add exposure of organism class
Closed this issue · 41 comments
In ECTO, we are wanting to adopt COB as the upper-level ontology and align ECTO's exposure event class with RO's exposure event or process class. We have submitted ticket to RO about this. See oborel/obo-relations#599.
To help with this, it would be very useful if exposure event was added to COB. For background on the ECTO issue see EnvironmentOntology/environmental-exposure-ontology#58.
I am happy to make the PR. I was wanting feedback first.
Minor clarification: That's an ExO class not an ECTO class.
I think this is in scope for COB esp if it's a domain of an RO class
@bpeters42 I am happy to coordinate with you. @matentzn @diatomsRcool and @laurenechan need to be involved too. Do you want to set up an ad-hoc call or join one of our meetings?
COB SOP wise, is it necessary to determine the intended design patterns for getting a class into COB? I think if we agree that a term is a useful (of course given its precise human readable definition) upper level concept, we can create in COB (even defer BFO alignment if necessary!) to parallelise the efforts of COB alignment (ontology->COB), BFO alignment (COB->BFO), and design pattern definition (exposure event has stimulus some COB:stressor or whatever).
@wdduncan can you put forward the working IAO:115 definition for "exposure event"?
For induction of autoimmunity, certainly in many cases there is exposure to a self-antigen in a particular context, such as tissue damage caused by a pathogen or a wounding event, in combination with a danger-signal present in that context (bacterial product like LPS, or even self-proteins, like histones or forms of IL-1) that changes the functioning of dendritic cells presenting a self-antigen from tolerogenic to capable of driving an immune response. The exposure in this case is internal, though there may be a precipitating event from an external stimulus. And there are probably spontaneous types of autoimmunity that do not even require the external precipitating stimulus, but still have an internal exposure event that is not tolerogenic but rather immunogenic. (should probably look up a bunch of reference here if I had time)
At a general level, we all seem to agree that these entities are involved:
- exposer (poser?): the entity that in some way interacts with another entity
- exposee: the entity that is acted upon
- exposure event: process in which the exposer interacts with the exposee
We can (of course) define roles for exposure/exposee. E.g:
exposure event equivalent to:
process and
(realizes some exposer role) and
(realizes some exposee role)
But, as noted, there will problems with things like exposure to noise.
Perhaps we can use relations (e.g., involves exposer/exposee) to specify the exposer/exposee. E.g.:
exposure event equivalent to:
process and
(involves exposer some entity) and
(involves exposee some entity)
Or if we assume that the exposee is always a material entity that is exposed to to something else, this might work:
exposure event equivalent to:
process and
has participant some (material entity and exposed to some entity)
@bpeters42 I agree that what counts as an "exposure" can be vague. I left the domain as material entity b/c I was thinking cases in which soil is exposed to a contaminant. But if the least controversial route is to restrict the domain to organism, then that works too. We may want to modify the label, though.
@bpeters42 We discussed this issue on today's ECTO call. We think limiting exposure event to organisms is too restrictive. Reasons for this are situations such as:
- exposure of a cellular component to some X
- exposure of DNA to some Y
cellular components and DNA are not organisms.
If OBI or ECTO needs a more specific exposure of organism class, the class can be a specialization of cob:exposure event.
I agree with @wdduncan. @bpeters42 is there a specific reason you would like exposure events to be restricted to organisms? Since OBO covers quite a few environmental and agronomy ontologies, there are even more examples I can think of, like exposures of soil to fertilisers etc.
We agree that the scope of ecto is about organisms and things that organisms are exposed to. However, cob is meant to be general, and it doesn't seem crazy (to me at least) that non-organism exposure terms of the kinds suggested might need a parent class in cob.
How about we add both exposure event and exposure of organism:
- exposure event
- exposure of organism
The existence of a hierarchy with only one child class may cause some concern. But, there are already a number of one-child hierarchies already in cob (cf. measurement datum, cellular organism, function, etc.)
Hey folks, I recently discovered this thread and wanted to provide some input since I've been doing related work and was actually preparing to bring some proposals to the ECTO/EXO folks. Some of the issues identified thus far in this thread with how exposures are represented in ECTO and EXO are similar to what I have identified, but there are a few more things worth mentioning:
- Not all exposure events involve physical contact with a material entity (e.g., victims of sexual harassment may never come into physical contact with their abuser)
- The definition of 'exposure stressor' in EXO implies that exposures are harmful, yet many have a protective or beneficial effect on the organism (e.g., exposure to sunlight can be psychologically beneficial)
- Whether the exposure is harmful or beneficial is often determined by the quantity of the thing that the person is exposed to acutely or the duration of the exposure (e.g., excessive exposure to sunlight can be harmful to one's skin)
- Many exposures are more or less harmful/beneficial depending on what stage in life the organism is in (e.g., exposure to lead in indoor paints is harmful to children, but not so much adults)
- In some exposure events, there is a separate process that the thing to which one is exposed is a participant in and that is relevant to a particular health outcome (e.g., exposure to gamma radiation or exposure to rain). Therefore, in some cases, it will be necessary to capture in one's representation of the exposure the fact that the material entity that a person is exposed to is simultaneously participating in a secondary process.
