New terms request: Diet and Human Diet #108
Comments
|
A related issue is which axioms to use to connect diet and human diet to the following processes:
For now, SDGIO says:
|
|
You may be interested in a schematic we were thinking of in FoodOn that separates out "food selection process" and digestion process as parts of food consumption. None of this is finalized, but may provide some ideas. Esp. for humans it distinguishes between foods that are desired to eat, vs. those that are possible to eat. Feedback welcome! What kinds of consumption patterns do you anticipate modelling?
|
As defined, this is synonymous with 'eating'. I think it should contain qualifiers to the effect that this is a habitual, long term process . And is it just the process? Under the existing definition, if I change the process (say changing from mastication and swallowing to feeding through a tube) and the materials remain the same (let's say I eat Soylent under both processes), am I changing my diet? Under this definition, yes, because the process changes. But I don't think this is intended. I think diet is more like climate EnvironmentOntology/envo#237 either a sum total of qualities (quantities of types of food by time) or a disposition. I prefer the former as being able to break down diet into its parts is clearer. E.g. someone's once a week big mac and fries habit is part of their diet. |
|
I realize I'm throwing a spanner in the works, but wouldn't diet be more appropriate to describe the suite of things an organism eats rather than the eating process itself? |
|
agreed
…On Fri, Jun 7, 2019 at 4:42 AM diatomsRcool ***@***.***> wrote:
I realize I'm throwing a spanner in the works, but wouldn't diet be more
appropriate to describe the suite of things an organism eats rather than
the eating process itself?
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#108?email_source=notifications&email_token=AAAMMONWSFMNNFFKASTE2ADPZJCUDA5CNFSM4HT4LE62YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODXFSWIQ#issuecomment-499854114>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAAMMOL2JR6UGI26IALYJHTPZJCUDANCNFSM4HT4LE6Q>
.
|
|
Yep! |
|
@Mttbnchtt I'd go with @diatomsRcool's position on this. Place the temporary class for human diet under object aggregate for now |
|
I don't think that's quite right either. The objects are spread out in
time. It could be construed as an aggregate of qualities. We also have to
think how this fits in with ECTO. It ciuld be conceived as a series of
interactions between an organism and edible material entities
…On Fri, Jun 7, 2019 at 1:09 PM Pier Luigi Buttigieg < ***@***.***> wrote:
@Mttbnchtt <https://github.com/Mttbnchtt> I'd go with @diatomsRcool
<https://github.com/diatomsRcool>'s position on this.
Place the temporary class for human diet under object aggregate for now
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#108?email_source=notifications&email_token=AAAMMOOTOFONW23Q3QUJ7STPZK6AZA5CNFSM4HT4LE62YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODXG3ZCI#issuecomment-500022409>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAAMMOOBGIYWFVNCHL7LCP3PZK6AZANCNFSM4HT4LE6Q>
.
|
|
An aggregate of qualities? Not following, sounds a bit too abstracted. There's temporal extension, agreed. The sum of interactions would be the sum of feeding processes rather than the diet though. Given the sum of feeding processes the organism is the agent of over some time period, all ME participants in that sum that realise a food disposition (or simply which have been consumed?) could then be the constiuents of the diet. But what would this sum of MEs be placed under? Feels like a some form of temporally scattered OA to me (4D worms rearing up again). |
|
Thanks so much to everyone for the great inputs. Following a suggestion of @pbuttigieg (in a personal conversation), I temporarily put diet under material entity in the SDGIO and human diet under diet. I will watch how this discussion develops and think more about the issue. I will make adjustments as necessary and describe them here. Thanks again very much. |
|
I'm not usually one to argue finer philosophical points of upper ontologies, and in general I vastly prefer usage of plain and straightforward classes from BFO like "ME" rather than the more esoteric ones. However, if we are using UOs we do need to be sure we use them consistently. One of the key conceptual distinctions in BFO is between continuants and occurrents, and the key things that distinguish them is the nature of parthood. With a C, parthood holds at any given time. From the BFO2 reference: Continuant entities are entities which can be sliced to yield parts only along the spatial dimension, yielding for example the parts of your table which we call its legs, its top, its nails I don't think this is true for diets. A plate of food can certainly be sliced in this way, dividing my big mac and fries into two parts, with the big mac further subdividing into burger, bun, lettuce, etc. However, a diet is necessarily subdivided into temporal parts (assuming normal