Thus, the idea came up, can we create a set of 1 million unique IUPAC names found in literature?

We started out with using Europe PMC to get JATS XML files for the full texts of open access articles. Parsing the XML is easy and the text paragraphs are passed through OSCAR and OPSIN. That has not changed.

What did change last weekend is something I had long on my todo list (but life interfered). The first approach was to ask for named entities using the Europe PMC APIs. But I quickly realized that with OSCAR and OPSIN we could get more names out of the articles. The next step was to move from Google Colab to a command line script. That gave another boost, as explained in this second post in the series. We reached 200 thousand names in june 2025 but then things slowed down again in the growth. Two months later we only had 75 thousand more. However, plenty of discussion was happening and there turned out to be other, larger collections of IUPAC names under an open license. Millions of names, actually.

But another problem emerged. We were still using the Europe PMC API and were basically asking for open access articles between two dates. Practically, the API could answer requests between 1 and max 3 days. Beyond that, times outs and 404s became an issue. Moreover, because these dates are publications dates and not the dates on which the JATS were deposited, I had to got back to previous months and redo the queries. That gave another 5 thousand names since last August. Something had to change.

The new Approach

Europe PMC, however, also provides the JATS XML files as download on their FTP site. Already that august 2025 I had a prototype and knew it would change the game. These gzipped XML files are about 150 to 250 MB. Unzipped, about 1 GB each. Better, these files are based on Europe PMC identifiers, hopefully resolving the issue with using dates in the queries.

Now, parsing a 1 GB XML files is a total non-issue. I have done it plenty of times before. Just use a Simple API for XML (SAX) parser. This is a streaming parser giving you full control of how to parse things. It is ideal for this siutation: you just keep the current paragraph of text in memory and release that when done with that paragraph. That is, you do not have to read the full file in memory, just the bits you are interested in. I used this for my Chemical Markup Language patches for Jmol and JChemPaint back in the nineties.

Last weekend I finally made the jump. Use SAX to extract the <p> elements one by one, running OSCAR on them, filter with OPSIN, output that name, and clear the memory. Effectively, each gzipped file processes with a Groovy script in about 1 to 2 hours.

The output is a mesmerizing stream of scientific literature (which I will use until someone points me to a Java CLI library that creates a Matrix-style falling letters equivalent), tho less so as a static image:

In this plot, an x means a new article to be processed. Each . and o that follows is a single <p> element and the difference is that an o means at least one IUPAC name was detected in the paragraph.

Each gzipped file gives 400 to 500 new IUPAC names. Indeed, going from 288 thousand to 300 thousand was a matter of a day and a half. And earlier this afternoon we passed the 400 thousand IUPAC names. With about 230 gzipped files. Now, I am going back in time, and the sizes of these files are shrinking: Another 500 files and the size has dropped to around 125 MB, so a rough estimate suggests that we will end up with 650 to 700 thousand names this way. This will be completed in a few weeks (and mostly because I need to focus first on other things again, because I can use our computing cluster do this).

Regarding the original goal, fortunately, we are still publishing at a higher rate every year, and more and more articles are available as open access. So, I still have good hopes we will reach the 1 million IUPAC names. Also, keep in mind, we know how to boost this by simple name variations to several millions, even with the 400 thousand we have today.

Oh, and our next milestone will be in the pocket before I visit Christoph Steinbeck’s cheminformatics team in Jena!

Recently, I was close to writing up the context, because it is related to a new feature of the profile pages, where you now can look up citations to pathways that you first authored (see this post). And it also relates to the data I have been collecting around citation intention annotations: articles that cite one of the WikiPathways papers and mention a specific pathway, could be considered cito:usesDataFrom (see doi:10.1186/s13321-023-00683-2).

A third angle to citations to specific WikiPathways is the following. WikiPathways is used a lot in data analyses and putting experimental data in biological context. How researchers do this varies a lot, in multiple ways. But just thinking about this factually, research output cite specific biological pathways. And there are some interesting phenomena there. Back in 2015 at the Metabolomics Society meeting in San Francisco (apparently, I only blogged about the meeting only once?), when I visited the 500+ posters looking for interesting biological pathways, there were a lot of studies on different species, different diseases, different toxicities. The biological response had one thing in common: it always was the TCA cycle that was key (see doi:10.1096/FJ.11-203091 for a 2012 comparison of TCA models).

