Hi, I have 36 animals and got two tissues of each animal for RNAseq.

My design model is: sample+tissue

and I got DEG results like that:

out of 16445 with nonzero total read count adjusted p-value < 0.05 LFC > 0 (up) : 7276, 44% LFC < 0 (down) : 7508, 46% outliers [1] : 0, 0% low counts [2] : 0, 0% (mean count < 4)

90% of genes are DEGs at the level of 0.05. Is there any problem with my results?

Below I attached two figures:

1.Normalization figure. I plot the density function of the gene vst expressions in each sample. The figures showed that two tissues are different, but all normalized to the median.

1. Gene expression data. I rank the average gene expression of genes in Tissue1, and the then plot all genes in Tissue2 based on the ranking order in Tissue1.

From this figure, it seems that a lot of genes between these two tissues are quite close to each other, there should not be so many DEGs.

Any suggestions?

Thanks for your time & regards,

Raymond

You are testing a point null and have many samples. We discuss this situation in the DESeq2 paper. Why not specify an LFC threshold?