eating patterns). My diet may consist of a plate of cereal in the morning, a salad at lunch, a large meal at dinner, repeated with variation over a week, etc. I agree that treating as a occurrent is pretty odd, but if you do want to conform to BFO then this is more like a history than a ME. I agree aggregate of qualities/quantities sounds kinda wacky but let's look at some use cases. The concept of a high fibre diet would be very useful, for example, to add statements in a knowledge graph stating high fibre diet --> increased microbiome health (to simplify the biology immensely). Similarly we may want to make a claim increased red meat diet --> {increased GHG, increased risk of certain types of cancer}. Here, just representing the diet as a ME yields incorrect statements. It is not the presence of some portion of fibre or some portion of flesh that is the subject of the statements, it is the levels. Of course, representing a concept like "increased level" is hard (relative to what?) but we are at least in the same boat we are with phenotypes and environmental variables. It is use cases like this that lead me to the aggregate of quantities representation. The aggregate part is necessary as we will want to break down diets into their parts. E.g. a specific carnivorous diet that has high calory and low fibre components. |
|
I'm not buying the aggregate of qualities, I'm afraid. |
|
There are different senses of diet. We will adopt a "high fibre diet" and a few others, these operate at a more generalizable analytic level. One thing we're anticipating doing in FoodOn is distinguishing food availability, and food selection according to diet. Need to describe something like "subsistence diet" that includes a minimal type of food required - but how that translates out to food selection is tough. I can eat bugs to survive, fulfilling elements of a subsistence diet. In other cultures that can be a mainstay. Would we call a "personal comfort diet" the list of food choices a person has to describe their eating pattern? |
Sure, but we need a good basis for these that can be applied across ecology and food science.
What do you mean by "adopt" here?
Doesn't availability precede selection, which precedes diet? Unless you're talking about a predicted diet (what an organism typically eats), rather than a "realised" diet (what it actually eats / has eaten)
So you're bundling food requirement with diet here? I think we need to unpack that carefully.
Sure. A robust definition of 'diet' should be able to accommodate this. In the SDG context (nutrition, poverty semantics), subsistence means just that: is the diet of sufficient quality to allow survival.
Again, this is more a diet profile or something similar, not the diet itself. The diet is just what has been eaten. The behaviours and preferences that go into selection of foods that compose the diet would be handled separately, I think. |
|
I see what @cmungall is saying about an aggregate of qualities, but how about just making diet a quality? We often categorize organisms by their diet, similar to other qualities like being arboreal or terrestrial. There's almost three senses of diet. There's the quality, like having a monophagous diet, and then there's the behavior like dieting, but then there's the actual diet, like the mediterranean diet that is a collection of potential food items. Maybe it would be easier to think about a "high fibre diet" as a collection of food items that someone could eat and still be conforming to that diet instead of the cumulative collection of food items that person has eaten over time? that way it could be an object aggregate |
|
I'm still mulling this. In addition to diet, USDA also uses "food patterns" (https://www.choosemyplate.gov/brief-history-usda-food-guides), "food style" and "eating pattern" (https://www.choosemyplate.gov/dietary-guidelines). It seems like "Mediterranean diet", "high fibre diet" enumerate material components of a diet, or mixture ratios of them, not qualities. Would people agree that food qualities are reserved to directly measurable aspects of the food - color, smell, texture, taste, weight? |
|
qualities can encompass ratios, as well e.g. increased levels of X
… |
|
Ah right I see you made that point above. I'm looking at PATO to look for increased ratio rather than just level. I see "increased amount" or "increased intensity" but no "increased ratio" or "increased proportion". Seems like a new general class is needed under PATO "increased amount"? |
|
Food qualities should be color, smell, etc. but I'm not sure that diet qualities have to be this way. I'm not really sure what to do about the additional USDA terms you mentioned. I'm still hovering around thinking of a diet as a collection of potential food items, not a specific list of items that a specific organism has eaten throughout its life. Like a mediterranean diet includes X. A vegan diet includes (or excludes) X. A carnivorous diet includes X. I'm not sure that helps at all. |
|
Surely not all diets are boolean exclusion/inclusion combos?
