Different tendency of results between two pipelines?

[super0925]

Hi All
As you know there are two kinds of pipeline in RNA-Seq analysis.
RPKM (Cuffdiff) and Count-based (DESeq/edgeR).
I have 3 conditions.
C1C2(the DE between Condition 1 and Condition 2)
C1C3(the DE between Condition 1 and Condition 3)
C2C3(the DE between Condition 2 and Condition 3)

In Cuffdiff , I get ~1000 DE genes in C1C2, ~1200 DE genes in C1C3, ~only 500 DE genes in C2C3
However, in edgeR , I get ~500 DE genes in C2C3, but only get ~300 DE genes in C1C2 and C1C3.
Is it normal?
All I used are default settings and the threshold is Q<0.05 (not considering fold change)
Thank you!

[mbblack]

Do you have replicates for these conditions? If so, how many?

And yes, different normalizations and analyses will often give very different results for numbers of differential expressed genes, especially so if you have few replicates (or worst, none at all).

[super0925]

Do you have replicates for these conditions? If so, how many?

And yes, different normalizations and analyses will often give very different results for numbers of differential expressed genes, especially so if you have few replicates (or worst, none at all).

I have 3 replicates in each group and all I used are default paramter settings.

[Asaf]

Do you see a correlation between the p-values of all the genes between the two pipelines? Do the differentially expressed genes in the two pipelines overlap?