Template talk:Drug links

Working here

I am working to improve this template. -DePiep (talk) 23:26, 30 January 2021 (UTC)

@Ajpolino: -DePiep (talk) 23:29, 30 January 2021 (UTC)

Excellent! I've been offline for a couple of days so I'm just taking a look through the WT:MED thread now. As you can see, most of the template fields now are just leftover from {{Medical resources}} as I quickly realized I was past the limits of my technical understanding (sadly, a low bar for me). I was imagining at the very least moving all the fields that show up under "Identifiers" in {{Drugbox}} and consist of an external link to some database to this box? But we'll see what others think. Thanks a million for putting some time into this! I hear you about the limited hours in the day. They just slip by. Ajpolino (talk) 23:42, 30 January 2021 (UTC)

We'll do fine. I'm building this base table to check the data, more easiliy. Good OP. -DePiep (talk) 23:48, 30 January 2021 (UTC)

Thoughts

To consider:

• Use WD data wherevert possible. Maybe local overwrite possible. [todo: example here].

• Allow 2, 3, n chemical compounds, somehow.
• Allow 2, 3, n compound drugs, somehow.
• Think hard why Chem and Drug composites are same/different?

• use same data-point handling: parametername, el, formatting, ... for example |CASnumber=

-DePiep (talk) 23:53, 30 January 2021 (UTC)

More design concepts

Let us keep good design concepts in mind.

What "Drug links" is not

−1. Not a Auth Control {{Authority control}}, not just referencing external IDs

−2. Not a box of repositioned infobox parts

−3. Not an infobox

What "Drug links" is (I hope)

+1. Part of article body = content, and in the TOC

+2. Legible, readable

+3. A data sheet, basically

+4. Using page width and space

What it has

Indexes, that is: repeated 'Drug list' data for multiple, separate items. Like:

Indexes for multiple drug compounds (eg Efavirenz/lamivudine/tenofovir)

Indexes for multiple chemical compounds (eg Propylhexedrine, CAS)

Allowing multiple input (eg ATC's)

Thinking wide usage

Usable on {{Chembox}} articles. {{Chembox}} sometimes handles drugs, as in: drug properties of a chemical as a drug ([1])

Thinking Wikidata

WD-1: read Wikidata automatically, allow setting qid

WD-2: allow local (=enwiki) overwrite by parameter (eg, current ECHA InfoCard ID)

This setup is well-provided by templates etc in enwiki

Do not

Do not use CheMoBot verification & validation handling N.

Do not create an {{AuthContol}}-like code only box N, keep it a readable section, in TOC

So, change thinking that ...

...it is some 'infobox things, replaced'

...it is an EL linkfarm

...info cannot be repeated in the Infobox (it can, is what Infoboxes are about :-) )

-DePiep (talk) 19:58, 12 February 2021 (UTC)

Multiples: compounds, combination, stereo

Drugs can have multiple chemical componds (eg ivermectin), and/or can be a combination drug (eg Artemether/lumefantrine)

Also, to keep in mind ({{Chembox}}-related): a single article can contain multiple compounds (eg by Stereochemistry)

• Three compounds = three {{Drug links}} boxes? -DePiep (talk) 23:47, 13 February 2021 (UTC)

  Hm, this is an interesting point I hadn't considered. The drug page that got me started on all this is ivermectin, a mixture of two compounds that aren't sold separately (contrast with, say, Artemether/lumefantrine or some other combo drugs). For some of the links (i.e. anything related to pharmaceutical products) it would just have one link. For anything related to chemical properties, it'd probably have two? I'd love for it to have just one Drug links box. So I suppose that would necessitate many fields accepting several inputs, as you suggested above. That said, I'm not sure what our technical limitations are, so you'll have to let me know if we need to be thinking about this differently. Ajpolino (talk) 01:25, 14 February 2021 (UTC)

  @Ajpolino: technically, under the hood, templates and especially modules can handle a lot. I don't think current drugbox-articles would cross a technical limit (= when article would not render completely, too many template tasks etc). We can design as it there is no tech limit.

To think about is: how do we present such multiple information? Simple drug = simlpe list-like, OK.

Then a drug can have chemical compounds end drug combination parts. I understand these are not the same - full stop. This can be shown: each drug (in a combination) gets its own main section (visual box). Within every single-drug section, properties of chemical compounds are listed in a regular form (pattern or tabular-like), so readers can oversee the list easiliy & compare values.

Of course, overall properties (for the drug) are in to listed once.

