Hello,
I am new to this forum so please forgive me if my question is a bit naive :)
- I have several tissues: A, B and C (and 2 replicates per tissue)
- I have many genes: gene1a, gene1b, gene2, gene3...
- genes 1a and 1b are paralogs
In edegR I can do multiple comparisons of gene expression for each of those 4 genes between the different tissues and the different replicates, which is great!
What I would like now, is to compare the expression level between gene1a and gene1b, in tissue A, B, C and in all the tissues. I would thus compare the gene expression level of the two copies. So this is not DE for one gene between samples, but between 2 genes among multiple samples... I wanted at first to use the counts but I fear that there will be a normalization problem at some point!
I hope I am clear, and if anyone has any idea of how to proceed (with edgeR ideally), that would be much appreciated! :)
Thank you very much,
Thibault Lorin
University of Basel
Thank you very much!
1. So do you mean it is better to use FPKM or RPKM instead of raw counts?
2. And what do you mean by "there is no statistical test?" If the FPKM value for one gene is always higher (in every tissue) than for another gene, I cannot say that the RNA level is higher?
With regard to #1, RPKM/FPKM is an attempt to account for differing gene lengths. With regard to #2, RNA-seq is not an "absolute" measure, only relative. Therefore, even if the FPKM is higher in all cases for Gene A than for Gene B, one cannot conclude much about the relative expression of Gene A vs. Gene B.
Sean, FPKM is indeed a relative measure, but you can compare FPKM between genes within a properly normalized experiment. That is the point of normalizing for gene length. However, as Gordon says, there's no statistical test that you can use to assess the significance of any difference between two genes, because any such test would need to account for the correlation between genes, which is unknown and cannot be calculated from just a few samples.