[BioC] Post-justRMA "quality control"

James MacDonald jmacdon at med.umich.edu
Wed Aug 20 12:58:58 MEST 2003

The hist and boxplot functions you are trying to use are intended for
AffyBatch objects, not exprSet objects, which is what you have after
running RMA.

Also, I don't think you will be able to simply normalize an AffyBatch
if you use the default quantile normalization. I reported a bug in
normalize.AffyBatch.quantiles about two months ago, where I couldn't
normalize 9 U-133A chips with 2 Gb RAM. To my knowledge this has not
been fixed. You might try normalize(abatch, method="invariantset") or
one of the other normalization procedures.



James W. MacDonald
Affymetrix and cDNA Microarray Core
University of Michigan Cancer Center
1500 E. Medical Center Drive
7410 CCGC
Ann Arbor MI 48109

>>> "Michael Barnes" <Michael.Barnes at cchmc.org> 08/20/03 11:23AM >>>
I am trying to follow the "Load Data" Vignette in the section entitled
"quality control through data exploration".  I am using R1.7.1 and
updated Bioconductor Tuesday (yesterday) on a Windows XP machine.  I
have 23 U133A chips that I am trying to examine.

As I followed the process I was able to do all the actions (histogram,
boxplot, RNA degradation plot) on my dataset. 
> data<-ReadAffy()

> data
AffyBatch object
size of arrays=712x712 features (91097 kb)
cdf=HG-U133A (22283 affyids)
number of samples=23
number of genes=22283

I then wanted to see what my data looked like following normalization
(similar to what is described from pg 23 on).  However, due to memeory
issues I could not run the example in a manner similar to that

> normalized.data<-normalize(data[1:23])
Error: cannot allocate vector of size 91091 Kb

 so  I ran justRMA

> normalized.data<-justRMA()
Background correcting
Calculating Expression

> sampleNames(normalized.data)
 [1] "1082_U133A.CEL"       "1083_HG_U133A.CEL"    "1085_U133A.CEL"   

 [4] "1087_U133A.CEL"       "109335_U133A.CEL"     "109338_U133A.CEL" 

 [7] "109341_U133A.CEL"     "109404_HG_U133A.CEL"  "1095_U133A.CEL"   

[10] "110256_U133A.CEL"     "110259_U133A.CEL"    
[13] "110443_HG_U133A.CEL"  "110444_HG_U133A.CEL" 
[16] "1349_U133A.CEL"       "1460_U133A.CEL"      
[19] "7113.3_HG_U133A.CEL"  "7118.3_HG_U133A.CEL" 
[22] "813.31_HG_U133A.CEL"  "F1089_U133A.CEL"     

> normalized.data
Expression Set (exprSet) with 
        22283 genes
        23 samples
                 phenoData object with 1 variables and 23 cases
                sample: arbitrary numbering

Now I can not get boxplot or histogram.  I have not tried RNA
degradation plot.

> boxplot(normalized.data,col=c(2,2,3,3))
Error in boxplot.default(normalized.data, col = c(2, 2, 3, 3)) : 
        invalid first argument

> hist(normalized.data[1:2])
Error in hist.default(normalized.data[1:2]) : 
        `x' must be numeric

If Ihave created an incompatible objecct with justRMA, how do I create
the correct object type?  If not, what suggestions do you have.

Michael Barnes, Ph.D.

Bioconductor mailing list
Bioconductor at stat.math.ethz.ch 

More information about the Bioconductor mailing list