cell-death/ lesion associated terms

[ValWood]

Hi Alayne could you check that all of the "pathogen-associated host lesions" have some cell death synonyms (also need to look at parentage, I will do this tomorrow)

[ValWood]

When I search on "cell death" which many will search on this is what I see

I'm not sure why terms which do not have"cell death" are higher than some that do. @jseager7 @kimrutherford could you look into the searching behaviour?

[kimrutherford]

The one at the bottom ""host necrotic cell death" is a related synonym so I think it makes sense for it to be not be near the top. If it was an exact synonym it would probably be higher up.

[ValWood]

To me, it seems odd that repeating a word twice (cell) scores higher than the exact phrase in any synonym (cell death). i.e. I expected an exact multiword string to always score higher than a repeated word. Repeated words are clearly important, but slightly less so than exact matches (I would imagine?) Maybe we have examples where this is not true? Can anybody recall any?

Although so far, I haven't encountered this issue using PomBase- so it could probably be resolvable by adding appropriate precise cell-death synonyms for the terms I was trying to locate

For example presence of pathogen-associated host lesions should have presence of pathogen-associated host cell death. There is a PHIPO ticket about this https://github.com/PHI-base/phipo/issues/306

But I am not sure they can be "exact synonyms" even though we describing exactly the same observation. Need @mah11 input here.....

[kimrutherford]

To me, it seems odd that repeating a word twice (cell) scores higher than the exact phrase in any synonym (cell death).

Currently related synonyms have a lower weight in the search than exact synonyms and names. We can tweak the weighting but we'll need to take care as related synonyms are quite general and can match a lot of terms.

If "cell" and "death" were part of an exact synonym, that term would be near the top.

[mah11]

What, exactly, do you need from me?

presence of pathogen-associated host lesions should have presence of pathogen-associated host cell death

In this example, "presence of pathogen-associated host cell death" sounds weird, just because I tend to use "present" or "absent" for physical things (continuants, in onto-jargon) more than for processes (occurrents).

In addition to the ontological sense in which these two strings aren't exactly synonymous, they don't look biologically identical. I'm assuming lesions in this context are always due to cell death, but I would naively assume that some pathogen could also cause cell death in some host in places or on a scale that doesn't form a lesion.

[ValWood]

That helped. was trying to figure the relationship between "cell death" and "lesion". The label is lesion(s) because they apply to multiple cells.

@CuzickA you could maybe add exact synonyms

pathogen-associated cell-death resulting in host lesions, present to get cell death into the search term?

~I'm not sure why PHIPO has "presence of blah" rather than "blah present" @CuzickA do you remember?~

Also, should the terms be lesion (singular?)

[ValWood]

~I'm not sure why PHIPO has "presence of blah" rather than "blah present" @CuzickA do you remember?~

ignore that part, I misinterpreted.

[jseager7]

Also, should the terms be lesion (singular?)

Personally I think singular, since that's applicable in the case of one or more lesions, whereas 'lesions' implies the formation of more than one lesion. Using the singular should also keep the term label more semantically compatible with the logical definition (some lesion).

My suggestion could be misinterpreted in the opposite way though (implying exactly one lesion), so if you want to be very precise, you could use 'lesion(s)'.

[jseager7]

you could maybe add exact synonyms pathogen-associated cell-death resulting in host lesions, present to get cell death into the search term?

If this term has the semantics of the pathogen being associated with cell death that causes lesion formation in the host, wouldn't a more effective term name be something like '[formation of] pathogen-induced necrotic lesions in host'? Then the synonym could be 'pathogen-induced host cell death causing necrotic lesions in host'.

Of course, that assumes the cell death is necrotic, and that the pathogen is actually inducing the necrosis and not just loosely causally 'associated' with the cell death somehow.

[ValWood]

The term here is "associated" because this is a grouping term. The plant also induces lesions as part of it's immune response. The plant will locally kill its own cells to prevent disease spread (this is probably the most common reason for lesion formation/localized cell death). In this situation (without further information) all we know is that we can see lesions. These might be disease (by product of infection by biotroph), plant induced (immune response), or pathogen induced (necrotic lifestyle).

[jseager7]

Thanks for the clarification. Just so I follow the semantics here, does 'pathogen-associated' imply the quality of PATO's 'associated with' (PATO:0001668)?

That is, "a structural quality inhering in a bearer by virtue of the bearer's being in close proximity and physically interacting with another entity." In this case, a pathogen being in close proximity and physically interacting with a host causes cell death in the host, and that causes lesion formation.

[ValWood]

PATO:0001668 sounds right to me.

This would probably be the correct quality for all "pathogen host interaction phenotypes" because we are describing phenotypes that are observed as a consequence of the pathogen and the host interacting.

In this instance that interaction causes cell death. The observed phenotype is a lesion (a group of dead cells)

@CuzickA @mah11 does that sound correct?