Thus, with so many articles mentioned specific pathways and deriving biological knowledge from this, what is reasonable to expect? Do we expect co-citation effects? That is, if two articles found the same set of pathways of interest to their data, is the data showing a similar biological response? Do we expect a similar thing like the above TCA cycle in metabolomics, something similar to the notion of frequent hitters (see doi:10.1021/jm010934d)?

Of course, to test this hypothesis we need data and the Cited In feature comes in. At the time of writing of this blog post, we can see on this page that 878 pathways have been cited a total of 2715 times. We are getting somewhere. This blog post will not analyze this data, which is one reason why I had not blogged about it. But from the above you can understand that I want to :)

The Cited In feature

This Cited In feature was introduced along with the new website (see doi:10.1093/nar/gkad960), where we change how GPML files are stored and how web pages are created from that. Because we are no longer confined to the MediaWiki platform (which has served the project for very long, very effectively), it is easier to integrate information from other sources. For example, from literature databases. This feature was developed by Alex Pico at the Gladstone Institutes (see this 2022 commit), where he uses the NCBI eUtils API to access PubMed Central. The data is then collected into this YAML file which then gets used to generate webpage content (like the section in the above screenshot and the page mentioning the current statistics).

Where is the data coming from?

As just explained, originally the data was only coming from NCBI. However, because I found many articles citting specific pathways that were not picked up by this approach, and I wanted more data, so I started searching Europe PMC the European partner of PubMed Central. However, I am not automating this. I want to see the data, the articles, and how people cite the pathways. I need to see that so that I can better understand how people are using the data/knowledge from WikiPathways. I cannot keep up with checking why people are citing my own research, but I once was. I learn(-ed) a lot from that.

I normally use a search that requires the word “WikiPathways” to be mentioned in the article (in most, but not all of them; citing literature you extend sounds like a core scholarly value, but is factually not systematically complied with), and then manually searching for “WP”. With close to 1000 PubMed Central articles mentioning WikiPathways in 2025 and that these are mostly full texts, I can see if the cite specific pathways. A good number of article mentions the WikiPathways identifier, e.g. the aforementioned WP4846. If the article only mentions a pathway title, I cannot confidently identify which pathway is cited, so I exclude that.

I originally started out manually editing the YAML file where the citations are collected, but by now use a script similar to Alex’ R script. This makes it far easier to scale up, as I just have to populate a three column TSV file, which is used by my R script to update the YAML file. This manual approach ensures that I am not looking at text mining results, but see the citation of the WikiPathways identifier with my own eyes. That’s just how I like it.

The full history of the YAML file content can be found on this GitHub page and this git blame tells you if the information came from PubMed Central via the API, or was added by me:

This is Open Science in action: added transparency and making it easier for anyone to verify, so that no one needs to be stuck in (dis)trust.

Of course, as we know from the CiTO ontology and real-world data, there are so many different reasons why journal articles are cited (just an example), the data in the YAML file and on the WikiPathways website in the Cited In feature does not have direct meaning. Just like a high citation count for an article or even a journal impact factor cannot be directly interpreted (despite so many researchers just blindly doing just that).

What’s next?

Well, while I did not do any analysis yet, and do not even know yet how much citations we need to reach some level of statistical significance, there are some observations I can mention:

• if your analysis included anything like linking your data to pathways, citing those pathways is a good way to give credit to the researchers that created that pathway
• if you cite data, please cite that as accurately as possible, see e.g. DataCite
• I wish all journal articles citing specific pathways from WikiPathways would include the pathway identifier
• I congratulate those authors that even mentioned the revision of the pathway! well done!

And about biological interpretation, our group has long published that some genes with differential data mapping to a pathway does not imply that that pathway is really affected. Gene-set enrichment and over-representation analysis are a starting point; not a conclusion. I wish more people were more aware of the work in our (now) Translational Genomics research group. Like that of Martina Kutmon (now as MaCSBio²), whom I have had the pleasure of collaborating with for quite some years now (and long time archtect of WikiPathways).

There is so much more I want to write up about WikiPathways, but I leave it to this for now.

We have now integrated this curation badge into the author and community pages on the (not so) new WikiPathways website too. Authors can now find curation reports for pathways they started and also for the community pages:

A second new feature is the “Citations” tab on both pages, which link to Europe PMC with a dedicated search for articles mentioning those author or community pathways:

We hope you like it!

