Hello !

I have RNA-seq data from different cell strains/tissues. The datasets differ in the following way:

dataset a : two columns: one treated and one untreated.
dataset b : 3 treated 2 untreated.
dataset c : 4 treated 4 untreated.

I merged all datasets together so it has 15 columns. Currently i did
differential expression analysis. However i am just starting with this
and i am not sure if i did it correctly. The concern i have is that
i might be missing out on something.

Is it okay to merge datasets from different experiments ( they do have the same genes and same treatments ) together.
and do RNA-seq analysis. Or am i supposed to do different steps that the

default manual of DESeq provides ?

​Also maybe merging is okay but can i also merge dataset b in this scenario because it has 3 treated and 2 untreated, does this cause problems ?

​
Any critic is appreciated !!!!!

​
I look forward to an reply !!

Regards,

ben

​

Take a look at the PCA plot of the merged dataset (see vignette). If the treatment effect is common across the three, it may make sense to merge them and use ~experiment + treatment, to find common treatment effects.

Note that, dataset A alone doesn't provide any statistical power to detect differences, because there is no replication, but it is still useful if you add it to B and C in determining an average treatment effect.

There is no problem with the sample size imbalance of 3 vs 2.