PATO quality is very general, not just smell, color, etc. Anything that
characterizes some aspect of a food or diet is game, including ratios,
quantities, propensities to have some kind of effect, higher order patterns
…On Fri, Aug 2, 2019 at 1:33 PM diatomsRcool ***@***.***> wrote:
Food qualities should be color, smell, etc. but I'm not sure that diet
qualities have to be this way. I'm not really sure what to do about the
additional USDA terms you mentioned.
I'm still hovering around thinking of a diet as a collection of potential
food items, not a specific list of items that a specific organism has eaten
throughout its life. Like a mediterranean diet includes X. A vegan diet
includes (or excludes) X. A carnivorous diet includes X.
I'm not sure that helps at all.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#108?email_source=notifications&email_token=AAAMMON5US6THHVYDQC7YUDQCSKZTA5CNFSM4HT4LE62YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD3OYZII#issuecomment-517835937>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAAMMOJAF3V4VWJFYSD5XXLQCSKZTANCNFSM4HT4LE6Q>
.
|
|
Pontificating a bit: On the one hand, ontology wants to be descriptive, capturing the different senses of a term as reflected in diverse speech community usage, preferably with the sense reflected in the label. In the face of a word like 'diet' that has multiple semantics, ontologists may want to furthermore be proscriptive, reserving a root term for one sense (along the lines of the French Academy control of french language semantics) and providing variants for other senses, essentially re-separating homonym usage. So to make everyone happy, is there a generic diet definition that anticipates/provides the space for variants which reflect differentiae between intended eating pattern via quality "increased ratio" some X; and measured eating, a sample (enumeration/aggregate) carried out over some duration of time. See "diet" usage examples at: https://dictionary.cambridge.org/us/dictionary/english/diet diet: a pattern of food consumption pertaining to an individual or sample of consuming organisms of some type. And then the processes as desired to support above more specific diet objectives. Now, (aside from discussion on definitions) do you want to host these terms in SDGIO, or FoodOn? I do volunteer to add diet variants as they come up in FoodOn if you want, so it is an SDGIO question of whether you want to support the root term variants (Mediterranean etc) or not? I'll say our food consumer domain so far is just human and livestock though. |
|
Going to pick this thread up over in where we now have a longer list of diets: FoodOntology/foodon#74 |
|
So only humans have diets?
…On Fri, Dec 13, 2019 at 1:19 PM Damion Dooley ***@***.***> wrote:
Going to pick this thread up over in where we now have a longer list of
diets: FoodOntology/foodon#74
<FoodOntology/foodon#74>
There I am pitching the idea that a diet is like a protocol involving
something like an inclusion criteria (
http://purl.obolibrary.org/obo/OBI_0500027) but pertaining to food that
is chosen to be consumed.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#108?email_source=notifications&email_token=AAAMMOPUFWI55T7XWKIE6ATQYP35JA5CNFSM4HT4LE62YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEG3I5HI#issuecomment-565612189>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAAMMOLSNWXSY3XC6FG55FDQYP35JANCNFSM4HT4LE6Q>
.
|
|
Are you thinking of FoodOn's scope? Its true we're starting out human and pet centric. But we expect to move on to agricultural & lab animals as consumers. Other than providing general terms for all organism food consumption processes though, it will be beyond our capacity for a while to talk about natural food chains - hence folks like ECOCORE I think getting into that area. |
|
Most organisms don't follow a protocol when eating...
