Microbiome Taxa Analysis for two factor interaction

General Discussion
1

Hi,

I am working on analyzing a microbiome dataset that has two factors - Sex and Treatment. Microbiome data is non-parametric, and transformation is not recommended (since important information can be lost), so as a default, I always use the Kruskal-Wallis test, Mann-Whitney test, or Spearman Correlation. In order to evaluate interaction, the only test I can perform is a two-way ANOVA, which is meant for parametric data, so this test cannot be used for the microbiome dataset. In this regard:

  1. Is Kruskal Wallis an acceptable test to compare 4 groups (Male_Control, Female_Control, Male_Treatment, and Female_Treatment)?
  2. Is there an alternative test that can be used instead of a 2 Way ANOVA that can test the interaction between sex and treatment?

Your suggestions would be very helpful for proceeding with the analysis.

Thank you,
Best,
R

2

Hello R,

Are you testing which microbes are higher and lower between groups?
This paper may be helpful: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8763921/

Or perhaps you are testing if groups are different (not overlapping in a PCoA plot)?

1 Like
3

Hi Colin,

Thank you for sharing this information. I will have a look at this paper. I am testing how the relative abundance of taxa is affected by treatment and sex.

Best,
R

1 Like