Wikidata:Property proposal/species formed by

species is product of edit

Originally proposed at Wikidata:Property proposal/Natural science

Withdrawn

Description	product of some species by some chemical reaction as qualified by a fabrication method (P2079)
Represents	product (Q542929)
Data type	Item
Domain	Q11173
Example 1	preproinsulin (Q7240673) → preproinsulin (Q60495072), qualifier fabrication method (P2079) enzyme activity (Q22035510)
Example 2	starch (Q41534) → maltose (Q170002), qualifier fabrication method (P2079) enzyme activity (Q22035510)
Example 3	amide ion (Q4026895) → ammonia (Q4087), qualifier fabrication method (P2079) protonation (Q899488) (example from conjugate acid (P4147))
Example 4	protein (Q8054) → glycoprotein (Q187126), qualifier fabrication method (P2079) glycosylation (Q898365)
Example 5	ethanol (Q153) → acetic acid (Q47512), qualifier fabrication method (P2079) fermentation (Q41760) (see vinegar (Q41354))
Expected completeness	always incomplete (Q21873886)
See also	conjugate base (P4149) and conjugate acid (P4147) as subproperties. Also substrate of (P2414).

Motivation edit

Chemical reactions are abundant and we already have properties to handle some of them, such as conjugate base (P4149). However, there is not a more general property, which would be useful in many occassions. One specific one is peptide hormones that result from proteins, which are encoded by genes, such as human insulin, which is a chemical product (after splicing) of the protein preproinsulin (Q39798). I want to stress that this proposal is not meant to enable storage of chemical reactions generally, but meant to be limited to link notable chemical entities with a notable reaction.

The immediate use case is to properly encode (human) hormones in Wikidata, such as peptide hormones and have entries in different databases than the proteins the are formed from. See also Modelling a chemical reaction?

Finally, I can imagine that qualifiers may be used for further detail. Egon Willighagen (talk) 12:17, 8 January 2019 (UTC)[reply]

Discussion edit

Notified participants of WikiProject Chemistry

Oppose Use has part(s) (P527) David (talk) 12:56, 9 January 2019 (UTC)[reply]
- Bringing up has part(s) (P527) as alternative suggests my explanation is not clear, because it is totally unrelated. Can you please explain what you mean? --Egon Willighagen (talk) 08:36, 10 January 2019 (UTC)[reply]
Comment I don't think has part(s) (P527) is relevant. However, I don't understand the proposed property label - it should make sense as a predicate label in the sense of "subject predicate object", but is "ethanol species formed by acetic acid" really your intended meaning here? Also "species" could be confusing to the biological taxonomy community; I think you need a different label, but not sure what it should be. ArthurPSmith (talk) 19:03, 9 January 2019 (UTC)[reply]
- Point taken. I was struggling with that too, and followed the description of conjugate acid (P4147), but not well enough. I'll update. --Egon Willighagen (talk) 08:42, 10 January 2019 (UTC)[reply]
Oppose This property doesn't provide enough guidelines to avoid any possible relations between chemicals. Just an example: if I take n-heptane, I can say that all hydrocarbon smallest that heptane can a species formed by heptane (through cracking). A certain number of qualifiers should be added to provide more information about the reaction itself and perhaps about the conditions too. Instead of a simple link we need a model describing the reaction. Snipre (talk) 22:40, 9 January 2019 (UTC)[reply]
- OK, so we should basically give guidance on the qualifiers to be used, e.g. the process to causing the conversion: propane formed by (cracking (Q212749)) from n-heptane? The fermentation example can be modelled like that too. --Egon Willighagen (talk) 08:39, 10 January 2019 (UTC)[reply]
- Oh, regarding the idea of generally encoding chemical reactions, with pressure, temperature, etc. I did consider that and not against it, but that sounds very unbounded. But cool. I shall add an example this afternoon or tomorrow. --Egon Willighagen (talk) 09:51, 10 January 2019 (UTC)[reply]
@Snipre, ArthurPSmith: please see my replies and the revised proposal. --Egon Willighagen (talk) 08:59, 10 January 2019 (UTC)[reply]
I think there should be a different way to model chemical reactions in WD, not necessarily by adding properties to items describing chemical compounds, but by separate item for every chemical reaction (with properties for substrates, products, catalysts, intermediates, possible conditions etc. etc., but without putting everything into chemical compound items. For linking notable chemical entities with a notable reaction maybe it would be better to create something like may take part in a chemical reaction with reaction XYZ item as a value and qualifier as with a value being e.g. catalyst, substrate etc. But I'm not sure at this moment if this approach would be suitable, it's just a loose idea, because it could lead to thousands of statements in one item... Wostr (talk) 14:42, 10 January 2019 (UTC)[reply]
- Not at all against this, but I am not sure about how to name the reactions. "Splicing of human proinsuling into insulin", etc? And we first need properties then to indicate what substrate/reactants are and what are products, right? --Egon Willighagen (talk) 16:32, 10 January 2019 (UTC)[reply]
  - Reactions in WD would be in fact classes IMHO, as even if we specify all the conditions for specific reactions, it's still a type of reaction that can occur in a specific position and time (that would be an instance, but we wouldn't have that kind of items in WD). There is already RInChI for reactions (in fact, there's RInChI, three versions of RInChIKeys and RAuxInfo), but I don't know if it is used in any database right now. However, there are databases with reactions (e.g. Reaxys have reactions and each reaction has its own ID). So (1) we should first find out what kind if IDs are there for reactions. (2) I think we don't need to add any statement in chemical compound item for reaction, properties for reactions may not be inversed (i.e. there would be links to chemical compound from chemical reaction, but not from chemical compound to reaction) — it wouldn't be visible in Wikidata's UI, but it would be possible to query everything using SPARQL. (3) model: we should check haw it's done in other databases. I can imagine a basic set of properties for chemical reactions at this moment, but basing on something that exists can be more useful than inventing everything from scratch. IMHO naming is the smallest problem here. Wostr (talk) 19:19, 10 January 2019 (UTC)[reply]
@Wostr, Egon Willighagen, Snipre: I'm still not happy with the label. Also I think we could explore Wostr's idea to put these relationship statements on the reaction - do we have items for specific reactions now? I note that the fabrication method (P2079) qualifiers that Egon has proposed up above are somewhat generic - even fermentation (Q41760) involves multiple reactions with several different components. ArthurPSmith (talk) 15:22, 10 January 2019 (UTC)[reply]
- I am not aware we have specific reactions yet, but Reactome and WikiPathways would be a (CCZero) resource that could provide many. Also, alternative names are most welcome, of course! --Egon Willighagen (talk) 16:34, 10 January 2019 (UTC)[reply]
Better stop the process of property creation and start the dicscussion within the WikiProject Chemistry about what we want to model concerning chemical reactions. Snipre (talk) 21:42, 10 January 2019 (UTC)[reply]
- Yes, sounds good. The discussion already existed and can be found at Modelling a chemical reaction?. --Egon Willighagen (talk) 09:06, 11 January 2019 (UTC)[reply]