Hi folks, plenty of papers have been published about different normalization methods for RNA-seq data like TPM or RPKM and meaning of these papers (in my opinion) is that none of such approach is optimal. I am not talking about classical DE analysis by edgeR od DEseq2, because it is pretty clear in most cases. One organism, more conditions (control and some experimental), comparison against the control group and I can see up and down-regulated genes. OK.
So, my question is hypothetical and did not find a satisfactory answer (maybe I did not read the right paper or thread on some forum). Try to imagine that you have plenty of totally different organisms - human, mouse, fruit fly, zebrafish, Caenorhabditis elegans (roundworm) and some another non-model invertebrate organism. For all them, I have a control group and experimental group - hmm, I don't, maybe with an injection of some amount of lead. And I know (again, only hypothetically), that all organisms in my experiment have gene XYZ which is responsible for lead removal from the body. Classical DE analysis will tell me how many times is the expression of XYZ gene up-regulated in the experimental for each of my organism but that information is not transferable. And I would like to get more global comparison how the expression of XYZ gene is different across all groups in my all organisms and compare everything with everything.
For example. I can take TPM values, put them to the graph and it will give me exactly what I want to see. I will see that the expression of XYZ gene is relatively low for control groups in the range 10 to 20 TPM and in experimental groups quite high about 900 but for some organism, it is around 1,500 and for other ones about 500. And now I can discuss DE of XYZ gene... But, is it nonsense/acceptable/correct? According to recent publications not really, but what is the correct method? Is it eve possible to do analysis like this?
Let's assume that amount of RNA was the same for all biological samples, biological replicates are fine etc.
Let me know what do you think. Thank you in advance. Cheers, Jirka
