I ran edgeR and DESeq2 on two datasets each with 5 conditions. For each condition, I have at least three replicates. However, the results don't seem to match up between DESeq2 and edgeR. Only about 20-30% of the genes called by DESeq as DE are also in the set called by edgeR. DESeq2 also seems to call more genes than edgeR but there is no clear trend across conditions.
I tried running DESeq2 with the same initial filtering as in edgeR and with independentFiltering=FALSE and the agreement percentages between the resulting sets of genes are not changed.
Do you know what are the possible causes of this difference in results?
I tried different sets of inputs counts (transcripts from different sources, genes etc).