Consider adding other models to MaAsLin2

In reading the referenced paper for MaAsLin2 (Multivariable association discovery in population-scale meta-omics studies), the authors indicate that this program can work with other distributions:

" While the default MaAsLin 2 implementation uses a log-transformed linear model on TSS-normalized quality-controlled data, the software supports several other statistical models including count models (e.g., Negative Binomial [[17]), zero-adjusted models (e.g., Compound Poisson [[18]–[20]], Zero-inflated Negative Binomial (ZINB) [[21]]), and multiple normalization/transformation schemes under one estimation umbrella."

For microbiome data, the zero-inflated model would likely be preferred, given that much of the data sets include substantial 0 counts.

Is it possible to adjust the Microbiome Analyst page for MaAsLin2 to let us select a different model?

Good suggestion! We will look into this feature and add the support if compatible …

Update: all the models are now available!

This topic was automatically closed after 2 days. New replies are no longer allowed.