I know this question might not be related to Bioconductor but I couldn't think of anywhere else. When I take correlation between two genes, should I exclude genes that their expression is not meaningful? Additionally, should I exclude samples that their z-score is under e.g. z-score = 2? Or should I include all samples in a correlation?

Best regards,

Arman

When is expression not meaningful?

We typically filter genes based on an abundance threshold before doing analyses (e.g. a CPM cut off), is this what you mean?

Should you exclude samples based on z-score?

I wouldn't remove samples unless they are clear outliers based on a global gene expression profile (e.g. using a PCA).