Taxa associated with more than one clinical covariate

Downstream analysis and statistics MaAsLin
1

Hello all,

After running MaAsLin2 on 271 samples with an abundance file containing 100+ taxa and a metadata file containing a 6 clinical covariates, our output results show some taxa associated with a single covariate, and several taxa associated with more than one clinical covariate at q< 0.05. Are these few significant covariates associated with the single taxon adjusted for confounders?

For example, taxon X shows statistical significance for association with covariate A, which is straightforward. On the other hand, taxon Y shows statistical significance for association with clinical covariates A, B and C. Does this mean that taxa Y is associated with covariate A after adjusting for confounders? And that this is the same for covariates B and C?

The fixed effects are gender, age group, tobacco use, periodontal index, dentate, and medication count.
Here is an example of our output results for a single taxon showing statistical significance for association with a few clinical covariates:

image
image884×127 5.32 KB

|metadata|feature|value|coef|stderr|N|N.not.0|pval|qval|
|Perio_Index|X.g__Filifactors__alocis|Yes|0.383845327|0.110376314|271|0|0.000591472|0.00545183|
|Dentate|X.g__Filifactors__alocis|Yes|0.459811519|0.13849872|271|0|0.001027162|0.008484094|
|Tobacco_status|X.g__Filifactors__alocis|Yes|0.37819624|0.131819776|271|0|0.004450095|0.026954861|
|Age_Group|X.g__Filifactors__alocis|Yes|-0.260699101|0.114043761|271|0|0.023049404|0.103235357|

Thank you for your help!

2

Hi - MaAsLin2 always runs a fully adjusted model with the provided covariates. So any association detected should be interpreted as significant while adjusting for all other covariates.