Why EBSEQ and DEseq2 gives different number of DE genes

Question

[John]

Hi,

I have RNA seq expected counts from RSEM (n=3, two conditions). I get only 50 differentially expressed genes from DESEQ2 (FRD<0.05),while I'm getting 160 differentially expressed genes from EBSEq (PPDE> 0.95, that means FDR<0.05). Even though they are different statistics, why they are showing very dramatic differences?

I doubted my samples, so I did even run many other samples, I got more numbers of genes from EBSeq (180 DE genes) than DESeq (69 DE genes).

Is there any simple explanation?

Thanks in advance!

Answer 1

[h.mon]

No, there is no simple explanation. They are different methods, results are expected to be different, and indeed many papers report differences between several methods. To complicate even further, different methods can have better or worst performances depending on the data set being analysed.

All the papers bellow include a good number of DGE methods in their comparison. They are an interesting read, and will probably help enlighten a bit the issue, but they will show my point above: different results and performances from different methods and data sets.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4878611/

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4293378/

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3608160/

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4201821/

[John]

Thank you...........!

Answer 2

[swbarnes2]

50 versus 160? That's pretty close. 50 versus 5000 would be more worrisome.

My guess is, those 90 genes that DESeq didn't call are pretty close to the cut-off, and that overall, you have pretty good agreement between the reported fold changes.