[mah11]

I don't know this biology well enough to comment on accuracy from that point of view, but I'm not aware of any problem with this use of the PATO term.

[CuzickA]

This is what the branch looks like

Some of these terms have related synonyms containing 'cell death' but do not seem to appear under search for 'cell death'. Do I need to make sure each term has an exact synonym containing 'cell death'?

[ValWood]

an exact synonym will ensure that it appears on a search. But it needs to be completely exact in meaning.

[CuzickA]

Maybe presence of pathogen-associated host cell-death resulting in lesion(s) absence of pathogen-associated host cell-death resulting in lesion(s) presence of pathogen-associated host defense cell-death induced lesion(s)

[jseager7]

@CuzickA What's the exact definition of this term? Does it already exist in PHIPO?

'presence of pathogen-associated host defense cell-death induced lesion(s)'

Edit: Ignore my term suggestions; I didn't read the term name properly.

[CuzickA]

@CuzickA What's the exact definition of this term? Does it already exist in PHIPO?

'presence of pathogen-associated host defense cell-death induced lesion(s)'

Edit: Ignore my term suggestions; I didn't read the term name properly.

The term is PHIPO:0000461 presence of pathogen-associated host defense induced lesions Def: A phenotype where the process of host tissue cell death causing a host lesion is induced by the host activating its own programmed cell death pathways in defense, is present.

[jseager7]

I think this would be easier to understand, even if it is a long term name:

'presence of pathogen-associated host induction of cell death resulting in lesion formation'

Assuming the cell death is of a specific type (e.g. apoptosis), we could refine the term label:

'presence of pathogen-associated host induction of apoptosis resulting in lesion formation'

Although personally it sounds like this term contains two separate phenotypes: 'induction of cell death in response to pathogen infection' and 'lesion formation caused by cell death'. They're causally related, but only one of them is strictly causally associated with the pathogen: the pathogen causes the host to induce cell death, and the cell death causes the lesion formation.

Terms which include causal chains like this could be very difficult (even infeasible) to describe with uPheno patterns. That's not so important now because we can't define any pathogen-host interaction terms at the moment unless they have corresponding GO process terms, but I think it's something to be wary of.

[jseager7]

Note that the GO term that seems to correspond to 'induction of cell death in response to pathogen infection' is 'programmed cell death induced by symbiont' (GO:0034050), and from what I can understand from the definition, it's possible that GO subsumes the concept of 'pathogen-associated' under 'induced by symbiont'. The term is a subclass of 'defense response to other organism', so it seems unlikely that the intended meaning of the term is that the pathogen is inducing host cell death in order to benefit its life-cycle.

[ValWood]

'programmed cell death induced by symbiont' (GO:0034050) is not the right term. This term is for use of annotations by the host

Definition (GO:0034050 GONUTS page) Cell death resulting from activation of endogenous cellular processes after interaction with a symbiont (defined as the smaller of two, or more, organisms engaged in symbiosis, a close interaction encompassing mutualism through parasitism). This can be triggered by direct interaction with the organism, for example, contact with penetrating hyphae of a fungus; or an indirect interaction such as symbiont-secreted molecules.

comment Comments Note that this term is to be used to annotate gene products in the host, not the symbiont. To annotate gene products in the symbiont that induce programmed cell death in the host, consider the biological process term 'positive regulation by symbiont of host programmed cell death ; GO:0052042'.

the correct term is GO:0052042 positive regulation by symbiont of host programmed cell death

or more likely one of the descendants GO:0034055 effector-mediated induction of programmed cell death in host

[ValWood]

Although personally it sounds like this term contains two separate phenotypes: 'induction of cell death in response to pathogen infection' and 'lesion formation caused by cell death'. They're causally related, but only one of them is strictly causally associated with the pathogen: the pathogen causes the host to induce cell death, and the cell death causes the lesion formation.

This is a good point, but I think it is only a single phenotype we are trying to capture. The pathogen is causing cell death. Cell death is observed is via lesions (dead cells). So lesions are the observed phenotype, but the underlying process is cell death.

We don't want two sets of terms really, one describing lesions and one for cell death. we could have "abnormal cell death" process terms and has_output "lesion_formation"

but ideally I think we only want one set of terms with liberal synonyms otherwise it will be confusing for curators.

We could make the primary labels "cell death" terms (which match the GO processes better)

'pathogen-associated host induction of cell death present' narrow 'presence of pathogen-associated host induction of cell death resulting in lesion formation'

this would allow the correct term to be located whether the user searched on "cell death" or "lesion formation" which was my concern.

[ValWood]

...also looking at the terms as they are they are inconsistent because they oscillate between matching the GO process (cell death) and "lesion formation". So I'm not sure if they are under 'biological process phenotype" or "physical quality phenotype"?