I tried to find non-DGE genes from both microarray and RNA-seq. The definition I used is to find gene which has logFC small and close to 0. I found that if the gene has small logFC, the p-value tend to be big (>0.05). I found that after I filter the logFC and tried to visualize the distribution of p-value. My question is, if I keep this genes and said this is the genes that is not DE based on logFC, is it valid? Is it ok if I ignoring the fact that the p-value is big enough so that it might be not significant in term of statistic? If it is not okay, is there any tools to get the non-DE genes with strong p-value? Is there any suggestion to do this test if no tools available. Thank you very much.

The way I do this is to run topTable() with confint=TRUE. The top table results will then include a confidence interval for the logFC for each gene.

To find genes that are non-DGE you first have to specify what you mean by a small fold change. Suppose you want genes that you are confident have fold changes less than 50%. In that case, choose those genes for which the max abs value of the confidence interval is less than log2(1.5).

To me, it is not ok to use logFC alone without considering variability.