Not in this: a single drug can have multiple ATC-values. Later. -DePiep (talk) 17:10, 14 February 2021 (UTC)

I can get behind that. We may end up with at least two sections to the box then. Section one is links where the destination site describes the marketed pharmaceutical product (e.g. Drugs.com link for artemether/lumefantrine). Section two is where the destination site describes chemicals (e.g. PubChem has separate links for Artemether and Lumefantrine)... actually as I'm typing this I took a look and found out that PubChem does indeed have a database page for the combination artemether/lumefantrine! Perhaps if many other databases also contain combination pages, this whole splitting point will become moot. Ajpolino (talk) 17:21, 14 February 2021 (UTC)

Not yet thought of this: destination site setup (EL). So over there the splitup I decribe (drug → sub-combinator, → sub-compound: hierachic only) could come together for these items. Will think about this, but must prevent headache ;-) -DePiep (talk) 17:28, 14 February 2021 (UTC)

Yeah poking around these sites, it's a mixed bag. The next two I checked (Commonchemistry.org and IUPHAR/BPS) don't host a separate page for artemether/lumefantrine. Though the good news is that all three sites so far have a single page for ivermectin. Perhaps today I'll get a moment to make a table for all of the current links so we can start to sort out how we'd make this work under-the-hood. Ajpolino (talk) 17:34, 14 February 2021 (UTC)

Sounds good. After all, in top of the new box there could/should be "all of this drug" links; so far for our presentation. Could require manual (=enwiki editor manual research like you did in this). We could use a mockup by now. -DePiep (talk) 21:09, 14 February 2021 (UTC)

What links would be useful for a Drug links bar?

Hi @WhatamIdoing, Bondegezou, Boghog, Colin, Seppi333, Tom (LT), RexxS, and Whywhenwhohow:, although the WT:MED discussion got caught on some tangents, I think the general consensus is in support of spinning off some links from {{Infobox drug}} to a new template bar in the External links section (compare to {{Medical resources}}). I'm hoping we can discuss here which links should be included in this new template. As candidates, I'll list the external links currently in the infobox:

• Under "Clinical"
  • Drugs.com link(s) that go to an information sheet for the drug (e.g. aspirin).
  • MedlinePlus links to a similar info sheet hosted by the US Nat'l Library of Medicine (aspirin again)
  • Licensing links. Currently broken but my understanding of that thread is that given a drug name we can fix the European Medicines Agency link (example, idk what the old links looked like, but at a glance this page is mostly a lay-friendly drug description rather than something licensing-specific).
  • ATC codes, listed both a wikilink for the code lists that apply, and ELs to the WHO entry for that code position (example).
• Under "Identifiers"
  • CAS number - gives both the number and a link to Commonchemistry.org that gives synonyms and an image of the chemical structure.
  • PubChem - shows the Pubchem ID number and links to the Pubchem page for that drug.
  • IUPHAR/BPS - shows the Guide to Pharmacology number and links to their website.
  • DrugBank - same.
  • ChemSpider - same.
  • Unique Ingredient Identifier - links to the FDA Substance Registration System, which doesn't seem to have that much info on it (e.g. [2]).
  • KEGG - shows the KEGG number and links to the drug's KEGG page.
  • ChEBI - same.
  • ChEMBL - same.
  • PDB - Aspirin has two links here, one appears to search PDB for all structures that contain the drug? The other links to the chemical structure.
  • CompTox Dashboard - same. Shows the database number; links to their site.
  • ECHA Infocard - same.
• Under "Chemical and physical data"
  • 3D model - Links to a St. Olaf College-hosted site (I think) which gives an interactable 3D model (example).

So my questions for you all: (1) Are there any links that we should no longer include? Any we should add? (2) {{Medical resources}} separates links into two sections "Classification" and "External resources", are there categories we can separate these links into? Clicking though these, many of the database links seem to offer more-or-less the same information. Should we still list them all? Are there some that are inferior and should be trimmed to better guide readers to the superior links? I know this is a lot, so I'll just note that we're not in a huge rush here, and no one is planning to secretly remove links to your favorite database. Also we can always go back and add/move/change links, so let's not let seeking perfection prevent doing good. Anyway, pardon the large post. Happy to hear folks' thoughts. Ajpolino (talk) 02:36, 12 February 2021 (UTC)

At the risk of sending everyone screaming from the room, what do you think about including https://mshpriceguide.org/en/search-results-by-name-2/?searchYear=2015&searchString=Acetylsalicylic+Acid&searchType=Name ? It qualifies under WP:ELYES #3. WhatamIdoing (talk) 03:24, 12 February 2021 (UTC)

Looks good to me! We can sort out how to "categorize" it with the other links as we start to settle what else will be on the template. Ajpolino (talk) 03:51, 12 February 2021 (UTC)

Most of those identifiers look great as external links, but I'd take them out of an infobox. Bondegezou (talk) 09:44, 12 February 2021 (UTC)

I don't think we should link to it, for several reasons.

• When I follow the link or go to any page on the site, Kaspersky pops up to warn that it has blocked malicious javascript from iclickcdn[.]com. So it looks whatever platform they are using to construct the site has been compromised. (WP:ELNO #3) I emailed them months ago about a similar issue, and they took the site down for a while, but it is back up and seems no better.
• The link above is to search results, which fails WP:ELNO #9. I'm also not convinced there is reliably a 1:1 mapping between article subject and search text (e.g. "valproic acid" and "sodium valproate" are two forms of essentially the same medicine but are held separately in their database).
• Their dataset hasn't been updated since 2015, which makes the data about 6 years old, and that is ancient history really. A key user of this database was the WHO/Health Action International Project on Medicine Prices and Availability, which is now historical.
• As we discovered from the discussions about prices, correctly interpreting the often very limited data on individual drug products is very hard and mistakes were more often made than not. My feeling is that this site could be used in an difficult exam about statistics to see who is actually capable of reliably drawing conclusions about the data: it is filled with gotchas.

