[BioC] HTqPCR problems

[Simon Melov]

Thanks for the help Heidi,
but I'm still having troubles, your comments on the plotting helped me solve the outputs. But if I want to just display some groups (for example the LO group in the example below), how do I associate a group with multiple samples (ie biological reps)?

Currently I'm associating genes with samples  by reading in the file as below
plate6=read.delim("plate6Sample.txt", header=FALSE)
#this is a file to associate sample ID with the genes in the biomark data, as currently HTqPCR does not seem to associate the sample info in the Biomark output to the gene IDs 

samples=as.vector(t(plate6))
raw6=readCtData(files="Chip6.csv", format="BioMark", n.features=48, n.data=48, samples=samples)
#now I have samples and genes similar to your example in the guide, but I want to associate samples to groups now. In the guide, you have an example where you have entire files as distinct samples, but in our runs, we never have that situation. I have a file which associates samples to groups, which I read in...

groupID=read.csv("plate6key.csv")

but how do I associate the samples with their appropriate groups for biological replicates with any of the functions in HtQPCR? 
You recommend below using a vector, but I dont see how that helps me associate the samples in the Expression set.

thanks again!

s

On Jun 26, 2012, at 12:48 PM, Heidi Dvinge wrote:

>> Hi,
>> I'm having some troubles selectively sub-setting, and graphing up QPCR
>> data within HTqPCR for Biomark plates (both 48.48 and 96.96 plates). I'd
>> like to be able to visualize specific genes, with specific groups we run
>> routinely on our Biomark system. Typical runs are across multiple plates,
>> and have multiple biological replicates, and usually 2 or more technical
>> replicates (although we are moving away from technical reps, as the CVs
>> are so tight).
>> 
>> Can anyone help with this? Heidi?
>> 
>> raw6=readCtData(files="Chip6.csv", format="BioMark", n.features=48,
>> n.data=48, samples=samples)
>> #Ive read the samples in from a separate file, as when you read it in, it
>> doesnt take the sample names supplied in the biomark output#
>> #Next, I want to plot genes of interest, with samples of interest, and I'm
>> having trouble getting an appropriate output#
>> 
>> g=featureNames(raw6)[1:2]
>> plotCtOverview(raw6, genes=g, groups=groupID$Treatment, col=rainbow(5))
>> 
>> #This plots 1 gene across all 48 samples#
>> #but the legend doesnt behave, its placed on top of the plot, and I cant
>> get it to display in a non-overlapping fashion#
>> #I've tried all sorts of things in par, but nothing seems to shift the
>> legend's position#
>> 
> As the old saying goes, whenever you want a job done well, you'll have to
> do it yourself ;). In this case, the easiest thing is probably to use
> legend=FALSE in plotCtOverview, and then afterwards add it yourself in the
> desired location using legend(). That way, if you have a lot of different
> features or groups to display, you can also use the ncol parameter in
> legend to make several columns within the legend, such as 3x4 instead of
> the default 12x1.
> 
> Alternatively, you can use either xlim or ylim in plotCtOverview to add
> some empty space on the side where there's then room for the legend.
> 
>> #I now want to plot a subset of the samples for specific genes#
>>> LOY=subset(groupID,groupID$Treatment=="LO" | groupID$Treatment== "LFY")
>>> LOY
>>   Sample Treatment
>> 2     L20       LFY
>> 5     L30       LFY
>> 7     L45        LO
>> 20    L40        LO
>> 27    L43        LO
>> 33    L29       LFY
>> 36    L38        LO
>> 40    L39        LO
>> 43    L23       LFY
>> 
>> 
>>> plotCtOverview(raw6, genes=g, groups=LOY, col=rainbow(5))
>> Warning messages:
>> 1: In split.default(t(x), sample.split) :
>>  data length is not a multiple of split variable
>> 2: In qt(p, df, lower.tail, log.p) : NaNs produced
>>> 
> 
> Does it make sense if you split by groups=LOY$Treatment? It looks like the
> object LOY itself is a data frame, rather than the expected vector.
> 
> Also, you may have to 'repeat' the col=rainbow() argument to fit your
> number of features.
> 
>> 
>> #it displays the two groups defined by treatment, but doesnt do so nicely,
>> very skinny bars, and the legend doesnt reflect what its displaying#
>> #again, I've tried monkeying around with par, but not sure what HTqPCR is
>> calling to make the plots#
>> 
> If the bars are very skinny, it's probably because you're displaying a lot
> of features. Nothing much to do about that, except increasing the width or
> your plot :(.
> 
> \Heidi
> 
>> please help!
>> 
>> thanks
>> 
>> Simon.
>> 
>> _______________________________________________
>> Bioconductor mailing list
>> Bioconductor at r-project.org
>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>> Search the archives:
>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>> 
> 
>