[BioC] Artifact in RMA-normalized data
Rafael A. Irizarry
ririzarr at jhsph.edu
Thu Mar 4 23:07:58 MET 2004
try it again using the default version of rma... quantiles instaed of
quantiles.robus
> allnorm<-expresso(alldata, bgcorrect.method='rma',
> normalize.method='quantiles', pmcorrect.method='pmonly',
> summary.method='me
> dianpolish')
On Thu, 4 Mar 2004, Mary Putt wrote:
> Hi,
> I normalized the data from 13 arrays (6 group H and 7 group P) using
> rma. I found that the arrays from my H group were systematically lower
> than from the P group at the lower end of the expression scale, while my
> arrays from the H group were higher than the arrays from the P group at
> the high end of the scale. The differences are subtle but they show up
> in the MVA plots, as well as in the summary statistics seen below. I had
> a Warning message from Affy during the normalization. It doesn't seem to
> me that RMA should introduce this type of artifact--unless there's
> something about the warning message that I don't understand. I'm
> wondering if anyone has insights on this. Thanks Mary
>
> ########
> #Program to normalize the data
> ############
>
> library(affy)
> load('all.Rdata')
> exprs.allnorm<-exprs(allnorm)
> allnorm<-expresso(alldata, bgcorrect.method='rma',
> normalize.method='quantiles.robus', pmcorrect.method='pmonly',
> summary.method='me
> dianpolish')
> save(exprs.allnorm, file='exprs.allnorm.Rdata')
> #########
> #Warning msg following normalization
> ############
>
> #> source('expresso.all.r')
> #background correction: rma
> #normalization: quantiles.robus
> #PM/MM correction : pmonly
> #expression values: medianpolish
> #background correcting...done.
> #normalizing...Chip weights are 1 1 1 1 1 1 0 1 1 1 1 1 1 1
> #Chip weights are 1 1 1 1 1 1 1 1 0 1 1 1 1 1
> #done.
> #22283 ids to be processed
> #.........
> #Warning messages:
> #1: the condition has length > 1 and only the first element will be used
> in: if (remove.extreme == "variance") {
> #2: the condition has length > 1 and only the first element will be used
> in: if (remove.extreme == "mean") {
> #3: the condition has length > 1 and only the first element will be used
> in: if (remove.extreme == "both") {
> #4: the condition has length > 1 and only the first element will be used
> in: if (remove.extreme == "variance") {
> #5: the condition has length > 1 and only the first element will be used
> in: if (remove.extreme == "mean") {
> #6: the condition has length > 1 and only the first element will be used
> in: if (remove.extreme == "both") {
>
> #####################
> #descriptive statistics of normalized data
> #
> #note that h and p are different groups
> ##################
> >
> > summary
> h1 h2 h3 h4 h5 h6 p1 p2
> p3 p4
> Min 2.997 3.008 3.051 3.010 2.967 3.005 3.123 3.057 3.119
> 3.102
> 1stQrtl 4.719 4.679 4.762 4.739 4.771 4.771 4.895 4.717 4.926
> 4.891
> Median 5.924 5.901 5.950 5.970 5.970 5.961 6.015 5.942 6.015
> 6.018
> Mean 6.165 6.143 6.150 6.171 6.182 6.178 6.167 6.163 6.172
> 6.162
> 3rdQrtl 7.291 7.300 7.266 7.358 7.316 7.288 7.201 7.281 7.216
> 7.224
> Max 13.310 13.620 13.760 13.800 13.660 13.660 13.800 13.790 13.670
> 13.660
> p5 p6 p7
> Min 3.121 3.017 3.041
> 1stQrtl 4.938 4.829 4.835
> Median 6.031 5.993 6.015
> Mean 6.172 6.166 6.168
>
> >apply(summary[,1:6], 1, median)
> Min 1stQrtl Median Mean 3rdQrtl Max
> 3.0065 4.7505 5.9555 6.1680 7.2955 13.6600
> > apply(summary[,7:13], 1, median)
> Min 1stQrtl Median Mean 3rdQrtl Max
> 3.102 4.891 6.015 6.167 7.220 13.660
> > apply(summary[,1:6], 1, mean)
> Min 1stQrtl Median Mean 3rdQrtl Max
> 3.006333 4.740167 5.946000 6.164833 7.303167 13.635000
> > apply(summary[,7:13], 1, mean)
> Min 1stQrtl Median Mean 3rdQrtl Max
> 3.082857 4.861571 6.004143 6.167143 7.227143 13.602857
>
>
>
>
> --
> Mary E. Putt
> Assistant Professor of Biostatistics
> Department of Biostatistics and Epidemiology
> Center for Biostatistics and Epidemiology
> School of Medicine, University of Pennsylvania,
> 621 Blockley Hall
> 423 Guardian Drive
> Philadelphia, PA
> 19104-6021
>
> Ph. (215) 573-7020
> Fax (215) 573-4865
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor
>
More information about the Bioconductor
mailing list