I wanted to see what I could recover, and here I describe what I did and what could be done next.

A list of all articles

The first step is actually to create a list of all articles published in the IJC and collect as much metadata about them as possible. With just over 100 articles, I decided to use Wikidata, as a machine-readable database, supporting the curation and reporting. I wanted at least two independent sources, and for Wikidata, use public resources. That means, while Web of Science does have a list of all articles, I only used this for validation, and not as information source. Instead, I used citations to IJC articles and, of course, the Internet Archive (IA). It turns out a query like this does wonders (well, for the abstracts; I did not find full-texts archived on IA):

https://web.archive.org/web/*/http://www.ijc.com/abstracts/*

I found that all but one article had the abstract archived in the IA. Here’s an example:

This gave my a lot of information to add to Wikidata. Title, publication date, volume, article number, keywords, an absstract, and, of course, the list of authors. Some authors I know personally, many I did not. But it did allow me to enter all articles to Wikidata along with the authors and “author” (P50) or “author name string” (P2093).

The article authors

It also turned out that multiple authors listed their IJC article on their public ORCID profile. That greatly helped identification. I managed to link many authors to mostly existing Wikidata items:

I already mentioned that I used Wikidata to collect this information. Besides the interactive visualization with Scholia, it also gave me the option to track my progress with SPARQL queries. For example, this query helped me do that author FAIR-ification:

You can see here two columns with author information, one for P50 and the other for P2093. There is quite some identification left to be done, and additional information is welcome:

Sources

So, that brings us to this list of sources:

• Internet Archive: abstracts and metadata
• ORCID profiles: ORCIDs of (some) authors
• Google Scholar: metadata and citations
• Web of Science: independent list for external validation

Because there is plenty of work left to be done and I hope the collected information will further spread in library collections, I added sources as much as possible. This query lists for all articles the Web of Science identifier (recorded so that everyone can check the consistency), the link to the Internet Archive-d abstract page, and a link to a known full text (five).

If you wonder, neither OpenAlex or Europe PMC have a full list.

What’s next?

I do not have a formal training in archiving, but I am happy with the minimal viable metadata collection. I know more can be done (and love to hear your pointers and suggestions): more author identies, better coverage of keyword annotation, etc. But I think an important addition is adding citations to and from the IJC articles are important. The journal predates efforts like the I4OC and Open Citations, so I may have to manually recover citations from Google Scholar. I will have to report on that later. But you can enjoy the citations that are already there. And now that we have sufficient metadata, I can use this to find more full texts.

Btw, I have made contact with Prof. Steven Bachrach, who founded the journal and was the Editor-in-Chief.

In this post I want to give an update.

275k IUPAC names

I have continued running the scripts to detect new IUPAC names in full text, open access papers in Europe PMC, but something more awesome actually did much more since the June post: in July I received a pull request from mnietfeld with more than 40 thousand unique and new IUPAC names from the Beilstein Journal of Organic Chemistry (see also their LinkedIn post or this archived version that doesn’t require an account). While Europe PMC provides these articles too (and actually one of the first I analyzed), a lot of these names come from supplementary information, not provided by Europe PMC. Thanks!

This is focusing on names from primary literature, but there is more happening. Because I want to restrict the above project to names from primary literature (and supplementary information is still that), I have not been sure what to do with other collections yet, and they have been coming in. I have been taking notes in the project issue tracker, for future reference (like now, here). I have not forgotten about these!

Other large collections of IUPAC names

4M, CCZero
Let’s start with the news yesterday. The Chemical Biology Services team released 4 million IUPAC names from patent literature as CCZero! The CCZero license/waiver makes it compatible with our list. Their Zenodo release:

… contains IUPAC names text-mined from patents (US, WIPO, EPO, Chinese, Japanese).

The post also includes a nice example of the complexity of IUPAC names which makes the counting of unique names tricky: O-methylphenol and o-methylphenol. Thanks, Noel and the rest of the EMBL-EBI team!

2.3 million, CC-BY
And then Haydn Jones was one of the earliest to coin in, and released 2.3 million IUPAC names under the CC-BY license.

850k, CCZero
Wikidata also turnes out to have many IUPAC names. Adriano found more than 850 thousand IUPAC names, see this project.

Next week I will do some comparisons of the datasets with a clear Creative Commons license.