-- Colin°^Talk 10:38, 12 February 2021 (UTC)

• My thoughts for the list:

INN: Add formal "INN" explicitly (now implicitly through title, with exceptions like heroin)

InChI, InChIKey, SMILES: under "Chemical and physical data", add InChI, InChIKey and SMILES. These are not exactly having the external links issue at hand, but this info is technical only (structured illegible ascii text), and so doe not even belong in the infobox. Meanwhile, while in the "Drug info" data sheet, as true identifiers, they are visible in front for searching machines, helpfully.

3D model: Also, by adding SMILES, the "3D model" can be derived from this input. Example: Aspirin infobox has SMILES data (hidden); the infobox has the "3D model (JSmol)" EL using that same SMILES input.

-DePiep (talk) 19:15, 12 February 2021 (UTC)

The WHO link for the ATC code seems more like a reference than something that qualifies as an external link. For the identifiers, is there some way we can determine if these database sites are well used and relevant in 2021 or just hangovers from something interesting someone found in the early days of the web. -- Colin°^Talk 10:38, 12 February 2021 (UTC)

Thank for the ping. I really feel that, consistent with most areas of the encyclopedia, the better place for identifiers is in the {{Authority control}} template - particularly the most fixed and formal ones (including ATC). Excluding those ones, because in your proposed template the links will not be to conspicuous, I think we should just include all the ones in the existing template, and any others that are felt to be useful - I will support any that others propose as finding useful themselves where the sites satisfy our general wiki principles such as likely to be verifiable, not paywalled etc. Over time the links can be whittled as required. Cheers --Tom (LT) (talk) 08:09, 15 February 2021 (UTC)

Some ID's may be acceptable in the AC template. However, reading WP:Authority control I think not all are suitable for this. I think AC is not suitable for ATC, a classification only and not per sé singular for a drug. Note that AC values & links are outside of body content (do not even show in mobile view). OTOH, with info like ATC, FDA-data, InChi definition, c hemical names and INN: all relevant in teh article. So while some of these should be removed from the infobox, they should be present in the article body. (Some could be repeated in the infobox, like INN, ATC).

In chemistry there is a thing called "data sheet", a list of all properties of a compound. For every data point, we should consider it they should be in the "data sheet" list such as a template box, in a dedicated article (sub)-section.

IOW, for each data point listed: why should be moved to AC?, to the datasheet?, and should it be repeated in the WP:infobox? -DePiep (talk) 15:13, 15 February 2021 (UTC)

My basic opinion is that the infobox should contain only essential information, and should ideally contain information that our readers, who are normal people and not for the most part pharmacists, can read and understand. Whatever else should be moved out to the AC, datasheet, external resources box etc. IF there is duplication then that defeats the point of the box. Whatever is suitable to move to the AC could be moved there, otherwise should be moved to the datasheet / external resources box. --Tom (LT) (talk) 22:38, 20 February 2021 (UTC)

Unbelievably after a quiet week, Tom and I came back to this at the same time. Edit conflict, but here's some follow-up info for each link, sandbox style. Trying to sort through what is what:

• Drugs.com - patient-targeted info sheet on each pharmaceutical product. Claims to have occasional expert review. At a glance it looks like drugs are listed by tradename.
• MedlinePlus - looks like the same target audience as the Drugs.com sheets. Drugs listed by generic name.
• EMA - Again, same product. I can't find Coartem (artemether/lumefantrine) on their site, so may be a bit more limited in scope? Drugs listed by tradename.
• ATC codes - the external links are to unhelpful WHO pages. Whether we keep the codes or not, the links should probably go.
• Commonchemistry.org - gives just the CAS number, list of names, and structure of each chemical (e.g. aspirin). We probably don't need to keep these links, though we may wish to state the CAS number somewhere regardless.
• Pubchem - A full chemical info sheet with links to other databases, clinical trials, et al. Has a separate page for each chemical and for combination drugs (e.g. Coartem).
• Guide to Pharmacology - a couple of text blurbs and a smattering of assay data for each chemical (i.e. combo drug would have two separate pages). This link seems less useful to me.
• DrugBank - similar to Pubchem but with more text, and no page for combination drugs. Our DrugBank page says some of its text comes from Wikipedia, so I'm assuming it's not curated as actively as some of the other resources.
• ChemSpider - a stripped back version of what's available on PubChem: names, a few chemical properties, and links to papers that mention the chemical. Does have pages for combo drugs, but they're not very useful (e.g. Coartem) so probably individual links for each member of a combo would be preferable.
• FDA Substance Registration system - has a page for every drug, but not much info as I said above. I think we could scrap this one, but we'd probably still want to display the UNII somewhere.
• KEGG - datasheet that has a bit of chemical info, a bit of functional info, some classification info, and links to other databases. One page for each combo drug.
• ChEBI - datasheet with a bit of info and the lead from the Wikipedia article. One page for each chemical (multiple links for a combo drug).
• ChEMBL - this one is quite different. Quite a bit of biology-related info and a really neat graph of papers over time color-coded by journal. Individual page for each combo drug. Color me impressed.
• PDB - we have two links. The first is a search result for all PDB structures that include your chemical (I think we should scrap this one), the second is a datasheet for each chemical, mostly with the same old stuff. One page per chemical (multiple links for a combo drug).
• CompTox dashboard - datasheet that's somewhat more toxicology focused. One page for each combo drug.
• ECHA infocard - a slimmer datasheet with chemical info. ECA's answer to PubChem? Can't find some drugs, so either idk how to find things on that site, or it's a bit more limited in scope.
• 3D model - basically what I said above. Looks like you feed it a SMILES name and it spits out a 3D model. Neat, but the PDB link also provides a 3D model.

