can we perform Enrichment analysis for Insignificant genes (padj <0.05)

Question

[majeedmj.ict]

Dear Experts, I have very naive question. I have performed Deseq2 analysis on condition1 vs condition2 and none of the genes pass the threshold padj < 0.05 ( zero significant genes) . but when i see pvalue < 0.05 (20 genes significant) .

My question is can i perform gene set enrichment analysis for 20 genes (when pvalue < 0.05 threshold) or because there are zero genes are signifcant (when padj < 0.05 threshold) , it is not an good idea to do enrichment analysis ?

I was just wondering if there were no significant genes then what is the use to perform further enrichment analysis or pathways analysis ? .

please guide me .

Answer 1

[Mohammad Rezaei]

I don't think that enrichment analysis on the non-significant genes is a good idea! This insignificant result can result from a bad quality sample in your analysis. You should perform multiple quality controls (e.g., PCA, sample correlation, and density plot) to find your bad quality sample(s) and remove them from your raw data. Then analyze your dataset again. Then, if there was no significant result, it is probable that your dataset is not appropriate for analysis, and you should find new samples with high quality.

[Mohammad Rezaei]

I have to add this point to the previous answer! My previous answer was about the subsequent enrichment analyses (e.g., pathway) on the DEGs. But mainly about gene set enrichment analysis, I think you performed the wrong methodology. If you want to perform a GSEA analysis, you should not perform DE analysis with DESeq2! Instead of this method, you should prepare normalized count data (with genes' ID in the row names and sample characteristics in the colnames) and then perform GSEA analysis on the count data. GSEA performs DE analysis and pathway enrichment analysis on the up and down-regulated genes.