Even more

Beyond these five data releases, there is more. PubChem and other databses have millions of names, but often these are generated by proprietary software. These IUPAC name collections may be under some license agreement, and thus not compatible with Open Science. This is why it is so important that we very clearly know where these names are coming from.

5-6 million, license unclear
I also learned about ChemPile about which Adrian Mirza explained me it has about 5-6 million IUPAC names. But the source of this list of names is not yet clear to me.

Names from PhD theses and preprints
I also want to give a shout out to Peter Murray-Rusts proposal to start extracting IUPAC names from PhD theses. There have been projects to extract chemistry from PhD thesis in the past, and this will yield a lot of unique names. Please ping Peter, if you want to get involved in his idea!

What’s next

I am so excited with all these efforts and very grateful with the contribution by Beilstein. I really hope more Open Science publishers will follow, like perhaps the Royal Society of Chemistry for which it should be easy, with their Project Prospect background!

I am also excited by the release by ChEMBL under CCZero. That will allow the WikiProject Chemistry use this for Wikidata!

So, I have one week left to write the article about the work we started in March. The outlook is bright. I played last week with the Europe PMC full text downloads and can confirm that should yield thousands of additional names from the full texts. A single download file gave me more than two thousand new unique names. I think the 500k IUPAC names is absolutely in reach with purely the full texts from Europe PMC.

This brings us to the end of 2025. By then, we should have a many millions of openly-licensed IUPAC names. And by March 2026, I hope we reached the 1M IUPAC names extracted from primary literature. That will require some creativity and enthusiasm, but sounds feasible!

But in the most recent iteration, triggered by some work for WikiPathways, I was using Europe PMC to find articles that mention WikiPathways and then search in the full text for the string WP, as a trigger for the possible mention of WikiPathways pathway identifiers, which look like WP4846. The use of compact (resource) identifiers (see doi:10.1038/sdata.2018.29) is minimal, but at least some articles use identifiers.

That allows me to extend our WikiPathways knowledge graph of articles citing specific pathways. At the time of writing, we collected 2509 citations from 440 different articles to 883 different pathways. Now, I want to blog about that more, but it’s related to an observation.

Information loss

Now, back in the late ninities I learned about GNU/Linux and after playing with Red Hat and Suse, I settled for Debian. One of the things I learned is that, generally, information corruption (like data loss) is an absolute red flag, a no-go, a total showstopper.

And then we have this in publishing, the one area where data corruption must also be a no-go:

In this image, the left side shows a screenshot of the publisher version of the article and on the right side the version in Pubmed Central (PMC). PMC has been an important project to archive full text versions of articles:

11.2 million articles are archived in PMC.

So, this is really bad! The archived version is not really useful. As a human I already struggle to read the degraded image, let alone an algorithm.

Does that matter? Yes, projects like the awesome Pathway Figure OCR (see doi:10.1186/s13059-020-02181-2) depend on images to be FAIR enough to extract information. (Side note: yes, these images should be vector graphics, but commercial publishers decided about twenty years ago that they could not care enough.)

At this moment, I do not know where the information is lost. Maybe PubMed Central is storing the images in a low resolution. Maybe the publisher provides PMC with a low resolution image. But to me, this must be solved as soon as possible. This is utterly unacceptable.

I wonder what the authors of the article (doi:10.1186/s13287-025-04166-z) I took as example think of this.

Is that a big issue? Hell, yes. Where do you think the FAIR ideas came from? And why FAIR in ten years has not brought about the change it was hoping for?

For me, my fascination for curation started as a student, around 1995, with the Dictionary on Organic Chemistry. At that time, my interest came from wanting to learn about chemistry and biology. During my M.Sc. and PhD, it was obvious how essential it was to derivating correct scientific conclusions from your experiment. Data, knowledge, and software alike, imo. And because curation is expensive, not having to repeat it, I prefer to do it as Open Science.

Curation

Of course, curation has been part of doing science, but to a large extens is separate step from doing science. It is done by database developers, librarians, and chemo- and bioinformaticians. For example, Chemical Abstracts Service (CAS) started over 100 years ago and started indexing chemical structures in 1965. The curation is an ongoing process, also for old records.

