rCLR transformation


Thanks for this wonderful tool.

I had an inquiry regarding the transformation implemented in the package.

For the CLR, the imputation is calculated as half the min feature for each sample. Why did not you consider using the robust CLR? Did you find in your testing any issues with it?


hi there @MEladawi,

I don’t believe we have tested an implementation of robust CLR using Maaslin2 but @himel.mallick may be able to correct me here.

In essence I don’t see any issue as to why you couldn’t normalize your data first using rCLR and then pass it into Maaslin2 with the normalization parameters as NONE

Jacob Nearing

Thanks for the information.

Yes, I realize it is doable, just wanted to know if you know how it may affect the regression models compared to the CLR.


1 Like