Question: Plotting Expression of Differential Transcripts plotDEXseq
0
3 months ago by
rbenel0
rbenel0 wrote:

Hi,

I was wondering which counts are appropriate to use for plotting the expression of the differential transcripts? plotDEXSeq is giving me this error, even when using the pre-constructed matrix of counts from the workflow "Swimming downstream: statistical analysis of differential transcript usage following Salmon quantification":

plotDEXSeq( dxr2, "ENSG00000000457.13", legend=FALSE, cex.axis=1.2, cex=1.3,
lwd=2 )
#legend is false because we have not imported everything from the txdf.  
Error in plot.window(xlim = c(0, 1), ylim = c(0, max(matr))) :
need finite 'ylim' values
1: In max(coeff, na.rm = TRUE) :
no non-missing arguments to max; returning -Inf
2: In max(matr) : no non-missing arguments to max; returning -Inf

I am interested in outputting all normalized transcripts that I found to be to differentially expressed.

Further, I have looked at the function plotDEXseq and it uses featureCounts, which is just like requesting in Deseq2, counts(object, normalized = TRUE) of the specific groupID. I can't help but wondering if these counts are OK to graphically represent a *transcript* that was found to be differential expressed following DEXseq and StageR?

Thank You

rnaseqdtu stager dexseq • 193 views
modified 3 months ago by Michael Love21k • written 3 months ago by rbenel0
Answer: Plotting Expression of Differential Transcripts plotDEXseq
0
3 months ago by
Michael Love21k
United States
Michael Love21k wrote:

plotDEXSeq is fairly specific to the exon-based analysis. Our use of DEXSeq for DTU worked well, but it doesn't necessarily have the built in functions for plotting the final results. I think the plotting function from DRIMSeq is more appropriate, because it plots the proportions for each transcript within the gene. plotProportions requires that you have run dmPrecision and dmFit, which perhaps takes too much time if you just want to plot the proportions for genes. I will try to work on finding an easier way to plot these, but if you look in this section* there is some code for quickly estimating proportions for all transcripts. You can use this to get the proportions and then plot them across samples with, e.g. barplot(prop[idx,], col=condition). The simplest way to plot all transcripts for a gene would be just using par(mfrow=...) and multiple barplots, but you could also use ggplot2 to make a prettier version, like plotProportions.

Hi Mike,

I agree with you regarding the plotting function of DRIMSeq, especially the plotProportions function; however, when I try to run dmPercision on my entire dmDSdata  and not a subset I receive the error below:

d
An object of class dmDSdata
with 11120 genes and 32 samples
* data accessors: counts(), samples()
d <- DRIMSeq::dmPrecision(d, design=design_full)

! Using a subset of 0.1 genes to estimate common precision !

Error in optimHess(par = par, fn = dm_lik_regG, gr = dm_score_regG, x = x,  :
non-finite value supplied by optim
In addition: There were 50 or more warnings (use warnings() to see the first 50)

My recommendation for now is to make simple plots on your own, and not to run DRIMSeq all over again if you've used DEXSeq.

OK. So that takes us back to my initial question: Which counts are appropriate to use for plotting the expression of the differential transcripts? Would this be correct to use, following DE of a transcript(s) after running DEXSeq and StageR

count <- featureCounts(dxd, normalized = TRUE)[idx, ]

1

I’d recommend using scaledTPM from tximport for the reasons described in the workflow.