Biocuration is getting more and more attention:

The recognition and rewarding by having the International Society for Biocuration (ISB, Scholia page) should not be underestimated (doi:10.1038/455047A). Their Annual International Biocuration Conferences have been running since 2005. And with their awards, they give the biocuration work recognition and, literally, rewarding:

• Biocuration Career Award (2016-2021)
• Excellence in Biocuration Early Career Award (2022-)
• Excellence in Biocuration Advanced Career Award (2022-)
• Exceptional Contribution to Biocuration Award (2017-)

My curation Curriculum Vitae

I don’t have a good curation CV. For a large extend because the curation has been part of a study. The curation itself does not get recognized, and only the journal article does. With datasets slowly getting more recognition, so does data curation, but data curation is not really part of how we do FAIR at this moment, and via this route not getting the attention it gets.

But since I have been updating my CV anyway, I dug up some curation I am proud of:

• the Dictionary on Organic Chemistry, which no longer exists, but it started my Open Science chemistry research
• the Blue Obelisk Data Repositry (BODR), which has been part of various GNU/Linux distributions (see also doi:10.1021/ci050400b). A new version is long overdue
• I contributed hundreds of NMR spectra with uncommon nuclei to NMRShiftDb
• Wikidata, see this preprint, but also many small projects, like adding CXSMILES for polymers, and main subject annotation in Scholia
• WikiPathways (see these blog posts), where I started curating metabolites in 2012, set up a computer-assistent curation platform using SPARQL, and were an early curator of SARS-CoV-2 biological processes
• citation intent annotation with the Citation Typing Ontology, see this Scholia overview
• nanosafety ontology and data: the eNanoMapper Ontology (ENMO), NanoWiki, JRC nanomaterial index and the ERM indentifier database
• made RDF for supplementary information (e.g. this NanoE-Tox spreadsheet, full databases, like ChEMBL and NMRShiftDb
• organized an online ChemCuration event (inspired by the ISB annual meetings!)

I am also curation my blog, which was originally in blogger.com but being ported to Markdown with extra annotation. That includes updating URLs and annotation of blog posts with chemicals, grants, and intention-typed citations.

Long tail

Of course, I have my Wikipedia edits, and contributed to projects like Bioregistry.io, FAIRsharing, regularly submit missed mentions to Altmetric.com, etc. There is a long tail in curation. And there is a lot of curation hidden in my literature list.

And that long tail matters to me. I want every researcher to pick up the challenge to curate their own research output. Put your experimental data in databases, add important provenance, get the details rights. This is essential to reduce the cost of doing research, and that is more important than ever.

BTW, I must note that our bioinformatics team colleagues too have done a tremendous amount of biocuration, in WikiPathways (Denise, Freddie, Susan), in nanosafety (Jeaphianne, Ammar), and in toxicology (Marvin), just to name a few. Often together with B.Sc. and M.Sc. students (which can work really well).

Award nomination

And I hope this makes it clear why I am delighted to was nominated last week for an ISB Excellence in Biocuration Advanced Career Award. The list of past awardees is impressive, as are the other nominations: Laurel Cooper, Oregon State University/USA, Steven Marygold, University of Cambridge/UK, Saurabh Raghuvanshi, University of Delhi/India, and Kimberly Van Auken, California Institute of Technology/USA.

It’s an honor to be listed along these other nominees and being nominated is a great recognition! With a thank you to the person who proposed my nomination.

Additionally, it also does not have an RSS feed, and we should indeed be using RSS more. So, I hacked something up and impressions were positive. Based on Jekyll and the experiences I had with this blog, I modelled individual articles as blog posts and biohackathons as tags. That automatically gave me the RSS feeds:

• feed for new BioHackrXiv preprints (or this JSON Feed)
• feed for Europe BioHackathon 2021 (based on the BH21EU tag)

It’s still very much in progress, but it’s now live at index.biohackrxiv.org:

Extras

Some extra goodies already there include:

• the Altmetric.com donut
• links to Scholia
• links to Europe PMC

The overview of biohackathons looks like this (the tag size follows the number of preprints for that biohackathon):

The current count actually is just above 230 thousand IUPAC names, but further growth may require new approaches, such as the four ideas I posted earlier. I have gone through all full-text Open Access articles provided by the Europe PMC API. Now, this list is not static, but I wanted to start using their bulk downloads anyway.

The current results

I have been looking at the names coming in. Some are short, others long. The complexity is fascinating and I will have to brush up my cheminformatics skills to make chemical space splots and visualize the structural diversity. I also note the current workflow does a good job at unicode characters, and we have plenty of names like ε,ε-carotene-3,3’-dione. There are also names that I do not expect to be really valid, like hydroxymethyl methacrylate- that end with a hyphen (41 in total), but their overall count is low. And OPSIN is happy with it, so the name fits the rules.

