[BioC] single channel analysis

[James W. MacDonald]

Hi David,

David martin wrote:
> Hi,
> I'm have a custom array design with several blocks and each spot in 
> duplicate. I'm running a single channel experiment. Each sample being 
> labeled with the same dye.
> 
> My problem is that when spots are assigned  weight=0 (discarded) they 
> still all appear in the fitted object. I though that assigning a weight 
> of 0 would discard this spots (would be removed from thh analysis). In 
> the documentation this seems to be true for withinarraynormalization 
> SInce this is not the case, how can i remove all these spots ??

I think you misunderstand the documentation (and the basic idea behind 
weighting data). It never says that data with a weight = 0 are 
discarded. Instead, it says that downstream functions will use these 
weights when analyzing the data.

Since the weights for certain spots are zero, you will effectively 
remove those spots from consideration when normalizing, fitting models, 
etc, but they are not removed from the fitted object.

Best,

Jim


> 
> Here is the code:
> 
> 
> #
> # Load libraries
> #
> library(limma)
> 
> # This defines the column name of the mean Cy5 foreground intensites
> Cy5 <- "F635 Mean"
> 
> # This defines the column name of the mean Cy5 background intensites
> Cy5b <- "B635 Mean"
> 
> 
> # Read the targets file (see limmaUsersGuide for info on how to create 
> this)
> targets <- readTargets("targets.txt")
> 
> 
> #Read gpr files and weight negative spots as 0 for spots with Flags -50.
> RG <- read.maimages(targets$FileName,
>       source="genepix",
>       columns=list(R=Cy5,G=Cy5, Rb=Cy5b,Gb=Cy5b),
>       annotation = c("Block", "Column", "Row", "ID", "Name","Flags"),
>       wt.fun=wtflags(weight=0,cutoff=-50),
>       )
> 
> # remove the extraneous green channel values
> RG$G <- NULL
> RG$Gb <- NULL
> 
> #Read spotypes and assign controls
> spottypes<-readSpotTypes("spottypes.txt")
> RG$genes$Status<-controlStatus(spottypes,RG$genes)
> 
> #Do background correction
> bRG <- backgroundCorrect(RG$R,method='normexp')
> 
> #Normalize
> MA <- normalizeBetweenArrays(log2(bRG), method="quantile")
> 
> #Handle duplicates spots
> corfit <- duplicateCorrelation(MA,ndups=2,spacing=1)
> 
> fit<-lmFit(MA,correlation=corfit$consensus.correlation,weights=w,ndups=2,genelist=RG$genes$Name) 
> 
> fit<-eBayes(fit)
> topTable(fit,genelist=RG$genes$Name,number=NULL)
> 
> _______________________________________________
> 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

-- 
James W. MacDonald, M.S.
Biostatistician
Douglas Lab
University of Michigan
Department of Human Genetics
5912 Buhl
1241 E. Catherine St.
Ann Arbor MI 48109-5618
734-615-7826
**********************************************************
Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues