Found writing CommunityMech:000336 in PR #747.
InteractionTypeEnum classifies interactions by their sign — who benefits:
MUTUALISM Both organisms benefit (+/+)
COMMENSALISM One benefits, other unaffected (+/0)
PREDATION One benefits, other harmed (+/-)
COMPETITION Both negatively affected (-/-)
CROSS_FEEDING Metabolite exchange between organisms
SYNTROPHY, NICHE_PARTITIONING, STRAIN_COMPETITION, COLONIZATION_FACILITATION
The C. acetobutylicum-C. ljungdahlii pairing has three well-documented
mechanisms that no value covers, because they are mechanisms of transfer, not
signs:
- Interspecies heterologous cell fusion — the two species fuse cell wall and
membrane (PMID:32873766, by TEM and electron tomography). GO:0140253
cell-cell fusion exists but is a biological process, not an inter-organism
relation, so it can sit in biological_processes and not in
interaction_type.
- Bulk cytoplasmic material exchange — protein, RNA and ribosomes crossing
between species (PMID:32873766, PMID:39254339). Distinct from metabolite
cross-feeding: the currency is macromolecules, and it is bidirectional.
- Horizontal gene transfer through that fusion — plasmid DNA from
C. ljungdahlii acquired and integrated into the C. acetobutylicum genome
(PMID:38214507). Distinct from conjugation, transduction and transformation,
which is the whole point of that paper.
Why this is not just a naming gripe
Forcing them into the sign enum makes the record assert something no source
establishes. Curating DNA transfer as COMMENSALISM claims the recipient
benefits; none of the cited work shows that. Curating fusion as MUTUALISM
claims both benefit from the fusion itself, as opposed to from the syntrophy
and cross-feeding that the same pairing also has.
PR #747 therefore leaves interaction_type unset on those two interactions and
explains why in each description. That is schema-legal — the slot is optional
— but it is unusual: exactly 1 of 899 interactions in the corpus currently
omits it, so two more from one record is a conspicuous workaround rather than a
convention.
Suggested handling
This is what ground-or-propose-metpo is for. The candidates:
- an inter-organism cell fusion relation;
- an inter-organism cytoplasmic material exchange relation, with the exchanged
class (protein / RNA / ribosome / DNA) as a qualifier;
- horizontal gene transfer as an interaction type, with a mechanism qualifier
so fusion-mediated transfer is distinguishable from conjugation.
Ground to RO first if a suitable relation already exists there; propose to METPO
only for what genuinely has no home. If the enum does gain values, PR #747's two
untyped interactions are the first things to backfill.
A wider question this raises: interaction_type currently mixes sign
(MUTUALISM, COMPETITION) with mechanism (CROSS_FEEDING, SYNTROPHY,
NICHE_PARTITIONING). An interaction can be both a mutualism and a cross-feeding
and only one slot holds it. Whether to split sign from mechanism is a bigger
decision than this issue needs, but it is the reason the enum keeps running out
of room.
Found writing CommunityMech:000336 in PR #747.
InteractionTypeEnumclassifies interactions by their sign — who benefits:The C. acetobutylicum-C. ljungdahlii pairing has three well-documented
mechanisms that no value covers, because they are mechanisms of transfer, not
signs:
membrane (PMID:32873766, by TEM and electron tomography).
GO:0140253cell-cell fusion exists but is a biological process, not an inter-organism
relation, so it can sit in
biological_processesand not ininteraction_type.between species (PMID:32873766, PMID:39254339). Distinct from metabolite
cross-feeding: the currency is macromolecules, and it is bidirectional.
C. ljungdahlii acquired and integrated into the C. acetobutylicum genome
(PMID:38214507). Distinct from conjugation, transduction and transformation,
which is the whole point of that paper.
Why this is not just a naming gripe
Forcing them into the sign enum makes the record assert something no source
establishes. Curating DNA transfer as
COMMENSALISMclaims the recipientbenefits; none of the cited work shows that. Curating fusion as
MUTUALISMclaims both benefit from the fusion itself, as opposed to from the syntrophy
and cross-feeding that the same pairing also has.
PR #747 therefore leaves
interaction_typeunset on those two interactions andexplains why in each
description. That is schema-legal — the slot is optional— but it is unusual: exactly 1 of 899 interactions in the corpus currently
omits it, so two more from one record is a conspicuous workaround rather than a
convention.
Suggested handling
This is what
ground-or-propose-metpois for. The candidates:class (protein / RNA / ribosome / DNA) as a qualifier;
so fusion-mediated transfer is distinguishable from conjugation.
Ground to RO first if a suitable relation already exists there; propose to METPO
only for what genuinely has no home. If the enum does gain values, PR #747's two
untyped interactions are the first things to backfill.
A wider question this raises:
interaction_typecurrently mixes sign(MUTUALISM, COMPETITION) with mechanism (CROSS_FEEDING, SYNTROPHY,
NICHE_PARTITIONING). An interaction can be both a mutualism and a cross-feeding
and only one slot holds it. Whether to split sign from mechanism is a bigger
decision than this issue needs, but it is the reason the enum keeps running out
of room.