Hi all,
Please accept my apologize if you found the question is basic. I'm working on a non-model organism, after making de novo transcriptome assembly by Trinity and doing expression analysis with edgeR, I found that some of DE genes for a given protein were up-regulated and some of DE genes were down-regulated for the same protein. Could you please advise me how I should interpret such results, they should be considered as up- or down-regulated genes? I saw such results in several papers, but it is not clear to me how the author explained them and got the final conclusion for such DE genes.
As we all know, it is not possible to check all DE transcripts one-by-one, could you please advise me how I can intercept all DE transcripts to understand their possible role in my experiment? I also think of DE clustering, but I am not sure how clustering will be useful in this situation to get the meaningful biological concept from them. Please kindly give me some clues and suggestion on these issues.
Thank you very much in advance.
I think these results are from multiple alternative transcripts from one gene. Instead of looking DE at transcript can be checked at gene level for example cufflinks has this option
Thank you. That's right, some of these results are from different expression of multiple alternative transcripts of the same gene. However, I found that such results among the various genes, too.
Dear Seta, Hi, If you have used Trinity software for your de novo assembly, you can run it in the transcript or in the gene level for your DE analysis.
in the gene level investigation, the software will consider the sum of expression of all the isoforms of each gene and the results will be different from the transcriptome level.
about the "However, I found that such results among the various genes, too", that you have mentioned, I could not understand clearly what you mean. Do you mean that for example you have several genes (orthologos?) that all encode for example IGF1 and some of them are up-regulated and some are down-regulated?
Hi, Farbod. Thanks for your comment. You're right with doing expression analysis at the gene and isoform levels. However, I did the expression analysis at the isoform level. You mentioned that the results of gene and isoform level analysis are different. I prefer the isoform level expression analysis becuase of the annotation performs at the transcript level, so it is easier to assign homology and you don't have to worry about the validity of the transcript clusters. However, please let me know which one (isoform or gene) level analysis is more informative in your experinece?
Yes, right. I found several genes that annotated as the same protein, say IGF1, showed the opposit expression pattern, some up-regulated and some of them were down-regulated. How we should report and interpret such results to get a meaningful bilogical concept?
Does your organism have a closely related species with reference genome?
Unfortunately, no. It has not a closely related species.
Annoying. Sequence and assemble the genome? ;)
In all seriousness though, about how many cases are you talking? Seems like a rather hard problem to solve automatically, for limited numbers of genes you could consider a manual analysis by comparing both transcript from the same gene and see in which they differ (pairwise alignment). Do they differ in a functional domain (likely identifiable by blast) or are they essentially the same... I'm working on human data, so don't take my strategy as state of the art. Just a stab in the dark ;)
Thank you for your comment. No information about the genome is available. I have just done de novo assembled transcriptome. I don't know exactly how many, I have faced with this issue during careful analysis of some genes of interest. Regarding your suggestion to do pairwise alignment, if they aren't the same, it indicates that they are different genes and both up and down pattern is acceptable, yes? or what about if they are the same?
If they are exactly the same, why would mapping indicate a differential gene expression? Or did I misunderstand something in your analysis? You performed Trinity assembly, mapped individual reads to the assembly, counted the mapping reads and performed differential expression analysis?
Sorry, if my question wasn't clear. That's right, assembly, mapping, counted the mapping reads and DE analysis.