[BioC] marray and limma error with lmFit

[Paquet, Agnes]

Dear Birger,
 
I just tried the code from the book on R 2.4.0 and R 2.5.0 and I was not able to reproduce the error you are mentionning. My Bioconductor installation on R2.4.0 is the following:
 
> sessionInfo()
R version 2.4.0 (2006-10-03) 
i386-pc-mingw32 
locale:
LC_COLLATE=English_United States.1252;LC_CTYPE=English_United States.1252;LC_MONETARY=English_United States.1252;LC_NUMERIC=C;LC_TIME=English_United States.1252
attached base packages:
 [1] "grid"      "tools"     "tcltk"     "methods"   "stats"    
 [6] "graphics"  "grDevices" "utils"     "datasets"  "base"     
other attached packages:
       beta7 arrayQuality RColorBrewer     gridBase       hexbin 
     "0.6.0"     "1.11.3"      "0.2-3"      "0.4-3"      "1.8.0" 
  colorspace      lattice      convert       marray    tkWidgets 
      "0.95"     "0.14-9"      "1.8.0"     "1.13.5"     "1.12.1" 
      DynDoc  widgetTools      Biobase        limma 
    "1.12.0"     "1.10.0"     "1.12.2"      "2.8.1" 
> 

Could you please email us the result of sessionInfo() on your machine so that we can try to reproduce the error you noticed?
 
The package limma has changed since the book was published, and the column header of the result table generated by topTable are different from the book example. You can find out more on the output from topTable by looking at the help file for this function: ?topTable.
The "P.Value" column that you can see in the book example is the p value for each gene, adjusted for multiple testing. The corresponding column of the new version of topTable would be the column "adj.P.Val", generated using Bonferroni correction.
restable=topTable(LMres,number=10,resort.by="M",adjust.method="bonferroni")
 
Best,
 
Agnes
 

________________________________

From: bioconductor-bounces at stat.math.ethz.ch on behalf of Hauchecorne Birger
Sent: Thu 4/19/2007 6:11 AM
To: bioconductor at stat.math.ethz.ch
Cc: birgerhauchecorne at hotmail.com
Subject: [BioC] marray and limma error with lmFit



Hi,

I've got a little problem with lmFit and I was hoping you could help me.

I'm following a book in order to analyse microarray data. The book I follow is "Bioinformatics and Computational Biology Solutions Using R and Bioconductor" by Gentleman et al.. More precise chapter 4 "Preprocessing Two-Color Spotted Arrays" by Yang and Paquet.
In order to get used to the protocol, I followed the example used in the book.
These are the procedures used:
>library(beta7)
>library(arrayQuality)
>Targetinfo<-read.marrayInfo("TargetBeta7.txt")
>mraw<-read.GenePix(targets=Targetinfo)
>normdata<-maNorm(mraw)
>library(convert)
>mdata<-as(normdata,"exprSet")

And then in the book, they used this:
>LMres<-lmFit(normdata,design=c(1,-1,-1,1,1,-1),weights=NULL)
Error in fit$Amean <- rowMeans(unwrapdups(object at maA, ndups = ndups, spacing = spacing),  :
        object "fit" not found

giving me an error.

I searched for this in the net, and found a solution from you
>LMres<-lmFit(as(normdata,"MAList"),design=c(1,-1,-1,1,1,-1),weights=NULL)

So I could continue with
>LMres<-eBayes(LMres)
>restable<-topTable(LMres,number=10,resort.by="M")
Error in order(na.last, decreasing, ...) :
        argument 1 is not a vector

giving me another error. I can't sort by M (a parameter I need)
By sorting on another parameter ("T" for example), I could generate a list nonetheless. In the book there was a solution for the analysis, so I checked it with the list I found. Oddly enough the values found for "M" in the book where equal to my "logFC" values, unfortunately I couldn't sort on "logFC".

Another odd thing that I found was the the resulting P.Value from the book (which was 1 for the significant cases), was different from my P.Values (although the genes are the same). Most of the time where it was supposed to be 1, I became 0. Unfortunately again, not all the time, so it was impossible to tell where it was correct.

I'm now thinking that the cause of this problem can be found in the solution you guys gave for the first problem (>LMres<-lmFit(as(normdata,"MAList"),design=c(1,-1,-1,1,1,-1),weights=NULL))

I was hoping you could help me with my problem why these P.values are different, and where is my "M" column?


I'm using R 2.4.1. on windows XP having installed bioconductor.

Kind regards and thanks in advance,

Birger Hauchecorne
Senior Student Bio-Engineering at University of Antwerp, Belgium



        [[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