[BioC] HTqPCR: Calculate CV between replicates with in a plate

[Heidi Dvinge]

Hello Jeremy,

there's no default way for calculating within-plate CVs in HTqPCR, since
it'll depend on the exact layout of the plate. Below is an example of how
this can be done, using the dataframe "design" to indicate how wells on
the plate belong together. In your case it might be different of course.
How is this information stored in the input data files and/or the qPCRset
object?

> library(HTqPCR)
> # Use test data from the package as example here, with 6 plates
> data(qPCRraw)
> sampleNames(qPCRraw)	<- paste("Plate", 1:6, sep="")
> # What's the order of genes/rep/samples
> design	<- data.frame(Sample=paste("S", rep(1:4, each=96), sep="_"),
+ 		Gene=paste("Gene", rep(rep(1:32, each=3),4), sep="_"),
+ 		Replicate=paste("Rep", rep(1:3, 128), sep="_"))
> # Calculate the CV
> sd.gene	<- aggregate(exprs(qPCRraw), by=list(design$Sample,
design$Gene), sd)
> mean.gene <- aggregate(exprs(qPCRraw), by=list(design$Sample,
design$Gene), mean)
> cv.gene	<- sd.gene[,-c(1:2)]/mean.gene[,-c(1:2)]
> rownames(cv.gene)	<- paste(sd.gene[,1], sd.gene[,2], sep=":")

HTH
\Heidi


> Dear list,
>
> I see that the package considers each plate as different sample and
> calculates CV  for genes across plates (samples) but not with in a
> plate.
> Unfortunately my plate design is different. I have 384 well plates
> with 3 replicates for 32 genes in first 4 rows of the plate. Likewise
> other 3 samples in the next 3 sections of 4 rows. So in total I have 4
> different clinical samples on a single plate.
> 32genes*3replicates*4samples=384 wells.
> Now I have gene 1 for sample 1 in first 3 wells of a plate. I want to
> calculate CV for this gene in 3 wells. Similarly for the remaining 32
> genes of the sample. Like wise I need to calculate CV for the same
> genes in 3 more samples in the same plate. I see that the package has
> no function to calculate CVs for plate in this pattern. My apologies
> if I am missing any thing. I would really appreciate any suggestions
> or else I will have to write my own script to analyze my Hi-Throughput
> data.
>
> Thank you
>
> _______________________________________________
> 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
>