Work in progress. Ajpolino (talk) 22:51, 20 February 2021 (UTC)

Proposal

Alright folks after spending some time clicking around today (see above), here's my proposal:

• Stay in infobox drug - ATC codes (but lose the unhelpful external links), IUPAC name.
• Move to Template:Drug links - Two categories:

1. Pharmaceutical resources - MedlinePlus, EMA link, licensing links (if available, see WT:MED discussion), MSH Price Guide (if available)
2. Biology/chemistry resources - PubChem, ECHA infocard, KEGG link, ChEMBL, PDB link, CompTox link. The others seem less useful to me, but I'm happy to be talked into them.

• True identifiers moved to authority control, chembox, or somewhere - UNII, CAS number (without the commonchemistry.org link), InChI, and SMILES. We'll sort this out later. Here, since we're at Template talk:Drug links, let's just sort out what we're going to put in {{Drug links}}

Note that in this proposal, Template:Drug links would contain only external links to useful resources. So it would be a bar in the External links section, analogous to {{Medical resources}}. Also just to be clear this means we would lose the Drugs.com link (redundant to the Medline and EMA links); Guide to Pharmacology, DrugBank, ChemSpider, FDA Substance Registration System, ChEBI (all redundant to PubChem and ECHA); PDB search results (but not the database page), and St. Olaf 3D model (redundant to PDB page). I've got no personal beef with any of the above. If anyone feels strongly about them, happy to re-add. I just personally found them to be redundant to the existing resources. Happy to hear folks thoughts! Ajpolino (talk) 06:25, 21 February 2021 (UTC) Setup a new subheading to make clearer what's being proposed. Pardon the mess. Ajpolino (talk) 20:04, 27 February 2021 (UTC)

Ajpolino, I didn't see any consensus for "MSH Price Guide (if available)". Even if it is updated beyond 2015, and even if they fix their security issues, the data remains statistically difficult and weak, but fundamentally, they are not a resource on drugs but a database of product prices. It is like adding a search for "cookies" on a supermarket website to Cookie. That isn't the kind of external link that is permitted. -- Colin°^Talk 08:52, 21 February 2021 (UTC)

Oh, apologies, I somehow missed your reply above about the MSH Price Guide. Striking it from my proposal for now, and we can further discuss below or in the future. I'll take a look later this week and share my thoughts, but I don't want to hang up everything else based on that link. Ajpolino (talk) 18:51, 21 February 2021 (UTC)

Drugs are products, and it being statistically difficult isn't a bar. But let's talk about it later. It's easy to add or remove one or two things, once we've got the overall structure in place. WhatamIdoing (talk) 01:12, 27 February 2021 (UTC)

"Drugs are products" is using both "are" and "drugs" ambiguously and in shorthand. A packet of 60 10mg generic tablets of wonderpam distributed by Drugs R Us International Aid and sold to the health ministry of ForeignLand in 2015 is not the subject of the article on wonderpam, indeed there are very few drug articles where such an end-product would even be mentioned. The same is not true of most (all?) of the other drug links we are discussing, where wonderpam is their page subject too.

Statistically difficult is a bar if we claim to have a generally usable link to an international reference price for drug X. We have a search-results link to a database, where readers can 'have fun' clicking on all the links and being disappointed. We don't have a way to educate the reader en-route to the results page, to say that a reference price requires a good number of supplier prices, for which you should use the median, to ignore buyer prices, and to apologise for wasting their time. There is a reason why WHO/HAI restricted themselves to a few dozen products (drug+dose+formulation+quantity) when doing their price comparisons among nations: those were the small number of records for which sufficient supplier prices were available to create a reference. It is like we are pretending to have an EU price for wonderpam, but in fact only having a six-year-old price for three dose/formulations from only two of the six pharmacy chains in Poland, one of which has since gone into administration and the other taken over by a multinational. -- Colin°^Talk 19:30, 27 February 2021 (UTC)

@WhatamIdoing, Bondegezou, Boghog, Seppi333, Tom (LT), RexxS, and Whywhenwhohow: Hi all, I'm sorry for the second ping, but if folks could take a quick look at this it'd be much appreciated. I've gone through each of the current links a few posts above and given a brief summary. Here's a proposal for what to put in the Drug links bar to start. We can always add more resources as we find them, but we've got to start somewhere. I'd feel better if we could get a couple more sets of eyes on this so that DePiep and I don't do anything idiotic while no one is watching. Thanks for your time! Ajpolino (talk) 16:28, 25 February 2021 (UTC)

