[BioC] affy: Trouble Creating Custom Processing Methods for summary expression

[Ben Bolstad]

Try renaming "computeExprVal.huber" to "generateExprVal.method.huber".
That should make it work. I think summary methods for expresso need to
have the "generateExprVal.method." prefix. I will fix the vignette
accordingly.

Ben






On Fri, 2005-06-10 at 16:52 -0400, Alexander C Cambon wrote:
> I am trying to use the "Custom Processing Methods" available in package "affy" to create new expression methods. But for some reason I am getting "NA"'s for the expression values. I start out using 4 cel files with rae230a data. 
> 
> > x
> AffyBatch object
> size of arrays=602x602 features (11330 kb)
> cdf=RAE230A (15923 affyids)
> number of samples=4
> number of genes=15923
> annotation=rae230a
> 
> ## And I proceed by using the example in the "Custom Processing Methods (How To)" vignette on page 4:
> 
>  > computeExprVal.huber <- function(probes) {
> +  res <- apply(probes, 2, huber)
> + mu <- unlist(lapply(res, function(x) x$mu))
> + s <- unlist(lapply(res, function(x) x$s))
> + return(list(exprs = mu, se.exprs = s))
> + }
> 
> >generateExprSet.methods <- c(generateExprSet.methods, "huber")
> 
> ###Then I use the expresso function... and get an error message (see below)
> 
> >eset.bg.huber<-expresso(x, bgcorrect.method="mas",  pmcorrect.method="pmonly",  
> + summary.method="huber")
> background correction: mas 
> PM/MM correction : pmonly 
> expression values: huber 
> background correcting...done.
> normalizing...done.
> Error in match.arg(summary.method, express.summary.stat.methods) : 
>         'arg' should be one of avgdiff, liwong, mas, medianpolish, playerout
> 
> ## So, after looking at the error, I decide to add (perhaps wrongly)  the line
> 
> > express.summary.stat.methods<-c(express.summary.stat.methods,"huber")
> 
> ## and I rerun...
> 
> >eset.bg.trm<-expresso(x, bgcorrect.method="mas",  pmcorrect.method="pmonly",  
> +  summary.method="huber")
> 
> 
> ## Now I get:
> 
> background correction: mas 
> normalization: 
> PM/MM correction : pmonly 
> expression values: huber 
> background correcting...done.
> normalizing...done.
> 15923 ids to be processed
> |                    |
> |####################|
> 
> ## This looks reassuring, and I start  thinking I am over the hump. However when I ask for the first few lines, I get all ##"NA"'s for expression ..
> 
>  > exprs(eset.bg.huber)[1:10,1:4]
>            AK_0A.CEL AK_0B.CEL AK_4A.CEL AK_4B.CEL
> 1367452_at        NA        NA        NA        NA
> 1367453_at        NA        NA        NA        NA
> 1367454_at        NA        NA        NA        NA
> 1367455_at        NA        NA        NA        NA
> 1367456_at        NA        NA        NA        NA
> 1367457_at        NA        NA        NA        NA
> 1367458_at        NA        NA        NA        NA
> 1367459_at        NA        NA        NA        NA
> 
> ###But when I do the same thing, using a "built-in" expresso method (using "x"), I get 
> 
>  > exprs(eset44)[1:10,1:4]
>             AK_0A.CEL AK_0B.CEL AK_4A.CEL AK_4B.CEL
> 1367452_at  564.68667 400.93697 476.22935 396.00641
> 1367453_at  330.41054 281.98687 279.02867 305.24909
> 1367454_at  158.99028 165.78677 134.08273 151.77046
> 1367455_at  302.39252 223.56433 192.13621 204.74223
> 1367456_at  881.07080 931.81819 792.18972 583.09672
> 1367457_at  139.75233 142.67949 140.01944 153.02465
> 1367458_at  144.34063 154.16194 176.33189 190.89218
> 1367459_at 1010.72613 709.47988 886.92512 957.64834
> 1367460_at   48.19831  51.70533  59.43114  54.13907
> 1367461_at  213.03203 194.16510 210.99492 182.22481
> 
> Any ideas what I am doing wrong?
> 
> Alex
> 
> 
> 
> Alexander Cambon
> Biostatistician
> School of Public Health
> Dept of Biostatistics and Bioinformatics
> University of Louisville
> 
> R 2.1.0
> Bioconductor 1.6
> Windows XP
> Dell Optiplex
> 1 GB RAM
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
-- 
Ben Bolstad <bolstad at stat.berkeley.edu>
http://www.stat.berkeley.edu/~bolstad