Hi to the experts of RNA-Seq analysis,
I have this burning question of knowing your experience about following;
There is my favorite gene from Arabidopsis which I over-expressed in a Arabidopsis system in different controls and treatment conditions. I used LIMMA analysis pipeline to derive the differentially expressed gene lists (DEGs).
My question; is it possible that I not see the over-expressed gene in any of the contrasts I make (e.g. even in the contrast between transgenic control vs. wild-type control) ?
Thanks a lot for your comments in advance.
Shani.
I don't know if I understand the question correctly, but the gene may not be differentially expressed if you have very few samples, due to low power. But then you could select genes based only on fold-changes, not p-values. I don't know how you over-express the gene, but if its not picked up by RNA-seq, then perhaps the method is not good enough?
Also, it depends on your cut-of. Many people uses FDR
First of all, which promoter are you using to over-express your gene??
If your using a strong promoter but is not constitutive, and this promoter is active only in specific conditions I suggest to carry out your analysis using Fold-change for these conditions and control.
Regards
Once a transgenic overexpressing line has been made, it is necessary to find out which T1 plants express the most (using RT-PCR for that gene), then one can use the most expressed line for further dowstream experiment (like RNA-seq). I would suggest to do RT-PCR first. Not all T1 I(primary transformants ) over express the transgene, mostly due to the location of insertion and differential chromatic structure/silencing there.
best
Thank you all for your comments. Yes, we checked the lines for the transformants using RT-PCR and selected only the ones that showed high expression levels of the transgene was sent for RNA sequencing. We use the 35S promoter (which I believe is a constitutive). We used TruSeq paired-end library preparation and sequenced to a 15 million depth. I do agree with the not well annotated genome for Arabidopsis. I did use the FDR of 0.05 but I can make it less stringent and check. Or better, I can assemble my own transcriptome. I will discuss these results with my supervisors and your ideas and will decide on what to do. Thank you very much!
Regards,
Shani.
Hi,
can you distinguish reads from the the endogenous gene and the transgene, e.g. the transgene has another terminator sequence and possibly 3'-UTR?
Tony Schaeffner
if you are interested in just one gene then just extract the expression for that gene from your normalized matrix, say X is the gene vector with n components, where each component represents a particular sample.
Z-scale this vector and plot it with the samples on the X axis and the scaled expression data on the y axis. If your expression level shows a profile that it crosses or is close to the 2 standard deviation mark in some samples, then it 'MAY' be overexpressed in those samples.
If you are comfortable with doing some basic statistics, then you can use a t-scale with degrees of freedom = n-1 to calculate your 95% cutoffs on your y-axis.
Furthermore, a gene will not be working alone, you can select a set of genes that perform similar functions to your gene X. Some selection criteria are correlations in expression or belonging to the same pathway. Do those genes also follow the same profile of your gene X, plot them together after Z-scaling them.
Then you may want to design a RT-PCR experiment to confirm this profile.
- Badges
- Science topic
- Similar topics
- Biological Science
- Anatomy
- Tissue