Taking all of this into account and with some input from a domain expert, I recently put together (independently of this thread) the following three proposals for representing different types of organism-environment exposures:
- environmental exposure process =(def.) A process p that is causally upstream of some response to an external stimulus [GO:0009605] that occurs in some organism by virtue of that organism participating in p.
- environmental exposure process via physical contact =(def.) An environmental exposure process that happens via some physical contacting between an organism and a material entity.
- environmental exposure process via physical contact with a material entity that is participating in an environmental process =(def.) An environmental exposure process that happens via some physical contacting between an organism and a material entity and that is a temporal part of some process the material entity is participating in.
Here is how I propose defining 'physical contacting with organism':
- physical contacting with organism =(def.) A process in which some material entity overlaps with or is adjacent to an organism that it was previously separated from.
The labels, especially the third one, can definitely use some tweaking. I think the three exposure definitions capture all of the complexities of exposures that have been mentioned in this thread, while also (1) avoiding the receptor approach and (2) linking the exposure event to internal biological processes via GO's 'response to an external stimulus'. Lastly, my definition of 'physical contacting with organism' tries to capture the fact that the material entity touches, enters, or penetrates the organism after having previously been physically separated from the organism.
I like where you are going with this @dillerm. I like the mechanistic approach. But this may be overextending for social exposures, where it's not clear the GO process is appropriate? And even for some chemical exposures, the chemical may cause damage without the organism responding?
Hi @cmungall , apologies but I somehow missed the email notification of your reply. GO's 'response to external stimulus' is currently defined as "Any process that results in a change in state or activity of a cell or an organism (in terms of movement, secretion, enzyme production, gene expression, etc.) as a result of an external stimulus." Unless I'm misunderstanding what you mean, I think this would cover chemical exposures that cause damage without the organism responding since that damage would still alter cellular activity in the regions affected.
Certain social exposures that have a protective effect or that act as a safety net for individuals might be trickier, but here is what I envisioned for something like "access to health insurance" (for the sake of argument, let's pretend there is a class labeled 'health insurance enrollment' that is a subclass of 'process'):
We can create a class, labeled 'access to financial resources', define it as being a subclass of 'environmental exposure process', and assert that it is causally upstream of some process that itself is causally upstream of some response to an external stimulus in some organism. I think this works because causally-upstream-of is defined as "p is causally upstream of q if and only if p precedes q and p and q are linked in a causal chain." We can simultaneously assert that 'health insurance enrollment' is causally upstream of some OGMS:treatment that itself is causally upstream of some response to an external stimulus in some organism. This idea still needs to be thought-out and developed more, as there are a few caveats that come to mind:
- It may be ideal to have a different object property than causally-upstream-of to avoid implying that every instance of 'access to financial resources' is causally linked to some subsequent process that affects a person's health. This object property should still allow for the exposure process to indirectly influence a response to external stimulus. I will look at what currently exists in RO and other ontologies to see if anything fits the bill.
- There are many things, like food, medical treatment, and housing, that access to financial resources are tied to that the definition will need to consider.
- The definition and axiom should also state that money has a causal role to play. Unfortunately, a good representation of 'money' is absent in OBO Foundry ontologies, although this is something we are currently working on adding to OMRSE.
After much discussion, the ecto group has decided that only the exposure of organism class would be of use.
Can some make a PR for this?
Depending on what you mean by "contact" (i.e., whether it's limited to physical contact or also includes social contact), you might want to change the label to something like 'physical exposure of organism'. Additionally, something to possibly keep in mind when creating the exposed-to relation is that some of these exposures are symmetric insofar as both parties are affected by the exposure process (this point is probably more salient when you think of two organisms interacting).
I think we do want to retain the "social contact" piece since we do have data that includes exposure to things like violence and poverty. We also need to represent the effects of "exposure to exercise" but I'm not sure that is a good way to model these behaviors.
Perhaps the notion of interaction may be of use.
exposure of organism = A process during which an organism an interacts with another entity, and as a result of this interaction, one or both entities undergo stress or tension.
This may be misread all exposures being harmful, but not all stress is negative. I'm not sure a better word to use.
I like your proposed definition @wdduncan , and might I suggest using 'stimulated' ? We had previously struggled with just having 'exposure stressor' in ECTO and wanting a way to suggest more neutral/positive exposure entities. We now include 'exposure stimuli' as a parent of 'exposure stressor' to fulfill that piece.