• There seems to be a lot of opinions on which information is not useful without any justification for why they are not useful. Redundancy is one of the few arguments I would find persuasive. Just to be clear, InChI, and SMILES are not names, but character representations of the structure. InChI is not human readable, but SMILES, at least the shorter ones, can be interpreted by organic chemists. The InChIKey is highly useful since it allows Google searches of the entire web (including Wikipedia) for chemical structures. Likewise, InChI and SMILES are highly useful, since many programs such as chemical drawing software can convert these into 2D and 3D representations of the structure. Concerning the PDB links, I find the link to structures that contain the drug very useful (I am biased, I do structure based drug design for a living). These PDB structures indicate how the drug is interacting with it biological target or perhaps enzymes that metabolize the drug. If you are interested in knowing how things work, this is a useful link. The other PDB link is probably redundant to many of the other links. Boghog (talk) 16:58, 25 February 2021 (UTC)

  I'm a bit confused as to what you're responding to in places, so forgive me if we're talking past each other. Regarding "usefulness": redundancy is the criterion I used to select the links I propose above. For groups of redundant-looking links, I've picked representatives that I thought covered all the information and were from particularly trusted publishers. If anyone thinks that some of the links I labelled as "redundant" are in fact not redundant, I would very much like to hear that. We're not particularly space-constrained here, so we can keep more links or add new links.

  I'm not proposing removing InChI or SMILES from the page. I'm suggesting that since they're widely used identifiers they should go somewhere on the page besides {{Drug links}} (which should contain external links) and {{Infobox drug}} (which should summarize key parts of the article). I suggested {{Chembox}} or {{Authority control}}, but I don't have strong opinions either way.

  Regarding the two PDB links, you can get to the search results by clicking the "Search" button just under the chemical name at the second link (aspirin example) and as Colin mentioned above we're discouraged from providing ELs that are search results pages, so my preference would be that we pick the second PDB link, rather than the search results one. That would put you a click away from the search result, while also providing the uneducated clicker with a bit more context. Ajpolino (talk) 02:31, 26 February 2021 (UTC)

• Sorry, I was not only replying to you, but the entire discussion. I was a little frustrated with suggestions through the wider discussion for removing certain links with no explanation why. If the PDB ligand link in turn contains a link to the crystallographic structures that contain the ligand, then I am fine with it. Thanks for the clarification. So basically there is no suggestions to removing any links (with the exception of one of the two PDB links), just redistribute them. Correct? I still generally opposed to spliting up the data into different templates. If this is scattered throughout the article (infobox, cheminfobox, authority control) it makes it more difficult for readers to find this information. Why not keep all the data in one place as it is now? Finally, the InChI and SMILES are representations of the structure and hence are essential parts of the article. Without structure, there is no drug. Boghog (talk) 11:15, 26 February 2021 (UTC)

  I think that splitting it up will make all of it easier to locate, once people have made the transition. Finding 'my' thing in 'my' short box, at least when 'my' short box is placed in a consistent location, is faster than finding 'my' thing in a single long box. WhatamIdoing (talk) 01:19, 27 February 2021 (UTC)

  The current {{infobox drug}} is clearly split into logical sections "Clinical data", "Legal status", "Pharmacokinetic data", etc. so that one can rapidly find the data one is interested in, even if the infobox gets long. Furthermore the infobox is almost always placed at the upper righthand side of the article and makes efficient use of space since it displays along side the table of contents, space that is otherwise wasted. As I understand it, the current proposal is to split the box into three parts. A condensed infobox, a new box in the chemistry section, and somewhere near the bottom in an authority control box. Many drug articles do not even have a chemistry section and if they do contain one, their positioning is often inconsistent. The proposal, if implemented, will decrease consistency, not increase it, and not just for me, but for everyone. Boghog (talk) 07:32, 27 February 2021 (UTC)

  I recall from your comments here that you're generally opposed to splitting up the infobox, but as I said at the top of this section, I think the general consensus of that discussion was in favor of peeling off some of the external links into a separate box. The purpose of this discussion is to answer the question: "What should we include in the external links bar?", not to relitigate that discussion (you're welcome to relitigate that discussion if you wish, but perhaps we could do so in a different section). Despite your preference to not be doing this splitting, I've pinged you here (and Seppi, the other editor who expressed a preference for keeping the infobox together) hoping to benefit from more eyes, perspectives, and experience as we sort this out.