…On Fri, Dec 13, 2019 at 1:59 PM Damion Dooley ***@***.***> wrote:
Are you thinking of foodon's scope? Its true we're starting out human and
pet centric. But we expect to move on to agricultural & lab animals as
consumers. Other than providing general terms for all organism food
consumption processes though, it will be beyond our capacity for a while to
talk about natural food chains - hence folks like ECOCORE I think getting
into that area.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#108?email_source=notifications&email_token=AAAMMOOLIINCDXUEPZWBYOTQYQASFA5CNFSM4HT4LE62YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEG3LYLY#issuecomment-565623855>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAAMMOOT7CT4C6CGXW2WMODQYQASFANCNFSM4HT4LE6Q>
.
|
|
Ah, right. The proscriptive (protocol & plan) and descriptive (observational) language distinction. A cluster of animals eating the same way, so we infer that biologically that type of animal in that environment eats a certain way, and we can generalize to vegetarian, carnivore etc. But we don't characterize their "hardware" layer as following a plan; there's no "intentionality" to their behaviour. Meanwhile for humans, we ascribe anything more than a survival diet as a matter of choice, preference, a plan to adhere to. Reminds me of the similarity in expressing an evolutionary function's goal - and a process objective - using the same language. Can we talk about diet as an objective/goal and leave it open as to whether evolutionary function is realized in a process, or a more intentional process is occuring, to achieve them? Can "Objective Specification" (http://purl.obolibrary.org/obo/IAO_0000005) do this? It is part of a plan specification which can be executed by a process; say a "carnivore objective specification" executed by a "food hunting" and "eating processes" of an organism? |
|
This is just making it too complex IMO. What if it's a sick animal, and its
diet is a result of neither evolution nor intention?
This is where I disagree with many other OBO people on modeling. I think
that many shadow classes for protocols for X are superfluous and confusing.
Just model X. And if you really want to talk about a protocol for X, just
make a class expression.
If you want to talk about inclusion criteria, then just use OWL axioms to
model this directly - e.g. N+S conditions using concept inclusion
(SubClassOf)
In GO, all BPs are evolved and have some objective honed by evolution. We
*could* model these using some kind of information entity... but why?
I can appreciate there may be some situations where this additional degree
of reification is required, though I am often skeptical, and I really don't
think you need it here.
|
|
A pox on over-modelling, all those axioms blowing in the wind! But there are competing senses of diet to accommodate in a class hierarchy, as I am reminded when reading above @diatomsRcool notes. Your examples @cmungall @diatomsRcool dowse for the most general concept under which variants ideally fit. I actually do have to model some stuff here though to anticipate describing data in a project we're working on for the USDA's Food Data Central portal onto a collection of databases, one of which is based on food surveys. So, persevering: That an individual animal or human's diet is potentially random or not healthy - and other extremes like force feeding - entail dropping individual intentionality, and even hunger as part (as causal factors) of one sense of diet, leaving just an observationalist record of food consumption. In OBI this would be derived from an organism history - a set of ME foods consumed during a series of an organism's food consumption processes/events. label: 'ingested food set': This could be derived from: label: 'ingested food log' label: 'ingested food' ... one could add terms for other statistics derived from the ingested food log Earlier diet was defined as a set of food proportions, so not referencing ME's directly. It is pointed out that PATO qualities are not just qualities of a thing observed at some instant in time, but can be counts - quantities, or calculations - ratios (whether categorical "more/less" or numeric "80%") so in the realm of IAO ICEs. The proportions are with respect to some open-ended "norm", or with respect more particular food groups - but lets avoid modelling that now, and just note that outright exclusions are needed, e.g. "A vegan diet contains no animal products." Technically this sense of diet sounds a lot like a "dataset of features" (http://purl.obolibrary.org/obo/OBI_0000649) about an organism's ingested food: A data set that is produced as the output of a descriptive statistical calculation data transformation and consists of producing a data set that represents one or more features of interest about the input data set. . This would cover both intentional plan/goal and descriptive perspectives, and could include constraints on timing of ingestion. label: "diet" label: "mediterranean diet" etc. One could add "objective specification" parts that contain diet constraints. |
|
I'd probably tweak the 'ingested food set' definition: |
|
Treating "diet" as a dataset of features, please no no no. At least call this something like "diet dataset". Diets exist outside datasets. Dinosaurs had diets. Diets have causal relationships to health and disease, datasets not so much. Last word. You should go the process (behavior) or quality route, in each case modeling the thing in the world directly, not modeling our model of them. I don't see any disadvantage to either. Finally responding to @diatomsRcool:
This is fine, aggregates of qualities are just qualities. In fact many things that we think of as 'atomic' qualities may be aggregates (e.g. color = H+S+V). I just emphasized the aggregate part to be clear that combinations of inclusions/exclusions would not be a problem.
