PCA does not show distinction in groups

Question

[druggable]

Dear All,

I performed pca on gene expression values. However, the resulting plot does not show distinction among groups...Is it still worth doing differential expression analysis?

Thank you!

[karl.stamm]

Note your first two dimensions are capturing only 12+7% of the total variance. There's about four fifths of the whole data variation not shown yet. Looks like these samples are related, maybe the same species and tissue, and your differential gene set is a small fraction of the whole genome. That's a good thing!

[swbarnes2]

If this is just a test of ordinary humans under ordinary conditions, I'm not sure you would expect to see large global differences in gene expression caused merely by obesity. If you'd collected tissue 30 minutes after a glucose challenge, then maybe you'd see something in the right tissues.

Answer 1

[swbarnes2]

Short answer: yes. The people who spent the time and money on the experiment deserve an answer.

You might have a a real effect in a small number of genes.

[druggable]

It turns out there were no significantly differentially expressed genes in my data at adjusted pvalue <0.05. I would still like to generate some patient-specific profiles. Maybe I can still do some analysis on my dataset with active module identification methods?