Hello,

So, I am working with RNA-seq data and voom/limma workflow. After I finished my workflow and get the list of DE genes, I tried to filter by p-value. I choose <=0.1 for my cutoff value. After that, I chekced one of gene and it isn't in my list. I noticed that it maybe because the p-value and it is. The p-value is big, almost 0.2. I tried to find the reason why p-value is big. I check the read count. The raw read count is actually quite big. Below is the readcount:

Below is values after I used voom:

I checked the Cat_1 vs Cat_3. What is the reason the p-value is big? With that read count, I hope I can get the p-value to be significant and that gene is one of the important gene to check.

Well, for starters, the variance is estimated using all samples. For this gene, there are several aberrant samples; Cat_2_1 in particular, in which the count is 10-fold lower than its replicates, but also Cat_3_2 (also 10-fold lower) and Cat_3_4 to a lesser extent. This results in a large variance for this gene, which leads to a larger p-value. Moreover, the counts don't suggest that there's a strong difference between Cat_1 and Cat_3. Both groups have counts from 1000 - 3000, so that's not very strong evidence for DE.