unfortunately plain object aggregate doesn't work for the reasons I've outlined above. I'm not sure I see a difference between your 1 and 3. I agree that behavior can be ontologically separated |
|
"diet dataset" .... Or "diet feature set" or "diet specification". Ok. So we keep the "diet" entity one step away from the specification of a diet. About "object aggregate" - is there something like "object class aggregate"? I don't mean to refer to particular instances of food items, but rather their classes. An "ingested food set" would hold the classes of things ingested, not instances which are time dependent (and indeed go out of existence as such if truly ingested.) |
|
I'm copying this over from FoodOn issues - it addresses both threads, though work below pertains to adding terms to FoodOn. OBI call (this morning) was informative. Call reinforced not to mint a basic "diet" term. All the more particular senses of diet live in different places in BFO etc. class/concept hierarchy. Instead, to connect the word "diet" as people might use it in a search engine to these varied term entities, we can use the hasBroadSynonym annotation. 'ingested food log' survived critique in the call as a clear case of a data set. Should we label it as "... log" or "... record" or "... data set"? And I just found "ingested food" already exists as an UBERON term so we will reuse that. The 'ingested food set' was agreeable but it does sound problematic to have it be an "object aggregate" as the aggregate is intended to reference a collection of object instances at some time t, whereas the proposed set is only meant to reference the classes of food object eaten. So I would revise "ingested food set" to simply be a type of dataset too, derived from the food log dataset. FoodOn may add this term in the future. My above "diet" term was roundly discouraged as the "dataset of features" parent was too tied to parameters of an actual dataset. People preferred "dietary regimen" as a subclass of protocol, so - and here I welcome definition edits - something like: label: "dietary regimen" Another suggestion (Bjorn's?) was to define classes of dietary food - a "vegetarian food", a "carnivorous food" - here one can define in an equivalency exactly what is permitted or not in the class by reference to other food types. I really like this. Under "FoodOn product type, I would add a convenience class "food product by dietary group" class to gather the various diet categories. label: "food product by dietary group" Then subclasses can have equivalency axioms that spell out the dietary group: label: "pescatarian food" This does expose that FoodOn is focused on "food products", a terminology that is somewhat alien to natural food ecosystems. The https://www.dietaryguidelines.gov/sites/default/files/2019-05/2015-2020_Dietary_Guidelines.pdf document conveys in its "eating patterns" not just the set of foods to eat, but also amounts, so I avoided using "eating pattern" in label as it may carry this extra level of detail. We can then simply say some organism is 'consumer of' [some | only] 'pescatarian food'. (It isn't defined in RO at moment but 'consumer of' would be inverse of 'has consumer'.) A study design can [paraphrasing] have an associated 'dietary regimen' protocol with objective (objective specification) to provide 'oral administration' (a planned process) of certain food (a 'food product by dietary group') according to some schedule. |
|
Note that "food product by dietary group" is a metaclass, this doesn't work in OWL. |
|
Its a convenience class - so would have no logical differentiae, no axioms other than being a subclass of "foodon food product". I presume that's tolerable? |
|
I'm only chiming in here to comment on the ecocore implications.
I think this is fine. We just come at it from a different angle, by saying that an organism is a piscivore or an omnivore, etc. I think the logic will work, but we don't usually refer to any organism other than humans as being pescatarian and I doubt FOODON wants to start including non-human foods. Otherwise, sounds good to me. |
|
Righto, now |
@pbuttigieg
I have just inserted the process classes Diet and Human Diet into SDGIO. Their definitions are temporary and will be superseded when the terms will be added to Ecocore.
For now, we have the following:
Prent: process.
Parent: diet.
The text was updated successfully, but these errors were encountered: