Combined MTX/MGX analysis, contamination removal

Hello,

I am trying to perform combined MTX/MGX analysis. When running the pipeline (Kneaddata/Humann), would it be more reasonable in case of the metatranscriptome to align it to the human genome or the human transcriptome, or even create a custom kneaddata database combining the two datasets. I would really appreciate some input. Thank you!

Best,
Mate