The ten longest names (so far) are these (with the lengths 322, 324, 332, 357, 371, 373, 376, 421, 429, and 626):

(5Z)-3-ethyl-5-[[4-[15-[7-[(Z)-(3-ethyl-4-oxo-2-sulfanylidene-1,3-thiazolidin-5-ylidene)methyl]-2,1,3-benzothiadiazol-4-yl]-9,9,18,18-tetra(nonyl)-5,14-dithiapentacyclo[10.6.0.03,10.04,8.013,17]octadeca-1(12),2,4(8),6,10,13(17),15-heptaen-6-yl]-2,1,3-benzothiadiazol-7-yl]methylidene]-2-sulfanylidene-1,3-thiazolidin-4-one
(Z)-[[4-[[(Z)-N’-carbamoyl-N-[2-[2-[2-[[3-[(4S)-6,8-dichloro-2-methyl-3,4-dihydro-1H-isoquinolin-4-yl]phenyl]sulfonylamino]ethoxy]ethoxy]ethyl]carbamimidoyl]amino]butylamino]-[2-[2-[2-[[3-[(4S)-6,8-dichloro-2-methyl-3,4-dihydro-1H-isoquinolin-4-yl]phenyl]sulfonylamino]ethoxy]ethoxy]ethylamino]methylene]urea dihydrochloride
2-((Z)-2-((6-(4-(6-((Z)-(1-(dicyanomethylene)-5,6-difluoro-3-oxo-1H-inden-2(3H)-ylidene)methyl)-4,4-bis(2-ethylhexyl)-4H-cyclopenta[1,2-b:5,4-b′]dithiophen-2-yl)-2,3-bis(hexyloxy)phenyl)-4-(5,7-diethylundecan-6-yl)-4H-cyclopenta[1,2-b:5,4-b′]dithiophen-2-yl)methylene)-5,6-difluoro-3-oxo-2,3-dihydro-1H-inden-1-ylidene)malononitrile
(2S,4S,5R,6R)‐5‐acetamido‐2‐[(2S,3R,4R,5S,6R)‐5‐[(2S,3R,4R,5R,6R)‐3‐acetamido‐4,5‐dihydroxy‐6‐(hydroxymethyl)oxan‐2‐yl]oxy‐2‐[(2R,3S,4R,5R,6R)‐4,5‐dihydroxy‐2‐(hydroxymethyl)‐6‐[(E,2S,3R)‐3‐hydroxy‐2‐(octadecanoylamino)octadec‐4‐enoxy]oxan‐3‐yl]oxy‐3‐hydroxy‐6‐(hydroxymethyl)oxan‐4‐yl]oxy‐4‐hydroxy‐6‐[(1R,2R)‐1,2,3‐trihydroxypropyl]oxane‐2‐carboxylic acid
(2R,3S,4R,5R,7S,9S,10S,11R,12S,13R)-7-[(benzylcarbamoyl)oxy]-2-(1-{[(2R,3R,4R,5R,6R)-5-hydroxy-3,4-dimethoxy-6-methyltetrahydro-2H-pyran-2-yl]oxy}propan-2-yl)-10-{[(2S,3R,6R)-3-hydroxy-4-(methoxyimino)-6-methyltetrahydro-2H-pyran-2-yl]oxy}-3,5,7,9,11,13-hexamethyl-6,14-dioxo-12-{[(2S,5R,7R)-2,4,5-trimethyl-1,4-oxazepan-7-yl]oxy}oxacyclotetradecan-4-yl 3-methylbutanoate
2-[4-[2-[[(2R)-1-[[(4R,7S,10S,13R,16S,19R)-10-(4-aminobutyl)-4-[[(2R,3R)-1,3 dihydroxybutan-2-yl]carbamoyl]-7-[(1R)-1-hydroxyethyl]-16-[(4-hydroxyphenyl)methyl]-13-(1H-indol3-ylmethyl)-6,9,12,15,18-pentaoxo-1,2-dithia-5,8,11,14,17-pentazacycloicos-19-yl]amino]-1-oxo-3 phenylpropan-2-yl]amino]-2-oxoethyl]-7,10-bis(carboxymethyl)-1,4,7,10-tetrazacyclododec-1-yl]acetic acid
