inferring interactions from 16s data


I’m dealing with some metabarcoding data and would like to know if I can use my 16s data to infer host-bacterial and bacteria-bacteria interactions. My data is not time-series. Any comments and suggestions is much appreciated.

Thanks in advance!

Hi @uth,

Umm… I have some luck on bacteria-bacteria interactions with q2-SCNIC, which builds a network based on SparCC. The plugin doesn’t work with the latest release (I think I was running it on 2018.11) but it made some nice networks.

As far as host-bacteria interactions go, Im not sure what you mean. Like, host social networks? Host phylogeny? Just general host features? Depending on what you want, there’s a lot of ways to go. Sort of choose your own adventure.



Hi @jwdebelius,

Thank you so much for your reply.

I have no clear idea about using 16s abundance data to infer interactions. My host is an alga and I want to look at the algal bacterial interactions using 16s data sets which are not time series data. I want to infer any possible ecological interactions between the alga and bacteria.

I’d be grateful if you could let me know, what you mean by “host social networks? Host phylogeny? Just general host features?”. How does the methods differ in the way you look at these mentioned entities? Please, do let me know if there are good literature that I should probably have a look to understand these concepts.

Many thanks!


Hi @uth,

With the caveat that I’m not an evolutionary (or co-evolutionary) biologist, there have been a handful fo really interesting papers recently around these topics, although most methods are not implimented in QIIME.

However, its hard to know where to point you if you’re not sure what you want to know about host-bacterial interactions. So, you could approach from a standard perspective similar to a case-control study where you use characteristics about hte host grouping. For something like that, you may want to check out the dataset tutorials (the first like 5-6 in docs page) and ignore the time series pieces.

If you have additional information about the hosts, you could layer in things like host phylogeny or host interaction, but again, that depends on yoru design. For those, a quick google scholar search should help come up with the papers, which are alluding me right now.


Thank you so much for your valuable comments. I found some good literature on interaction inferences. I hope things will work out!

Thanks again :grinning:. Really appreciate your suggestions!

1 Like