Relative (to control group) Log2FC values obtained by qPCR and RNA-Seq are as follows:
See here:
There is a difference in values as well a TREND. I have quite a few other such genes and this problem has effected them too.
I am concerned about the values not following similar TREND. However, the type (up/down) regulation remains conserved throughout for all my genes. Can anyone please help me with this? is this result acceptable? What can be the cause of it?
IMHO, if all the genes you validate by qPCR change in the same direction as in the RNA-seq, it's validating surprisingly well, especially if the logFCs are in similar orders of magnitude. There is even ongoing debate about whether qPCR validation is really needed for RNA-seq (see for example this reference, or this biostar post).
With regards to the "TREND" you mention, it is not quite clear if you are referring to always having lower fold-changes in the "Treated3" group, or maybe "Treated1, 2, 3" are individual samples. Nevertheless, you may find that logFC values are quite variable because they arise from the comparison of two values and may change with different qPCR assays, or endogenous controls, etc.
What I mean by TREND is the Order of Magnitude in which the gene was expressed while going from Treated1 to Treated3 (i.e. 1st dpi to 3 dpi). This TREND differs between RNA-Seq and qPCR in most of my genes. However, good news is that the sign of Log2FC for each gene is same between RNA-Seq and qPCR across all dpis.
Well it's hard to say with such little information. I guess "dpis" are days-post-infection or something like that, so this is a time-series? And did you statistically test/validate these "trends" in the RNA-seq or did you only do the cross-sectional comparisons? If not, it may be less surprising that they are less reproducible in the qPCR, etc. But it is hard to say.
Yes, time series and dpi=days-post-infection. This is my cross-sectional comparisons. I ranked the LOG2FCs for all three dpi (based on magnitude) and found the order to be different for qPCR and RNA-Seq.
IMHO, I don't think it is surprising that the "trends" do not match well. These "trends" you observe are qualitative observations of the data, because you did cross-sectional comparisons. If you want to detect "trends" with confidence, you could do longitudinal comparisons across the time-points. If there are some genes with very strong significant trends maybe these will be also detected in the qPCR. Nonetheless, even in that case do not expect qPCR to give you similar results. It's a very different technique. Overall I think it is good that your cross-sectional comparisons (not the "trends") are similar between qPCR and RNA-seq.
Well it's hard to say with such little information. I guess "dpis" are days-post-infection or something like that, so this is a time-series? And did you statistically test/validate these "trends" in the RNA-seq or did you only do the cross-sectional comparisons? If not, it may be less surprising that they are less reproducible in the qPCR, etc. But it is hard to say.
Yes, time series and dpi=days-post-infection. This is my cross-sectional comparisons. I ranked the LOG2FCs for all three dpi (based on magnitude) and found the order to be different for qPCR and RNA-Seq.
I used DESeq2 for DE analysis.
IMHO, I don't think it is surprising that the "trends" do not match well. These "trends" you observe are qualitative observations of the data, because you did cross-sectional comparisons. If you want to detect "trends" with confidence, you could do longitudinal comparisons across the time-points. If there are some genes with very strong significant trends maybe these will be also detected in the qPCR. Nonetheless, even in that case do not expect qPCR to give you similar results. It's a very different technique. Overall I think it is good that your cross-sectional comparisons (not the "trends") are similar between qPCR and RNA-seq.
Eureka! Thanks Papyrus .