Finding the Similarity/dissimilarity of RNAseq samples using correlation

[dena.dinesh]

Hi,

I am working on RNAseq data for 8 samples which biological duplicates. I want to find how close the samples are with respect to each other. So I obtained the count data and loaded the file into R. I loaded the library DESeq2 and later converted the count data into DeseqcounzTable and later into Dseq2 using following commands

ddsFullCountTable <- DESeqDataSetFromMatrix(countData = rnaseqMatrix,colData = conditions,design = ~ conditions)
dds = DESeq(ddsFullCountTable)

Later I transformed the dds to variance stabilizing transformation usig following command

vsd=varianceStabilizingTransformation(dds)

As I wanted to find the how similar/dissimilar the samples, I found the correlation matrix for the vsd transformed data and generated heatmap using following command,

heatmap.2(cor(assay(vsd)), col = redgreen(75), key =TRUE, symkey= FALSE, density.info = "none", trace = "none", margins = c(10,10))

Is my approach to use correlation matrix on vsd transformed data to find the samples similarity/dissimilarity is right?

[Michael Love]

You can calculate correlations, or if you take a look at the vignette, we also show how to calculate Euclidean distances of the VST data and perform hierarchical clustering.

[dena.dinesh]

You can calculate correlations, or if you take a look at the vignette, we also show how to calculate Euclidean distances of the VST data and perform hierarchical clustering.

Thanks for your reply. So if I take correlation on vst transformed data or calculate the the Euclidean distances of vst data and perform hierarchical clustering the resulting dendograms should be the same, isnt?

Kindly guide me