Question 1

Is dysbiosis a well-defined term?

Accepted Answer

Not particularly. It describes a departure from some reference state, but the reference varies between studies and healthy composition itself varies widely between individuals. Using the term is acceptable provided the work says what it means by it in that context, rather than relying on the word to carry an argument.

Question 2

Which confounders matter most?

Accepted Answer

Diet, medication and stool consistency as a proxy for transit time are the usual leading candidates, and all three associate with both disease and composition. Recording them prospectively rather than reconstructing them afterwards is what allows adjustment. Their omission is among the commonest weaknesses in this literature.

Question 3

Why does relative abundance complicate analysis?

Accepted Answer

Because sequencing measures proportions, not quantities, so an apparent rise in one taxon can result from a fall in another rather than any real change. Methods designed for compositional data address this, and stating which approach was used, with its assumptions, is expected in the analysis section.

Question 4

What is needed to move from association to cause?

Accepted Answer

Either temporal evidence, from sampling before disease onset or across a treatment, or an intervention, most often transfer of a community into germ-free animals. Both have limitations, and neither converts a cross-sectional difference into a cause on its own, so framing the claim to match the design matters.

Question 5

Do metabolite measurements add much?

Accepted Answer

They add function, which composition alone does not give. Two communities with different membership can produce similar metabolite profiles, and the metabolites sit closer to a mechanism than a taxonomic list does. Pairing the two is more informative than either separately, though it roughly doubles the handling requirements.

Question 6

How much does sample handling affect results?

Accepted Answer

Substantially. Time to freezing, storage temperature and extraction method all shift the observed community, and differences between collection sites can exceed differences between patient groups. Standardising handling across the study, and recording batch structure so it can be modelled, protects against attributing a technical effect to biology.

Question 7

Are animal models informative here?

Accepted Answer

Within limits. Germ-free transfer can show that a community is sufficient to produce a phenotype, which observation cannot. The recipient animal differs from a human in diet, physiology and immune background, so a transferred phenotype supports a causal role without establishing that the same mechanism operates in patients.

Microbiology

Microbiology

Research Frontiers in Microbiome Composition and Dysbiosis in Human Disease