Interpretation of additional column in MetaPhlan4

Microbial community profiling MetaPhlAn
1

Hi all,

I’m doing species-level functional analysis downstream of MetaPhlAn 4 (mpa_vJan25), where organisms are mapped to functional roles (e.g. mucin degraders) by matching on species name. I’d like to check my understanding of the additional_species column before I decide how to handle it, and hear how others approach it.

My understanding of what the column means

For a multi-species SGB, the clade_name gives one representative species, and additional_species lists other named species whose reference genomes are within <5% ANI of that representative. As I understand it, these additional species are not independently detected — there are no distinguishing markers for them; they’re listed because the SGB’s markers can’t tell them apart from the representative. So the abundance belongs to the SGB, and the true identity is any one of {clade_name species, additional species…}, unresolvable at the marker level. Is that correct?

Where it gets tricky for me

My downstream reference (functional roles) is keyed to species names. Sometimes the functionally-relevant species I care about is not the clade_name but appears in additional_species. Concretely (generic example): an SGB is labeled as species A in clade_name, but species B — which my reference knows to be, say, a mucin degrader, is listed in additional_species. If I match only on clade_name, that SGB’s abundance never reaches the “mucin degrader” role, even though the organism present might be B.

So my question is really about legitimate practice:

  1. Is it defensible to relabel/attribute an SGB’s abundance to a species found in additional_species (rather than the clade_name) for downstream functional grouping? Or should only clade_name ever be treated as the identified species?
  2. Given the abundance is the whole SGB’s and the identity is genuinely ambiguous among the <5% ANI members, is attributing it to any single one of them (mine or the representative) equally uncertain? i.e. is the clade_namerepresentative itself just a convention, not a stronger claim than the additional entries?

I want to avoid over-attributing specific-species abundance that the method explicitly can’t resolve, but I also don’t want to silently drop functionally-relevant signal just because the representative label happened to differ. Interested in both the “intended interpretation” from the developers and what people do in practice.

Thanks!