Hi all experts,
I have used Trinity software for doing de novo transcriptome assembly. Then, I used bowtie and RSEM within Trinity package for read mapping and counting. Finally, I got the raw count and FPKM value. I used raw count for doing differential expression analysis by edgeR software, say, I compared A-B library. I found that among DEG genes reported by edgeR software, some few genes with positive log fold change, meaning that the higher expression in A library in relative to B, have the lower FPKM value in A as compared with its FPKM value in B library. For instance, please consider the below example
logFC logCPM LR PValue FDR FPKM (B) FPKM(A)
c50354_g2_i5 2.825768038 5.659882391 41.70680491 1.06039E-10 6.64373E-07 118.187 18.093
As you could see the FPKM value of c50354_g2_i5 transcript are in 18.093 and 118.187 in the library B and A, respectively. But during the comparison A-B library, the log fold change is positive. Could you please explain to me what happened for such genes?
Thank you in advance
The first thing that comes to my mind is that you should probably include the code you used to obtain these values. And the output of sessionInfo().
If you plot edgeR's log fold change esimates vs. your manual FPKM log ratios, are they for the most part concordant with a few outliers, or most of them not concordant, at all?