NTRs: Terms for food products

[pbuttigieg]

Moving discussion from https://github.com/Planteome/plant-ontology/issues/624

@rvita requests several classes to handle food products. These will be pushed to a food ontology once it's up and running, with temporary accommodation in ENVO.

@rvita - could you add the terms with initial definitions here? Thanks! Please also include any terms that ENVO should import from other ontologies to form the logical defs or axioms.

[rvita]

I wanted to model the definition of vegetable food product after that of "plant fruit food product" ENVO_00003877 but it does not have a definition to copy. I guess the term should be "plant vegetable food product" to follow the same format. Its parent term "plant food product" ENVO_00002216 also does not have a definition to copy from, but refers to the wiki page for vegetables. plant vegetable food product is_a plant food product definition edited from wikipedia: any part of a plant that is consumed by humans as food as part of a savory meal, exclusive of other food derived from plants such as fruits, nuts and cereal grains, but includes seeds such as pulses. https://en.wikipedia.org/wiki/Vegetable will that work or do I need more?

[rvita]

according to wikipedia, a nut is a fruit. not sure if you'll want to classify it that way or not? plant nut food product is_a plant fruit food product ENVO00003877 definition edited from wikipedia: A nut is a fruit composed of a hard shell and a seed, which is generally edible. Common usage of the term often refers to any hard-walled, edible kernel. https://en.wikipedia.org/wiki/Nut%28fruit%29

[rvita]

you alreday have leguminous food product, so we do not need to request that one. Thanks!

[pbuttigieg]

We'll work on getting some reasonable definitions in there. These are likely to be updated as the Food Ontology projects mature. Thanks for the input! On 21 Jan 2016 18:53, "rvita" notifications@github.com wrote:

you alreday have leguminous food product, so we do not need to request that one. Thanks!

— Reply to this email directly or view it on GitHub https://github.com/EnvironmentOntology/envo/issues/284#issuecomment-173652845 .

[cooperl09]

I would avoid using the Wikipedia definition of vegetable for a number of reasons. It should not contain pulses (seeds), and they do not necessarily need to be part of a "savory" meal. I would be interested in contributing to the development of a food ontology. Sounds fascinating!

[pbuttigieg]

I would avoid using the Wikipedia definition of vegetable for a number of reasons. It should not contain pulses (seeds), and they do not necessarily need to be part of a "savory" meal.

@cooperl09 : Agreed. If there is a real need to include other definitional dimensions (e.g. part of a savory meal), we can create subclasses to handle them. The top level class should be fairly neutral.

For interest, here's the FAO definition, clearly showing that there is a need to broker between these different and often arbitrary groupings.

I would be interested in contributing to the development of a food ontology. Sounds fascinating!

We'd welcome expert input to help shape it, especially from the plant domain. I'll post on this thread when we get the repo going.

[pbuttigieg]

From: https://github.com/Planteome/plant-ontology/issues/624#issuecomment-173644897

So we would be happy if you could add 'vegetable food product' 'nut food product' and 'legume food product' for now and we can later move to the food ontology.

These are added in dd787fcc162260a43a3f9783a7186936f71d9917 However, I've added vegetable rather than vegetable food product as the latter could be derived from the former and not be, in actual fact, a vegetable at all. I think the same holds true for fruit, so I've added that too with nut as a subclass.

We do have pulse (formerly pulses) which needs to be reconciled with legume (seems to be an arbitrary separation.

These classes are likely to populated by inference rather than assertion as they seem to be overly broad and often fuzzy. This should boost rather than hinder usability if done consistently. A food ontology should be prepared to handle this.

@cmungall @rlwalls2008 is there a more intelligent way to say: 'derives from' some Viridiplantae or 'produced by' some Viridiplantae Feels wrong to have the taxon in there, it's more one or more individuals of that taxon.

@rvita @cooperl09 I'll make a test release so the classes are available, but please note that they are unstable. We'd value your feedback - please create new issues here as needed.

[rvita]

for the food allergies, we want to use the food product terms that you already have for fruit and legume and they are: plant fruit food product ENVO_00003877 leguminous food product ENVO_0010080 b/c the allergy is to the food products. we will also use these existing terms: meat food product ENVO_00002165 plant seed food product ENVO_0010085 cereal food product ENVO_0010002 so ideally for us, nut and vegetable would be similar terms to those existing terms. But all of your input makes sense to me.

[pbuttigieg]

I'll add some more axioms and definition content to those. We can say that they derive from legumes, nuts, etc.

[cmungall]

On 22 Jan 2016, at 8:33, Pier Luigi Buttigieg wrote:

@cmungall @rlwalls2008  is there a more intelligent way to say: 'derives from' some Viridiplantae or 'produced by' some Viridiplantae Feels wrong to have the taxon in there, it's more one or more individuals of that taxon.

There is a well-worked our algebra for developmental relations that could be abstracted and transferred to a more generic production or derivation set of relations.

But a few questions first: what proportion needs to be vegetables? What preparations are allowed? Is pizza a vegetable product? Kale chips? Is a US beefburger a corn product?

Somehow I feel the relational algebra of RO will be a hindrance to inferring arbitrary food classifications...

[rvita]

we spent some time reviewing different groups' food terms and concluded that many are arbitrary and some are controversial. we liked this one best, but it does not necessarily mesh with botany or meet everyone's needs: http://www.mhlw.go.jp/english/topics/foodsafety/positivelist060228/dl/r04.pdf

[pbuttigieg]

for the food allergies, we want to use the food product terms that you already have for fruit and legume and they are: plant fruit food product ENVO_00003877 leguminous food product ENVO_0010080 b/c the allergy is to the food products.

I understand this, but as @cmungall noted, we will probably have to differentiate foods that are primarily composed of some food stuff and foods that have that food stuff as a part. This would be like saying 'nut food product' and 'nut-containing food product'.

With ingredient data, both classes can be filled by inference.

[pbuttigieg]

Food is indeed a very complicated domain, but - after much work - we now have the right team (and ontology) to start to tackle this. Please check out the Food Ontology repository.

This request has been reassigned to https://github.com/FoodOntology/foodon/issues/23