Hi all,
I’m running DESeq2 on TCGA-LUAD RNA-seq counts comparing Primary Tumor (TP) vs Normal (NT).
I have 529 tumor samples (1 per patient) and 59 normals.
With padj < 0.05 and log2FC mire oe eq to 1, I get around 13k significant DEGs, which seems way too high. previously, a similar setup gave 3k.
I’ve checked:
All tumors are primary tumors No duplicate patients Factor for DESeq2 is set correctly: factor(group, levels=c("Normal","Tumor"))
I suspect my prefiltering might be too permissive, but I’m unsure how to go from here
1 answer
Large sample size, strong expected effects. Given that your QC made sure you do not have confounder eriving the DE even more beyond the biology then you can test against a fold change, dee vignette, to prioritize strong and effects and get fewer DEGs. This dataset is old and has been analyzed a million times before. Be sure to smartly use results to not reinvent the wheel.
Log in to answer this question.
