1
$\begingroup$

I have count data from a RNAseq experiment (2 samples are from normal cells and 3 samples are cells with a disease), and the data is already standardized by trimmed mean of M values (TMM). I want to do some plots: biplot of Principal Component Analysis (PCA) and a cluster dendrogram to see if the samples normal/disease are well separated (there is a clear difference between them). Since the data is already standardized (TMM) should I scale and center the data prior to perform PCA and cluster dendrogram??

thank you!

Improve this question
$\endgroup$
2
  • $\begingroup$ by TMM you meaning running limma or edgeR on it i supposed? Or obtaining the cpms? $\endgroup$
    – StupidWolf
    Oct 7, 2020 at 12:58
  • $\begingroup$ Using kallisto I get the TMM, and before looking for differentially expressed genes with edgeR I am doing PCA and cluster dendogram to look into the data. $\endgroup$
    – Mee
    Oct 8, 2020 at 8:19

1 Answer 1

Reset to default
1
$\begingroup$

Scaling (or centering) makes the genes comparable: Putting the expression levels of genes in the same scale (i.e. between 0 and 1) sustains that all of your genes contribute equally to the PCA or distance calculations. On the other hand, without this step, such calculations would be dominated by the highly/lowly expressed genes. I get "better" results with scaling.

Improve this answer
$\endgroup$
3
  • $\begingroup$ Yes, I understand that. But the data has already been standardized using by trimmed mean of the value. And when performing PCA with and without scaling the results are pretty similar. I am wondering if the scaling and centering are not needed because the data has already been standardized. $\endgroup$
    – Mee
    Oct 8, 2020 at 8:16
  • $\begingroup$ What TMM does is to "make different samples comparable" by adjusting for library size and more. It does not deal with individual genes. So yes, scaling would help when expression levels of different genes vary a lot (it is often the case). Since you are using edgeR, you can use the function cpm() to calculate log2 of CPM and then scale these before clustering for example. $\endgroup$
    – haci
    Oct 8, 2020 at 9:30
  • $\begingroup$ ok, thank you for your answer now I undestand. But, in terms of using the cpm() function: is it ok to use it over the data that is already TMM? Thanks $\endgroup$
    – Mee
    Oct 8, 2020 at 18:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.