I used MaAslin2 analysis to process the relative abundance of my microbial species. Age and gender are used as random_effects. I want to know whether the relative abundance result file after adjusting for age and gender can be found in the output file of MaAslin2 ? Or tell me ’ boxplots for categorical variables ’ is a graph generated from which file data
Hi,
Maybe I’m misunderstanding what you’re looking for, but the significant_results.tsv and all_results.tsv contain the associations with relative abundance after adjusting for covariates. However, there’s no file that shows the relative abundances themselves after adjusting for covariates. The data used for the boxplots is still just the original relative abundances, so what you’re seeing in those plots are the unconditional associations. That is, the coefficient and p-value reported are from the results table (adjusted), but the actual points in the plot aren’t adjusted; they just show the marginal association.
Will
Thank you very much for your patience. There is another question to ask you, why my results are not generated a heatmap, because there is only one fixed_effect in my metadata ? My fixed_effect is only divided into health group and disease group.
Assuming you actually have significant associations in your significant results table and you have plot_heatmap=TRUE, it should still generate a heatmap even with a single fixed effect. Still, if you’re not seeing the heatmap, it should be relatively straightforward to create the heatmap from the significant results table. You just have features on one axis, metadata variables on the other, and the cells are red/blue for positive/negative effects with the alpha determined by the log q-value.
Will
Thank you very much for your patience, I have learned a lot!
Could you answer my one last question why there is no file that shows the relative abundances themselves after adjusting for covariates? Is it because MaAslin2 analysis does not directly adjust relative abundance? I hope to get your answer, and I will be really grateful.
MaAsLin doesn’t directly adjust relative abundances (it’s just linear modeling on the relative abundances), so there won’t be a table of just the relative abundances themselves after adjusted for covariates. You could maybe create something like this by running a model with only your covariates and then storing the residuals, but those wouldn’t be relative abundances in any traditional sense.
Thank you very much for your reply !