Hello,
Do you have any insights regarding the parameters to use when running Maaslin2 for Metabolomic data ?
Thank you again for your help!
Best,
Nicolas
Hi @NicolasB - we refrain from recommending default parameter values in general due to data-, design-, and platform-specific differences but you can take a look at the iHMP paper (Multi-omics of the gut microbial ecosystem in inflammatory bowel diseases | Nature, Differential microbiome feature abundance) for an inspiration where we used a set of reasonable parameter values for various data types including metabolomics. I hope that helps as a starting point?
All the best,
Himel
We’ve been trying out MaAsLin2 for metabolomics data as well. I’m not an expert on this topic by any means, but what seems to work for us is to use something approximating Z-scores (zero-centered, unit deviation) as input, with normalization="NONE", transform="NONE", analysis_method="LM" as parameters.
Per the iHMP paper, a log transform may also be appropriate but not on data that are already normally distributed. From experience with our metabolomics data provider, a log transform applied to what they call “scaled imputed values”, appears to provide approximately normally distributed data.
1 Like