>> differences. however, in my dataset(s) I get a massive outlier >>> comparing the array means after normalization with old vsn and vsn2. >>> >>> scatterplot here: http://server7.pictiger.com/img/198354/other/vsn- >>> bioc.png >>> >>> my data are six replicates of dual color tiling microarray (193290 >>> spots), i.e. i pass a 193290 x 12 matrix to vsn. >>> >>> i guess that's a bug, is it? >>> >>> Tobias >>> >>> >>> ====================================================================== >>> Dr. Tobias Straub Adolf-Butenandt-Institute, Molecular Biology >>> tel: +49-89-2180 75 439 Schillerstr. 44, 80336 Munich, Germany >>> > > ====================================================================== > Dr. Tobias Straub Adolf-Butenandt-Institute, Molecular Biology > tel: +49-89-2180 75 439 Schillerstr. 44, 80336 Munich, Germany > ====================================================================== >
ADD REPLY

Login before adding your answer.

Traffic: 400 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