Hi there,

I am using RobiNA for my RNA-Seq analyses. I want to compare different stress conditions in plants with WT plants.
I am working already with the counts table for the expression of different plant genes.
Unfortunately, I have no replicates for my samples.
My question is whether the values in the raw_countstable.txt files are some kind of normalized from the log2-values (logFC) from the full_table.txt files? Or whether there is no normalization step without replicates? And if not what is the difference between these two values.
At the end I would like to compare the expression of the genes under the different conditions. But without normalization step this is not possible, or?

It would be great if someone might help me.
Thank you very much!

Best,
Schmiddie

Hi

I am not quite sure I follow. But in any case even if you work without replication, normalization factors are calculated for the the analysis of differentially expressed genes.

If you look in the source directory you should find an R source file ROBINAXXXXX.R where XXX depends on the name you gave.
Within there (if you used mostly standard settings) you will find

"d <- DGEList(counts=raw, group=groups)
d <- calcNormFactors(d)"

which tells R to calculate normalization factors to be used in the analysis.

That being said, without replication the p-values are almost completeley meaningless and not interpretable. Usually the least you should do is to pick your genes of interest and cofirm them in a well replicated qRT-PCR study. (i.e. biological replicates)

Best Wishes
Björn

Hi Björn,

oh I see. So that means that the values in the "xxx_results" textfiles in the column "d_xxx" are the same as the logFC values which are present in the "full_table_xxx" textfiles, and are already normalized?
Do you have an idea which algorithm is used for the normalization step?

Best regards,
Karin