I am having trouble understanding Cuffdiff q-values. I have a gene, for example, that appears to be differentially expressed but is not called so even when using either two or three replicates. In two replicates, the gene levels appear like this:
Condtion1 Rep 1 61.8278 58.4906 65.1649 OK
Condition2 Rep 1 0.416481 0.176026 0.656937 OK
Condition 1 Rep 2 52.4839 50.2009 54.7669 OK
Condition 2 Rep 2 3.07902 2.52377 3.63427 OK
but somehow, even though the p-value is significant, the q-value is almost one.
Can anyone explain to me how this q-value adjustment takes place, and what other information in my data might be causing the adjustment to be so high? This gene seems to me, along with other genes in the data, like it should have been called differentially expressed.
As a note: I ran cuffdiff on these two samples separately, not using replicates (just out of curiosity) and in the second pair, this gene is not called differentially expressed, though it is in the first. What's going on?
Condtion1 Rep 1 61.8278 58.4906 65.1649 OK
Condition2 Rep 1 0.416481 0.176026 0.656937 OK
Condition 1 Rep 2 52.4839 50.2009 54.7669 OK
Condition 2 Rep 2 3.07902 2.52377 3.63427 OK
but somehow, even though the p-value is significant, the q-value is almost one.
Can anyone explain to me how this q-value adjustment takes place, and what other information in my data might be causing the adjustment to be so high? This gene seems to me, along with other genes in the data, like it should have been called differentially expressed.
As a note: I ran cuffdiff on these two samples separately, not using replicates (just out of curiosity) and in the second pair, this gene is not called differentially expressed, though it is in the first. What's going on?
Comment