Is there a "general" accepeted dysbiosis index or score?

Dear all,

I am new to the science of microbiomics and just started to learn about qiime2 with the help of the available tutorials (which are really great).

Now, I was wondering if a "general" accepted dysbiosis index or score for microbiomes exist. If so, is such an index/scoring system already implemented in qiime2 (maybe as a plugin)?

In the literature I found several attempts to define dysbiosis; often from an ecological view point that in a disbiotic microbiome a niche is not occupied and so pathogens can subsequently fill this niche (or due to changes in the host enviroment, e.g. pO2-changes, pathogens can occpuy a niche). However, to develop a score for dysbiosis based on this definition, the core metabolic functions of the (human) microbiome (as well as the potential probiotic or commensal bacteria which constitute this core) must be known. Bacteria which do not belong to this core, could then be scored as dysbiotic.

In your opinion, is the development of such a scoring system (or a similar one which relies on another definition of dysbiosis) possible and could be robust/meaningful?

An answer would really help me :slight_smile:


Not that I know of. However, I'm more on the numerical ecology side of things, and the clinical microbiologists may have better ideas!

Maybe! :face_with_monocle:

In this article from 5 years ago, @.Yoshiki had a clever idea for comparing healthy gut microbiome variance vs unhealthy / inflamed gut microbiome variance using an ordination.

The healthy plane is then located in UniFrac space by fitting a line to the major axis of the points (d ) and defining a least-squares fit to identify a plane that minimizes the sum of squares of distances to the nearest point on the plane
Dynamics of the human gut microbiome in inflammatory bowel disease | Nature Microbiology

This 'healthy plane' method was built to put a number on a trend we saw, so it's not necessarily robust or directly translatable across studies. I remember us discussing other methods for scoring dysbiosis, including archetype analysis.

This is an interesting topic, and I would love to know if new metrics are being developed. :hammer_and_wrench:

Thank you for bringing this question to the forums! :qiime2:


Thank you very much for your response. I have read the article...the idea of an "healthy plane" is interesting. Unfortunately, I could not find much more publications regarding a "general" dysbiosis index...However, i will keep on searching and will post any news here :slight_smile:

1 Like

Hi @Bonic and @colinbrislawn,

Its interesting that this topic is coming up now, I was thinking about how "dysbiosis" has historically been defined as not!case, rather than by anything consistent. These are mostly thoughts, rather than any actual advice, so take them for what they are...

There's a "health index" developed by Gupta et al, (see q2-health-index for an implementation), however, it requires your data to be processed through Metaphlan2 for the database and assumptions to work.

I think a broader question is what "dysbiosis" means, because there are multiple community ways to define it. Do you think it has to do with stability, the way @colinbrislawn's study defined it? A shift away from a common "core" profile or change in taxonomic abundance? The change in a keystone? A loss or change in a diversity like the shannon index? A loss of core functional potential? A loss of core functions?

Do you expect it to be consistent across all populations, or is this dependent on selection/situational factors like age, diet, location, medication use, or other things?

Essentially, I think you need to look into eubiosis as closely as you try to define dysbiosis.



@jwdebelius your comments about how to really define dysbiosis reminds me of the plugin q2-karenina... I have not ever tried it, maybe this would be relevant to your study, @Bonic ?


I agree that there's not a widely used, generally useful dysbiosis index (at least not that I'm aware of). I came across this review a couple of weeks ago, which I thought was a helpful introduction to different approaches. And interestingly, I just came across this R package last night, which seems to have been inspired by that review.