Hi, I'm analyzing RNAseq in order to identify differential expressed genes between 2 conditions in 15 individuals (total of samples = 30). The normalization method that I use is TMM, with Limma R package, using voom and duplicateCorrelations for the individuals. The problem is that in the analysis I obtain 50% of the genes differentially expressed (around 10.000 DE of 20.000 genes). The assumptions of the TMM normalization are violated. How to proceed with such a case?

This has been answered here more than once. An example being this.