Functional gene normalization

Microbial community profiling HUMAnN
1

HUMAnN 3.9

Dear officer,

I have a question regarding functional gene normalization. Following the standard workflow in HUMAnN3, I have already normalized my functional gene data into relative abundance. However, I encountered an issue: the ungrouped and unmapped data were also included in the relative abundance of one sample. Typically, when most people perform functional gene analysis, they only use characterised functional genes (e.g., K00001 or K00002, rather than ungrouped or unmapped genes) for data analysis.

Given this, when I focus solely on the characterised functional genes, do I need to re-normalize the characterised functional genes to ensure they are on the same scale?

Best wishes,
Shuyuan

2

There is an option to exclude “special features” in the normalization script if you don’t want to include things like UNMAPPED and UNGROUPED. It is mostly a philosophical choice of whether you want to measure differences the amount of total read mass assigned to functions vs. the differences in the composition of the mass that was assigned. Both can be useful. I tend to favor the latter.