Question: Sample-level differential expression for TCGA samples
0
gravatar for rin
4 months ago by
rin30
rin30 wrote:

Hi all!

Is anyone aware of available sample-level differential expression from TCGA? I am trying cBioPortal, but for a reason the patient IDs I am using are not recognised. As far as I saw, Firehose does not provide DE data.

Thanks in advance. R.

deg rna-seq de tcga • 243 views
ADD COMMENTlink written 4 months ago by rin30

What do you acutally want to compare?

ADD REPLYlink written 4 months ago by kristoffer.vittingseerup1.8k

I have performed already a Tumor Vs. Normal DE with EdgeR, but I want to visualize the potential expression differences in sample level in a network I have built.

Edit: I am aware that cBioPortal does not have tumor vs. normal analysis when it comes to expression.

ADD REPLYlink modified 4 months ago • written 4 months ago by rin30

Still not clear what you want to do. When you say 'sample level', do you mean that you want your network vertices to be samples and not genes? In that case, just transpose your input data with t() (in R).

ADD REPLYlink modified 4 months ago • written 4 months ago by Kevin Blighe42k

Sorry I should have made it more clear :) I don't want different expression between groups (tumor Vs normal) in general, but differential expression of genes in every patient (tumor Vs normal for patient A, patient B and so on). The reason is that after doing the differential expression analysis in patient groups from specific subtypes of TCGA-COAD, I found that a lot of critical genes, known to be differentially expressed in colon cancer, were having a logFC ~O.4 or so. So I wanted to check whether the results are "diluted" with those genes being highly differentially expressed in some of the samples while not in others.

Hope that I made it clear enough now.

ADD REPLYlink written 4 months ago by rin30
2

You could do a paired analysis in order to adjust for the within-patient effects - in this way, you would obtain a p-value and fold-change for each gene. Otherwise, if you're just comparing 1 tumour versus 1 normal, the best that you can do is derive a ratio of expression for each T-N pair, and/or convert your entire dataset to Z-scores and compare these values in each T-N pair separately.

ADD REPLYlink modified 4 months ago • written 4 months ago by Kevin Blighe42k
1

Helpful as always! Thanks for your suggestions and help, Kevin.

ADD REPLYlink written 4 months ago by rin30
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1851 users visited in the last hour