search

ebi-chebi / ChEBI

Chemical Entities of Biological Interest (ChEBI) is a freely available dictionary of molecular entities focused on ‘small’ chemical compounds.
https://www.ebi.ac.uk/chebi
Creative Commons Attribution 4.0 International
42 stars 10 forks source link

ontology revision #409

Open muthuvenkat opened 19 years ago

muthuvenkat commented 19 years ago

We'd like to revise the ChEBI ontology quite dramatically. The actual editing will be done on the database but for a moment we are discussing the higher level structure. I am attaching Excel file created by Alan McNaught containing comments on some higher level categories.

Reported by: kiri11

muthuvenkat commented 19 years ago

ontology revision

Original comment by: kiri11

muthuvenkat commented 19 years ago

Original comment by: kiri11

muthuvenkat commented 19 years ago

draft ChEBI ontology in GO flat format

Original comment by: kiri11

muthuvenkat commented 19 years ago

Logged In: YES user_id=802919

Here is a draft in old GO flat format (easier to look at even without DAGedit). Please ignore all identifiers assigned by DAGedit. First, I have divided the "ChEBI" ontology into three domains: structure (StruChEBI), biological function (BioChEBI) and (human) applications (AppliChEBI). (These nicknames are just for fun until we find nicier ones.) I did not go into any details in structure since that will reproduce the ontology Alan works with. At the upper level we have three main categories: "molecular entities", "groups" and "elements" (same as "atoms"). Note that "elements" also are partof "molecular entities" and "groups" also are partof "molecules". The BioChEBI domain is something that GO people probably will use for molecular function while AppliChEBI does not have direct analogue. To illustrate: "vasopressin" belongs "hormones" in BioChEBI but also to "antidiuretic drugs" in AppliChEBI; "antibacterial antibiotics" belong to "antibiotics" in BioChEBI but also to "antibacterial drugs" in AppliChEBI etc. Of course there could be multiple use of the same compound in AppliChEBI (check the fluorescein example).

Original comment by: kiri11

muthuvenkat commented 19 years ago

Logged In: YES user_id=451873

It looks good. I think splitting it into the three sections works really well - the AppliChEBI should be useful to GO (although there's a debate as to whether applications, like drug, should remain in GO at all because they are human interpretations not naturally exisiting types, but that's another story!).

In the structure ontology, the two classes that stood out (this is on the Excel spreassheet rather than in the ontology file, so you may not actually be planning to include them at all) were natural product classes and organic functional classes - it may just be my ignorance of chemistry, but does natural product refer to how the chemical was synthesised (secondary metabolite)? If so, I'd be a bit wary about including because you're including information about the origin of the chemical rather than the structure - we've always had problems when we've tried to do this in GO (e.g. experimental v/s functional information). Likewise with organic functional classes - is this referring to the function of the chemical? If so, should be in your function-based ontology instead. But I may be just mis-interpretting what these classes mean in which case ignore me!

Original comment by: jl242

muthuvenkat commented 19 years ago

Logged In: YES user_id=1225878

We can reassure Jane:

(i) that 'natural product' refers to structure type (e.g. steroid, terpene, tetrapyrrole), not synthetic route;

(ii) that 'functional group' denotes structural feature (e.g. ketone, amide, lactone), not chemical activity.

Original comment by: mcnaughta

muthuvenkat commented 19 years ago

Logged In: YES user_id=451873

If that's the case then, I think it all looks great. Looking forward to when we can align the structural chemical ontology with GO.

Original comment by: jl242

muthuvenkat commented 18 years ago

Logged In: YES user_id=802919

I suggest now to include the formulae (equal to the symbol of the element) and graphic representation (elemental symbol without any charges) with element entries. This is a departure from current ChEBI practice. However, the element=(single) atom (even though it is not a molecular entity) and its formula is a valid concept, much like formula for a group. See an example (CHEBI:33353).

Original comment by: kiri11

muthuvenkat commented 18 years ago

Logged In: YES user_id=802919

As a result of today's (17/01/2006) discussion, we decided upon the use of singular terms for classes that truly belong to biological and application ontologies. Thus we will have "drug" instead of "drugs", etc.

Do everybody agree with this?

Kirill

Original comment by: kiri11

adekker2 commented 8 years ago

chebiX.ontology.zip