I’ve used the same normalization and chip to search a gene expression in R2 Genomics Analysis and Visualization Plataform, but I have compared very different cells (like lung cancer and melanoma). So my doubt is: is it correct/safe to say that a gene is more expressed in lung cancer, for example, than in melanoma (p
What is the normalization based on?
I think differences in gene expression between tissues are tricky on two levels.
In the first place, there is the normalization issue. In my experience, if you test for differential gene expression between tissues, a large proportion of the transcriptome comes out DEG, and it all depends on the normalizations used. So for a given gene: do you consider its expression relative to the rest of the transcriptome? Dangerous, because highly expressed genes in the background may "dilute" your transcript of interest, even if stoichiometrically, it is "upregulated" relative to the genes it interacts with. Or do you normalize relative to housekeeping genes? It is known that housekeeping genes exhibit tissue-specific expression. Do you consider it relative to the genes it interacts with? Interesting, but requires a lot of prior knowledge...
Secondly, there is also the question of what it actually means to be differentially expressed between tissues. Transcripts (and/or their corresponding products) hardly ever function on their own. They interact with the rest of the transcriptome/proteome, causing feedback to gene expression etc. These "backgrounds" are drastically different between cell types. Also, the same amount of mRNA may lead to different amounts of protein depending on the status of the cell (presence/activation of ribosomes, tRNA,...). As a result, the impact of a change in the number of copies of a transcript may be enormous in one tissue, versus negligible in the other.
These are actually considerations that to a large extent hold true even when comparing gene expression between conditions in the same tissue, but obviously the more different your samples are to start with, the more exaggerated the effects will be. So, my two cents is: be careful, it's a dangerous comparison. I would rather try healthy lung vs lung cancer and healthy skin vs melanoma.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools