I recently performed RNA-seq differential expression analysis. I first did a comparison between Condition 2 vs. Control. Then I did Condition 3 vs. Control. However, the raw counts do not seem to be substantiating the output data from edgeR. For instance, the Krt17 gene supposedly has a log fold change of 7 in the comparison between Condition 2 vs. Control, and the Krt17 gene does not show up on the list of differentially expressed genes (at an FDR of 0.05 or lower) for Condition 3 vs. Control. Yet, when I go back to the raw counts for the Krt17 gene, this is what I have: Control replicates(0, 0, 0), Condition 2 (25, 27, 25), and Condition 3 (16, 14, 11). Clearly, it seems to be "overexpressed" in both, but it doesn't show up in the other list. How do I fix this or account for such an observation?
For those wondering, I ran my analysis following most of what the Griffith Lab tutorial suggests.