(2R,3S,4R,5R,7S,9S,10S,11R,12S,13R)-12-{[(2R,4R,5S,6S)-4,5-dihydroxy-4,6-dimethyltetrahydro-2H-pyran-2-yl]oxy}-7-hydroxy-2-(1-{[(2R,3R,4R,5R,6R)-5-hydroxy-3,4-dimethoxy-6-methyltetrahydro-2H-pyran-2-yl]oxy}propan-2-yl)-10-{[(2S,3R,6R)-3-hydroxy-4-(methoxyimino)-6-methyltetrahydro-2H-pyran-2-yl]oxy}-3,5,7,9,11,13-hexamethyl-6,14-dioxooxacyclotetradecan-4-yl 3-methylbutanoate
(2S,4S,5R,6R)‐5‐acetamido‐2‐[(2S,3R,4R,5S,6R)‐5‐[(2S,3R,4R,5R,6R)‐3‐acetamido‐5‐hydroxy‐6‐(hydroxymethyl)‐4‐[(2R,3R,4S,5R,6R)‐3,4,5‐trihydroxy‐6‐(hydroxymethyl)oxan‐2‐yl]oxyoxan‐2‐yl]oxy‐2‐[(2R,3S,4R,5R,6R)‐4,5‐dihydroxy‐2‐(hydroxymethyl)‐6‐[(E,2S,3R)‐3‐hydroxy‐2‐(octadecanoylamino)octadec‐4‐enoxy]oxan‐3‐yl]oxy‐3‐hydroxy‐6‐(hydroxymethyl)oxan‐4‐yl]oxy‐4‐hydroxy‐6‐[(1R,2R)‐1,2,3‐trihydroxypropyl]oxane‐2‐carboxylic acid
(2R,3S,4R,5R,7S,9S,10S,11R,12S,13R)-12-{[(2R,4R,5S,6S)-4,5-dihydroxy-4,6-dimethyltetrahydro-2H-pyran-2-yl]oxy}-2-(1-{[(2R,3R,4R,5R,6R)-5-hydroxy-3,4-dimethoxy-6-methyltetrahydro-2H-pyran-2-yl]oxy}propan-2-yl)-10-{[(2S,3R,6R)-3-hydroxy-4-(methoxyimino)-6-methyltetrahydro-2H-pyran-2-yl]oxy}-3,5,7,9,11,13-hexamethyl-7-({[2-(2-methyl-5-nitro-1H-imidazol-1-yl)ethyl]carbamoyl}oxy)-6,14-dioxooxacyclotetradecan-4-yl 3-methylbutanoate
N-[(2S,3R,4R,5S,6R)-5-[(2S,3R,4R,5S,6R)-3-amino-5-[(2S,3R,4R,5S,6R)-3-amino-5-[(2S,3R,4R,5S,6R)-3-amino-5-[(2S,3R,4R,5S,6R)-3-amino-5-[(2S,3R,4R,5S,6R)-3-amino-5-[(2S,3R,4R,5S,6R)-3-amino-4,5-dihydroxy-6-(hydroxymethyl)oxan-2-yl]oxy-4-hydroxy-6-(hydroxymethyl)oxan-2-yl]oxy-4-hydroxy-6-(hydroxymethyl)oxan-2-yl]oxy-4-hydroxy-6-(hydroxymethyl)oxan-2-yl]oxy-4-hydroxy-6-(hydroxymethyl)oxan-2-yl]oxy-4-hydroxy-6-(hydroxymethyl)oxan-2-yl]oxy-2-[(2R,3S,4R,5R,6S)-5-amino-6-[(2R,3S,4R,5R,6R)-5-amino-4,6-dihydroxy-2-(hydroxymethyl)oxan-3-yl]oxy-4-hydroxy-2-(hydroxymethyl)oxan-3-yl]oxy-4-hydroxy-6-(hydroxymethyl)oxan-3-yl]carbamate

