Hi,
I must run DGE analysis on bulk RNA-seq samples from abdominal aorta. Unfortunately, during the initial QC steps I detected a large variability in two lipid processing genes pervasive across all experimental conditions. This suggests that some of the sequenced samples had a larger amount of adipose tissue still attached to them. Thus, some samples' expression profile might be dominated by adipose tissue instead of vascular.
How should I perform the DEG analysis correcting by this? Can I simply add to the metadata the DESeq2-normalized expression levels of these 2 genes and run DESeq2 using as my design ~ adipose_score * grouping_variable
? where adipose_score is a continuous variable and grouping_variable is a factor with condition+genotype.
I have read about using RUVseq here. However, this dataset contains 3 experiments KOing a certain gene.
- WT vs whole-body KO
- WT vs KO only in celltype 1
- WT vs KO only in celltype 2
The proportions of celltype1 & 2 do also vary from sample to sample (even within the same experiment), although we cannot detect a clear pattern or marker for that. I worry that, by running RUVseq to remove unwanted variation, the software is going to remove also the celltype-of-interest variation instead of only the adipose-specific one.
What would you recommend me to do here? And how should I interpret the results/model?
Thanks