So then:
exposure of organism = A process during which an organism interacts with another entity, and as a result of this interaction, one or both entities are stimulated
@bpeters42 , for photons, can we just say that physical contact occurs between an organism and either a material entity OR a subatomic particle? I think for sound you can say that the exposed organism is coming into contact with some gas, liquid, or solid matter that is participating in some propagated wave, although 'propagated wave' would have to be defined and given a URI. The third class in my above proposal (see below for definition) was meant to handle stuff like this where the definition of the exposure needs to capture that the thing to which the organism is exposed is participating in another process simultaneously.
environmental exposure process via physical contact with a material entity that is participating in an environmental process =(def.) An environmental exposure process that happens via some physical contacting between an organism and a material entity and that is a temporal part of some process the material entity is participating in.
@wdduncan and @laurenechan , I definitely agree that we don't want to imply that the outcome is negative for the organism, which I think the usage of "stress or tension" does. I'm also not sure that saying the organism is stimulated is the best approach since "stimulated" often implies that there is an excitatory response. Avoiding this issue is actually what first motivated me to try to causally relate the exposure process to some downstream response to an external stimulus.
I think there is very good reason to aim for an exposure class that covers both material and social exposures, even if the processes that the latter depends on haven't been fully represented yet. There is a lot of interest currently in studying how the environment affects health using -omics techniques and "big data," and therefore in building linked datasets facilitate such studies.
I also don't think the current proposed definition of 'exposure of organism' needs to be modified much, if at all, to cover non-material exposures, although this depends on what is meant by "interacts with another entity." If this covers non-material exposures where the organism is exposed by virtue of participating in a process, like exercise (i.e., "participates-in some process" can be used in place of "interacts with another entity"), then the current proposal is fine as is, in my opinion.
As for the example of immune exposures where the immune system does not detect the pathogen (e.g., viral latency), I think it still can be argued that the infecting pathogen still acts in a manner that changes that state or activity of one or more cells. In some cases, such as with certain herpesviruses, this evasion of the immune system is caused by them directly downregulating components of that immune response.
.. things like 'teaching, talking, fighting, whatever can very well be defined as specific processes ..
I agree with your general point here. I don't know what guidance to give as when to use exposure to X vs. X itself. However, the notion of exposure seems to be important in some scientific domains.
The examples of viruses and such escaping detection by the immune system are interesting. Not sure how to fit this into the current approach.
Ah, I see that I misread @bpeters42's previous comment. I don't think that representing social exposures simply as participates-in X where X is a social process is useful simply because participating in X does not always mean you are exposed to X. Useful examples that illustrate this include sexual harassment, battery, education, and employment, where there are multiple parties involved who we would not want to say were exposed even if they participated in one of the relevant processes. In battery, for instance, we would say that both the batterer and the victim are participants in an violent battery process, but we would not want to assert that the batterer's participation is the same as being exposed to battery. In other words, it's often the case that the nature of one's participation in a process is what determines whether their participation counts as a health exposure.
@dillerm The difficult part is trying express the nature of the (so-called) "exposure participation" without sounding circular; hence, the language about "interaction", "stimulation", and the like.
It seems like there are two types of exposure to a process. First is something like teaching where the person being exposed (the student) is a participant in that process. Second is something like a hurricane where the person being exposed is not part of the process. Unless there is some sort of a "bystander" role?
I agree that we should avoid saying that someone can be exposed to a process, but I think one way around this would be my previous suggestion to represent an exposure as a process that has participant an organism and is causally upstream of some change in the state or activity of the organism or some part of the organism (note again that this change can be minuscule and not clinically relevant). The process that the organism is a participant in can be a social process or some process of physical contacting. I think we'll need to develop something different for handling social "exposures" that are defined by having access to financial or social capital (e.g., access to health insurance, access to a social support network), but the above definition should cover most physical and social exposures without being circular.
I also think this approach allows you to represent exposures where, because the organism is in the vicinity of an environmental process, it comes in contact with something that is participating in that process. For hurricanes, this would mean that you are exposed via physical contact to some raindrop(s) that are participant in a water-based rainfall process, some gaseous environmental material that is participant in a wind gust, and/or some liquid water that is participant in some coastal flooding (note: I'm reusing a lot of ENVO term labels here). For radiation, you would be exposed to some photons that are participant in an electromagnetic radiation process (assuming we allow for 'physical contacting' to be between an organism and a subatomic particle).
Of course this again raises Bjoern's point of whether we even need exposure classes for social exposures instead of just representing 'exposure to social process X' as participates-in 'social process X' since my above-mentioned proposal does not represent the nature of the organism's participation in the process. I think the answer to this depends on if we want to assert in every class that represents a social process (e.g., teaching, fighting, working) that that process is causally linked to some change in state or activity of the organism. If we don't, then I think it would be useful to have classes that distinguish between social exposures and social processes.
I'm inclined to agree with Bjoern. We should make progress where we can and have further discussions about exposures to processes. It is clearly a complicated issue. As long as we don't make decisions that box us in for the future.