Dear all,

I am trying to run EBSeq on a RSEM count matrix containing 2 conditions with 3 replicates in each.

I ran EBSeq following the tuotrial and it worked well. However, I have 3 questions :

• How could I determine which normalization method is the best ?
• My replicates are named AC1, AC2, AC3 for condition 1 and AI1, AI2, AI4 for condition 2 and I would like to make AC as reference for comparison (AI over AC), but when I ran the str(GeneFC) command, EBSeq output is "Direction: chr "AC Over AI". I tried to change the order of columns in the data matrix and replace AI1, AI2, AI4 by AI (and the same for AC) but it did not change.
• I also used DESeq2 on thisdataset and I would like to compare results between DESeq2 and EBSeq (even if counting tables from RSEM are not suitable for DESeq2 because they have to be rounded first) . In DESeq2, I selected DE genes at a p-value adjusted (method Benjamini) of 0.01 and a log2FC <-1 or >1. I wonder what is the equivalent in EBSeq output ? I guess it is something as genes with a PPDE < 1-0.01 and setting FDR < 0.01 ? 

Could you please help me ?

Thank you in advance,

Jean