Scaling by feature after CLR transformation for microbome data?

Analysis
1

Dear MixOmics Team:

I am working on integrating the proteomics data and microbiome data with sPLS-DA (block.splda). I have performed the CLR transformation for the microbiome data, as shown in mixMC tutorial. I was wondering if we should keep scale = TRUE while running block.splda function. Would scaling data by feature after CLR over-normalize the data and damage the characteristics rendered by CLR transformation?

Thank you!

2

Hi @garyz,

I would recommend setting scale=FALSE as you have already scaled your microbiome data using the CLR transformation, however it might be a good idea to try running both with and without scaling and inspect the outputs such as sample plots to see what impact it has.

Cheers,
Eva