Microbiome influences on infection

Downstream analysis and statistics MaAsLin
1

Hi, I am trying to test if my samples microbiome is predictive of my infection severity. Another way of saying this is I am wanting to know which bacteria are associated with an increasing infection integral (continuous value). I am worried the math would not make sense if I put infection integral as a fixed effect-as the microbiome should be influences the infection not the other way around.

I have the “pre-infection microbiome” of my 16s samples (n=40 samples) those samples were than infected with a pathogen and I have resulting “infect integral”. I am wondering what variables I should put into Maaslin.

In MaAsLin2, can I place “infection integral” as fixed effects, or would that imply it was causing the community difference?

2

Hi @Rachael_Kramp - not sure if this has been resolved yet but we generally do not recommend MaAsLin 2 if the end goal is to predict an outcome. Based on your description, it looks like a prediction algorithm such as random forest or elastic net might be more appropriate for your analysis?

3

Hello
I am reopening this topic because I think I have the same kind of question
I have a set of > 100 microbiome samples from different segment of the gut in 20 persons with HIV. My goal is to determine if some taxa are associated with higher level of virus. my initial approach was to consider the virus level as the outcome?
I can run maaslin like below

Maaslin2(
    input_data = df_input_data, 
    input_metadata = df_input_metadata, 
    output = "maaslin_output", 
    fixed_effects = c("virus_level","tissue","age","sex"),
    random_effects = c("pid"),
       min_prevalence = 0.1,
    min_abundance = 0.01)

I get results but should not it be the other way around with each taxa being tested and virus level being the outcome?
btw:impressive tool!
thank you

4

Hi @Antoine,

MaAsLin currently only supports taxa to be the response variable. If you want to try and use taxa to predict virus levels I would suggest checking out random forest variables.

Cheers,
Jacob

5

thank you Jacob. any package/tutorial you’d suggest?
really appreciated!

6

If you want to stay in R I would check out this package:

and this one:

cheers,
Jacob

7

thank you! Will definitely look into it.