What does “non-log-linear space” mean for genomic data?

Question

[willy]

I’m reading a paper (CIBERSORT, by Aaron Newman et al, 2015) and the authors mention that all data are “in non-log-linear space.” Authors also mention adding “non-log-linear noise”.

What does this mean?

Sorry if this is a really noob question. I’m new to this field.

Answer 1

[seidel]

It's nothing too complicated. It simply means that the data is not log transformed. Sort of like advertising cholesterol free potato chips (plants don't produce cholesterol, so it's not going to be there unless you put it there). Often, in gene expression studies or publications involving analysis of gene expression signals for algorithm development, the gene expression data is log transformed. This is because the data approximates normality upon log transformation, and the variance is more readily dealt with (for instance see Irizarry et al., Exploration, normalization, and summaries of high density oligonucleotide array probe level data (2003), as well as a gazillion other references of that era). So I think Max and Ash are referring to and addressing assumptions that the data might be log transformed vs. linear.