Suggested min_prevalence and min_abundances for mataphlan or humann inputs in maaslin

Downstream analysis and statistics MaAsLin
1

Hello community,

I am trying to run maaslin on results from metaphlan and humann. In both cases I have relative abundances as suggested from their github guide.

I want to use maaslin on:

  • taxonomic profile
  • gene families
  • pathways
  • reactions

While running some trials of maaslin, I noticed that of course I can not use the same min_prevalence and min_abundance threshold for every file. In particular, for gene families file I have relative abundances in the order of e-05 and lower, as for metaphlan they increase to e-01.

What are your suggestion for min_prevalence and min_abundances in the different cases?

Thanks

2

If it’s computationally feasible, I would just run without any min_prevalence or min_abundance since MaAsLin 3’s prevalence model is designed to model 0s vs non-zeros. However, if that’s computationally infeasible, I would use min_prevalence=0.01 and min_abundance=0.0001

Will