Determine if a microbial community is predictive of another

Hi QIIME2 community,

I am analyzing data from a study and would appreciate any advice on what analysis to run to answer my main research question.

Background: This study was a repeated measures design on 12 subjects. Half of these subjects had a cannula (allows access to collect intestinal fluid) and the other half did not. We collected intestinal fluid, feces, and swabs from the cannulated subjects and feces and swabs from the non-cannulated subjects over 4 timepoints during a 24 hour period. We took measurements that include pH (intestinal fluid and feces), taxa abundance, alpha diversity (Shannon's index), and beta diversity (Bray-Curtis, UniFrac). This study did not include an intervention, we did that in subsequent studies, this was just to assess a normal baseline over a period of time, but the question will be the same in those studies as well.

We want to know how feces and swabs relate to the intestinal samples? In other words, can feces and swabs be predictive of microbial changes in the intestine? This information would be useful in order to survey non-cannulated subjects in the future using a non-invasive method.

I'm not sure what analysis to use to answer this question.

I've looked into indicator species analysis and linear discriminant analysis, but those seem to provide statistical differences in features between communities. I'm also looking into SCNIC, but I'm not familiar with co occurance networks yet.

Thanks in advance!

1 Like

Hi @Mahasti,

I do not have much to recommend other than these papers:

There are many publications out there that compare how well swabs, and other sampling strategies, compare to direct sampling of the targeted material. These are just a few that I could recomend.

3 Likes

@SoilRotifer Thank you taking the time to reply and sharing those papers with me. The first one looks to be especially helpful in determining my methodology. It looks like supervised learning might be what I'm after in this case.

Really appreciate the help!

1 Like