Hi,
I have a question about varianceStabilizingTransformation and creating heatmaps.
The data is without biological replicates, therefore:
Since I just had technical replicates and no biological I guess it is not a good idea to perform any differential expression analysis. However, I would like to have an heatmap to display the similarities/dissimilarities in my miRNA dataset.
So:
When creating the heatmap using:
I result in the first graph (slide 1 in "heatmaps_MMu_samples.ppt" (in attachment))
When performing varianceStabilizedTransformation for the count data, the heatmap looks better (slide 2 in "heatmaps_MMu_samples.ppt")
However, when plotting the SD of the transformed data neither log2 nor vsd seems to work for the dataset, since the SD is not constant throughout the range. Am I right? (please see graph in attachment: "vsd graph.ppt", left graph=log2, right graph=varStabTransf)
Can I conclude from this that:
a)neither log2 nor variance stabilizing trasformation should be applied for this dataset and b) I can't present the data on the heatmap using varStabTansf?
Is there any better way to show the similarities/dissimilarities in the data? Or can I use some other type of data transformation to visualize the differences on my heatmap?
Thanks to everybody for answering
I have a question about varianceStabilizingTransformation and creating heatmaps.
The data is without biological replicates, therefore:
Code:
cds0_trial=estimateDispersions(cds0_trial, method="blind", sharingMode="fit-only")
So:
When creating the heatmap using:
Code:
>select=order(rowMeans(counts(cds0_trial)), decreasing=TRUE)[1:50] > hmcol=colorRampPalette(brewer.pal(9, "GnBu"))(100) >heatmap.2(counts(cds0_trial)[select,], col=hmcol, trace="none", margin=c(10,6))
When performing varianceStabilizedTransformation for the count data, the heatmap looks better (slide 2 in "heatmaps_MMu_samples.ppt")
However, when plotting the SD of the transformed data neither log2 nor vsd seems to work for the dataset, since the SD is not constant throughout the range. Am I right? (please see graph in attachment: "vsd graph.ppt", left graph=log2, right graph=varStabTransf)
Can I conclude from this that:
a)neither log2 nor variance stabilizing trasformation should be applied for this dataset and b) I can't present the data on the heatmap using varStabTansf?
Is there any better way to show the similarities/dissimilarities in the data? Or can I use some other type of data transformation to visualize the differences on my heatmap?
Thanks to everybody for answering