  I think the layout of this page has become a bit confusing, so I've made a new subheading "Proposal" above. To your there is no suggestions to removing any links.... Correct? after going through the current external links, I am proposing to remove some that I thought were redundant to existing links (see the first post under the "Proposal" subheading and the last post before the "Proposal" subheading). If you think some of the links I'm proposing we remove are not redundant and ought not be removed, I'd very much like to hear that. Similarly, if there's any resources that come to mind that you think we should include, we can probably do so. We do not have to remove any links at all, I just figured as long as we're looking at this, we could sort out which ones are redundant. Fewer redundant links means readers get to new information more quickly. I'd appreciate any comments. Ajpolino (talk) 20:03, 27 February 2021 (UTC)

• Happy with the proposal... in fact literally any proposal which moves stacks of links out of the infobox. I am happy with redundancy, and in my opinion the most important element is the reliability of the source we're linking to (rather than selecting the best of several). In my opinion, simply starting by moving all the existing links out into the new template is the best starting point with the most consensus. I don't feel it's likely to be particularly constructive to have a very lengthy discussion in advanced about its contents as here we have always had an agile, iterative approach to doing things and I don't sense there is that much disagreement about any specific items that would block the whole move. Specific discussion about entries can then occur, including about authority control etc. Anyhow that's my two cents. --Tom (LT) (talk) 03:30, 28 February 2021 (UTC)
• I will support the proposal if we can remove the proposed further fragmentation of data into Authority control (with a few exceptions for common drugs) and the ChemDrugBox (for reasons that I have stated elsewhere is impractical). I think most of the current links in the infobox are appropriate, hence I think it is more efficient to focus on what should not be included supported by strong arguments, keeping in mind what is not useful to you may be useful to others. As I have stated before, I think it is important that we first have a fully populated mock up in place to see what the proposed split looks like and then decide what if anything to remove. Boghog (talk) 18:12, 1 March 2021 (UTC)

Moving some to authority control

Hi DePiep and others. Having browsed through the above, could we consider moving these to the authority control template?

• • Unique Ingredient Identifier - links to the FDA Substance Registration System, which doesn't seem to have that much info on it (e.g. [3]).
  • KEGG - shows the KEGG number and links to the drug's KEGG page.
  • ChEBI - same.
  • ChEMBL - same.
  • PDB - Aspirin has two links here, one appears to search PDB for all structures that contain the drug? The other links to the chemical structure.
  • CompTox Dashboard - same. Shows the database number; links to their site.
  • ECHA Infocard - same.

These seem to be official databases that could conceivably be moved. If we agree on that, then there are less things to put in your resource box.--Tom (LT) (talk) 22:42, 20 February 2021 (UTC)

Honestly, I'm still not sure I understand the purpose of authority control. Is the template just intended to hold widely used identifier codes for the page it's on? If so, you can see from these databases which identifiers they consider worthy of listing. DrugBank lists just the Unique Ingredient Identifier, CAS number, International Chemical Identifier (InChI), IUPAC name, and SMILES name. The first two are short(ish) alphanumeric codes. The latter three can be a bit long. Those other sites have their own internal IDs, but to my (limited) knowledge their internal IDs aren't widely used. Ajpolino (talk) 23:02, 20 February 2021 (UTC)

As an extra data nugget, I'll note that under the "Identifiers" section of its datapages, PubChem lists IUPAC name, InChI, SMILES, CAS, and "DSSTox Substance ID" which looks to be the same as the CompTox Dashboard link we currently have. Ajpolino (talk) 23:07, 20 February 2021 (UTC)

@Tom (LT):

Yes these should be removed from {{Drugbox}} tomorrowtoday. Per WP:INFOBOX.

No they should not go to AC. Per WP:Authority Control. AC is not to provide EL's to databases with new/extra info. Not. Mainly, AC is to to distinguish names, aka WP:DISAMBIGUATE. AC is not to serve info WP:External Links. (but elsewhere we can ;-) see below)

Let's do create section ~"==Data sheet==" that has this info, including the EL's.

-DePiep (talk) 23:59, 20 February 2021 (UTC)

I disagree with your interpretation of authority control (WP:VIAF). The first line says that it is important to help disambiguate but if you continue reading it clearly indicates that it can be used to link unique identifiers in other circumstances: The template links Wikipedia articles (and user pages) to the corresponding entries in library catalogs of national libraries and other authority files all over the world. The entries typically correspond to people, book titles, and similar well-defined entities and in "Purpose" Authority files, however, are not restricted to information about people. Some files contain data about organizations, geographic names, work titles, or general subject terms or keywords... By adding authority data to Wikipedia articles, software developers from both the Wikimedia and library communities will be able to create new tools to link and combine information from Wikipedia and databases and online catalogs that use these identifiers. Happy to wait for other editors to comment on this.

With respect to the data sheet, I'm not sure why this can't just be a little box like {{Medical resources}} at the end of the article, and I also think that the title is unclear (I wouldn't know what to expect if I was reading and saw a section titled "Data sheet" - perhaps "Additional resources" might be more readily understood). Overall I support the general idea to move things out of the infobox and to a place that's less prominent.--Tom (LT) (talk) 00:12, 21 February 2021 (UTC)

It's not just about WP wordings. AC, INFOBOX, EL, and yes DATASHEET too have their own information concept. For example, AC does not show in mobile phones, for a reason. -DePiep (talk) 00:24, 21 February 2021 (UTC)

AIUI the main reason that AC doesn't show on the mobile site (regardless of your device) is because it's built on navbox, which is a wide table, and wide tables display poorly. (Also, we have no good evidence that readers use navboxes anyway – nobody ever complains about their absence.) WhatamIdoing (talk) 01:21, 27 February 2021 (UTC)

IMO, first candidate for AC WP:VIAF is CAS number. Truly an identifier, assigned by an authority (CAS). -DePiep (talk) 18:01, 21 February 2021 (UTC)

