obsoletion request: p-block molecular entity and others defined by block of periodic table

[cmungall]

ChEBI stands for chemical entities of biological interest. It's not clear why the top level contains groupings like p-block which AFAICT have of no interest to biologists

This leads to massively unintuitive classifications, for example this is the classification of the nsp3 sars-cov-2 viral protein in PRO:

I propose

1. obsolete p-block entity and similar classes
2. if this information is useful to someone, replace with a non is-a relationship. For example:
   • include a small branch of the ontology for things like orbitals
   • all existing children of p-block entity would have a relation like has-electron-in some p-orbital
   • this way 99.9% of us could only use is-a, has-conjugate-base etc and ignore this part of the ontology, but those who still need it can easily find it

[K-r-ll]

I believe the information of an entity being in a particular group of periodic table is useful, e.g. "halogen molecular entity", as these entities share many properties. Also a higher level distinction "main group molecular entity" vs "transition element molecular entity" is useful. So what's wrong with blocks? It's just an extra level between particular group and main group / transition element.

On the other hand "has-electron-in some p-orbital" is not particularly useful info as most chemical elements have electron(s) in p-orbitals. Just saying.

Why the sweeping opening statement? ChEBI is used not only by biologists. And even biologists know a bit of basic chemistry.

[amalik01]

Hi Chris,

I don't think this would be feasible and our user community is diverse (consisting of both biologists and chemists). Some of our users are interested in the type of periodic block a particular element belongs to. However, you can develop your own slim version of the ChEBI ontology using a tool developed by OLS (https://github.com/zoependlington/chebi_slim/tree/slim). This tool will allow you to keep or remove terms from the hierarchy.

I hope this helps.

[cmungall]

@amalik01 thanks for the suggestion. We use robot for making the slims. In this case we'd probably make an anti-slim since we want to retain 99.99% of the terms, it's just the upper level that is an issue. I think such a slim would be useful to the majority of users, maybe we can discuss making it available via an official chebi OBO overlay.

Note that my proposal still retains the information, users would just navigate over a different relationship type, no information would be lost. This is a very standard design pattern in ontologies to avoid is-a overloading.

Also note that I am not proposing obsoletion of p-block element atom, just p-block molecular entity.

However, you know your users best. If you think this would be too disruptive for them this is your choice. I might suggest though that simplicity and clarity for the core user base of CHEBI might take priority.

Can we all agree that classifying a specific protein such as SARS-CoV-2 as a "p-block molecular entity" (see screenshot) is not useful?