In searching the Integrated Authority File, only the most common medications like aspirin are listed. The vast majority of medications are not. There does not seem to be an equivalent authority control system for chemicals or drugs. Strictly speaking, CAS numbers are proprietary to the Chemical Abstract Service and one needs a subscription to access their database to determine a CAS number. An authority control system by definition, must be completely open, and the Chemical Abstract Service which creates CAS numbers is not. Granted, CAS numbers have become the de facto standard for identification of chemicals by chemical vendors and CAS has loosened up access over time, but there are still restrictions on their use. A prerequisite for using CAS numbers as a Authority control is a working external link however the vast majority of CAS links from Wikipedia drug articles to www.commonchemistry.org return no hits.

I don't think it is a good idea to split the drug infobox into more pieces than is absolutely necessary. Better to place all the identifiers/external links in a single {{Drug links}} template and skip the authority control file which because of very limited coverage of drugs by established authority control systems is probably not appropriate for drugs to begin with. Boghog (talk) 16:15, 28 February 2021 (UTC)

One additional potential short coming of CAS number for authority control is that there is often more than one salt form of a drug, and hence more than one CAS number assigned to a drug. It seems to me that a one-to-one relationship is needed between an entity and its authority control identifier whereas there is often a one-to-several relationship between a drug and CAS numbers. Boghog (talk) 14:54, 1 March 2021 (UTC)

Some relevant statistics about Authority control in drug articles:

• Infobox_drug & Authority_control – result count: 105, and many that do contain the {{Authority control}} template are dormant (i.e., no corresponding Wikidata).
• Infobox_drug – result count: 7993

Boghog (talk) 16:49, 1 March 2021 (UTC) One additional

What is being proposed here?

@Ajpolino and DePiep: I had not been following this discussion very carefully, but I was recently pinged and now you have my full attention. My starting position is if it ain't broke, don't fix it. The current {{infobox drug}} is logically laid out into sections, and even if it is long, it is easy to find the information one is looking for. Furthermore, I do not see a coherent plan for replacing it. At various points of this discussion, I have seen proposals to split the infobox into at least four pieces, a slimmed down drug infobox (at the top), a chemistry box (in the chemistry section), a links box (I assume in the external links section), and authority control. This is crazy. As far as I can tell, there are no authority control links that would be appropriate for a drug, so this proposal is a non-starter. Creating a chemistry box where many drug articles do not contain chemistry sections also makes no sense. So essentially we are talking about moving some external links to the {{Drug links}} template and no other templates. Correct? If so, it is essential that you first create a fully populated mock-up for what the slimmed {{infobox drug}} and {{Drug links}} would look like in a draft article. But I am still not convinced that splitting up the drug infobox is a good idea. Seems like a lot of work for dubious benefit. Boghog (talk) 20:24, 27 February 2021 (UTC)

You are right in noting that current proposals are manyfold, and so not straightforward (now).

You are wrong in claiming "that is crazy" full stop.

Point is, that there is information in the infobox that does not belong there (see WP:INFOBOX design principles).

As for {{Chembox}}: the same issue occurs there. Both since it has similar links, and since it also can contain drugbox data.

Whether these links should be in Authority Control or in regular article body text (as an overview list/see-also/extrenal links/info table ie not running text) depends on the data point (is one of my points).

So far, my first replies from top of my head. -DePiep (talk) 20:36, 27 February 2021 (UTC)

Moving information from one into four templates, particularly when one of the templates has no drug specific fields, is crazy. Full stop. Splitting one into two templates is less crazy, but still dubious. Boghog (talk) 20:43, 27 February 2021 (UTC)

My quick reply was, also, to note that "crazy" is not a good way to start a wiki discussion. Especially when the premisse is "I had not been following this discussion very carefully, but ..". Pls allow me this bumping. This said, the issue is complicated and can use good fleshouts & demo's sure. (My personal opinion is, in short & with broad brush: yes, some data points *must* be removed from the IB; yes, some info should/could be in the article, yes data can be present in IB, AC, body text, seealso, ELs AND repetition is possible). -DePiep (talk) 23:00, 27 February 2021 (UTC)

Boghog, I understand that you feel the current infobox is fine as is. We already discussed here whether the infobox should be trimmed, and you had your say. Your preferred option did not carry the day. Now I'd like to move on and implement the least bad solution we can come up with. I've proposed this new template {{Drug links}} to hold many of the external links that are current in {{Infobox drug}}. We're currently discussing which links should be moved to this new template. I've given my thoughts above and proposed a set of links to put in the template. Your thoughts on that set of links would be helpful to the discussion. Repeatedly reminding us here that you really don't wish to split the infobox at all is less helpful. I already know that's your wish. I disagree that "it ain't broke". If you wish to object to the whole process, your objection is heard. Here, you may contribute to the new setup, if you wish. I understand the discussion above has been freewheeling, but that's ok. Now I'm trying to focus us on a discrete proposal at the subheading "Proposal" above. I'll annotate it to make extra clear what is being proposed. Your comments are welcome. Ajpolino (talk) 02:15, 28 February 2021 (UTC)

Repeating the question asked above: So essentially we are talking about moving some external links to the {{Drug links}} template and no other templates. Correct? Boghog (talk) 06:46, 28 February 2021 (UTC)

There are authority control standards for information such as books, locations, organizations, and people. To the best of my knowledge, a generally accepted authority control standard has not been established for drugs. Would you then agree then that it is not appropriate to move some of the links to an authority control template?

Consistent placement of templates is also important so that readers can easily find information that they are after. {{Infobox drug}} is almost always placed at the top of the article. The most logical place for the {{Drug links}} template is in the external links section. Hence for consistency, it should always be placed there. Would you agree? Boghog (talk) 07:30, 28 February 2021 (UTC)

It is broken. I think it is important that the Infobox in the lead directly contains information for readers, otherwise it isn't an infobox. Anything that is merely an external link (e.g., licence data links) or a database primary key code (most of the identifiers), is not in itself providing information, and is a candidate for moving, such as to external links, which we usually locate at the bottom. -- Colin°^Talk 12:33, 28 February 2021 (UTC)

OK, if the purpose is to move most if not all of the external links from the infobox to external links section, and there would be no need for a seperate chembox in the chemistry section. Correct? Boghog (talk) 15:34, 28 February 2021 (UTC)

To your questions: Here we're discussing moving some (most?) external links to the {{Drug links}} template and no other templates. Folks at WT:MED (and above, myself included) had discussed moving some of the chemical data from {{Infobox drug}} to a new box that would live in the chemistry section. That is not currently being discussed, but I'm interested in following up on that part of the conversation some day. I'll ping the participants in the previous discussion (yourself included) when/if we have that discussion. I agree that {{Drug links}} should be consistently placed in the external links section, and should only contain external links. No opinion on whether some identifiers should be placed in {{Authority control}} in addition to - or instead of - {{Drug links}}. I'll just note that I read your comments above, and I've struck mention of Authority control from the proposal so that we can focus on links here. Ajpolino (talk) 17:23, 28 February 2021 (UTC)

Thanks for your clarifications and I think we are making progress. As for the proposed ChemDrugBox, the following statistics are relevant:

• Infobox_drug & Chemistry section – result count: 672
• Infobox_drug – result count: 7993

Less than 10% of drug articles contain Chemistry sections and I assume that the ChemDrugBox would be added to this section. Are you proposing to add chemistry sections to the 90% of drug articles that do not already contain one? (I think that would be great, but I do not think it is practical.) I think a better solution would be to remove chemistry parameters that have little relevance to drugs (boiling point and density), Jmol and the 3D representation (which are arbitrary and for the most part do not correspond to the active or low energy conformation), and leave the rest of the chemistry section in place. I do get that Project Medicine is allergic to chemicals, but there is no getting around that drugs are chemicals and without chemistry, there would be no drug. Hence chemistry is an essential part of the article and belongs in the drug infobox. Boghog (talk) 17:37, 1 March 2021 (UTC)

Mock drug links bar

Hi all, per conversation above, here's an example of how the drug links bar would look. The links are for aspirin. Here's the full version containing all the links currently in the Identifiers section (and a few ELs in the Clinical section):

Clinical resources	Drugs.com: Monograph MedlinePlus: a682878 EMA: EMEA/H/C/000471 Licensing agency: [link] (TBD)
Chemistry resources	PubChem: 2244 ECHA: 100.000.059 DrugBank: DB00945 ChemSpider: 2157 UNII: R16CO5Y76E KEGG: D00109 ChEMBL: CHEMBL25 PDB: AIN

And here's the shorter version removing the links that I think are redundant:

Clinical resources	Drugs.com: Monograph MedlinePlus: a682878 EMA: EMEA/H/C/000471 Licensing agency: [link] (TBD)
Chemistry resources	PubChem: 2244 ECHA: 100.000.059 DrugBank: DB00945 KEGG: D00109 ChEMBL: CHEMBL25 PDB: AIN

This much could be done by just moving the links currently in the infobox into this new box. If we want the template to automagically grab links from Wikidata, I'm afraid that's above my abilities, so that will depend on how much time DePiep and others with technical know-how have to dedicate. Happy to hear anyone's thoughts. Ajpolino (talk) 16:41, 31 March 2021 (UTC)

Thanks for the mockup. This helps enormously. One could argue that the DrugBank link is the single most useful in the entire list. It has an incredible breadth and depth of information which provides an excellent starting point for expanding stubs. It also provides a reciprocal link back to Wikipedia which I think is admirable. Also I think we should limit the Balkanization of data to the drug links bar and skip the authority control which is a non-starter and chembox is impractical since a large majority of drug articles do not contain chemistry sections. Boghog (talk) 15:36, 3 April 2021 (UTC)

Poof, DrugBank re-added. DePiep, any chance you've still got the time and interest to codify this? I'd be perfectly happy with a version where the links are manually added at each page (as they are at {{Medical resources}}). But if you've still got the bandwidth to make a version where links are auto-magically pulled from Wikidata, that would be amazing. But certainly no pressure. Ajpolino (talk) 05:16, 17 April 2021 (UTC)