That last compound has the InChIKey DKPKDPKJVDQUPD-XGBIXEJNSA-M and cannot be found in Google nor in PubChem. It looks like this:

There are some closely related compounds, though.

Chemicals only published about once

Some related data was blogged by Henry Rzepa last week, with this quote by Lee from CAS:

38.5% of the current substances have only 1 reference

Apparently, based on CAS Registry data, about 1 in 3 chemical structures are only published about once. And two in three are published about at least twice. I agree with Henry here, with organic chemistry literature in mind, I would have expected that 38.5% to be higher.

Anyway, since this project is not tracking in which articles IUPAC names are found, I have nothing to study this.

1 million IUPAC names

So, the primary goal of this project is to reach one million IUPAC names. We are currently at around 23%. Not bad, considering we started in Februari. And we have plenty of untouched literature left.

But I also applied idea 1, the varying names. The idea is that this was I can explode the number of compounds. In that compounds above, just the number of variations by enumerating all OH replacements with OMe and OEt would help a lot.

Because I wanted to make sure I could answer positively at the ICCS if we made it to one million CCZero IUPAC names, I implemented a very simple enumeration script. Really dumb approach. But the results are interesting. I started with the 200026 names from the milestone. If I explode these names, I get 1,377,127 IUPAC names, well above the target. Even if I remove name variations due to unicode variations for hyphens, I still have 1,162,107 IUPAC names.

Something interesting I cannot fully understand at this moment yet, however, is the following. When I calculate the number of unique InChIKeys for the milestone, I get 117,726 keys, and when I do this for the list of name variations, I get 203,979 keys. So, while the IUPAC name list is about five times as long, the list of InChIKeys is not even twice as long. Well, I guess that is why this is called research.

This milestone release took a bit longer. Going from 50 to 100 thousand is a bigger step than from 10 to 50 thousand, but the open access chemistry literature was already done by then. Basically, I ran out of open access chemistry publications. The scripts are now finding names in all (open access) literature, and the number of new names per articles is a lot lower. Still about 1 in every twenty to 30 articles. But the diversity in names is not really going down, which is important.

The first few weeks, I used the Google Colab to run a Jupyter notebook, initial created by Magnus, but having to process more articles to get a reasonable number of new IUPAC names required longer and longer jobs, and then Google Colab is not really fit (well, the free version anyway). So, I started using a local script. That turned out to be able to handle up to 20 thousand articles in one go and runs at least twice as fast. Moreover, I can run three of them in parallel.

And that had impact. With each commit around 1000 new IUPAC names, the number of commits went up remarkably last week:

At the current speed, I think we’ll make it to 150k soon and I added a new milestone for 200k, which sounds doable in the next three week. That also means that 1M extracted IUPAC names from literature has become a reasonable goal. And we can start thinking about the 2, 5, 10, 50 and 100 million IUPAC names. Those are, at the current speed, rather unlikely to reach from the open access literature anytime soon. That brings us to the question, what will. Well, I have some ideas.

Idea 1: name variations

First, I am figuring out some ways to make variants of names (no, not based on hyphens and spaces; that’s too easy), but actual variations of the chemical structures. For example, I could exhaustively replace “methoxy” with “ethoxy”, and iterate the halogens and acyl chain lengts. I have little doubt that I can grow the list with this approach easily a 5-fold, maybe even a 10-fold.

Idea 2: hallucination

Another idea is that I could use tools that can generate IUPAC names for a limited set of compounds. I once wrote code for alkanes myself and if I can find that, I may be able to generate additional names. But perhaps more realistic is that I train a deep learning model and have it generate names for all compounds in Wikidata (~1.5 million) or PubChem (>100 million). STOUT needed 81 million compounds (doi:10.1186/s13321-021-00512-4), but I don’t need a good model; I just need a model that comes up with new, valid names. Hallucinated names, but valid.

While the list of valid names grows, I can retrain the deep-learned model and repeat. As long as the diversity remains high enough, one could hypothesize that the deep learning will learn new tricks. And then, that should be a near infinite source of additional names.

Idea 3: (semi-)closed access literature

Also, I haven’t touched closed access articles yet. This is all based on the collection of full texts in Europe PMC. For example, I could start with the green open access article in (Dutch) university repositories, particularly those with large chemistry departments. PDF to text tools are mature enough that this will provide a new source. Oh, and perhaps PhD thesis, which are now also increasingly archived in university repository under open access. And that reminds me of a Dutch project two decades ago doing exactly that. I wish I remembered the name.

Idea 4: alternatives to Oscar4 and Europe PMC

So, the first round of named entity recognition was with Europe PMC itself, as explained in the first post. The move to Oscar4 helped a lot. But there exist many other chemical NER tools, like (doi:10.1093/bioinformatics/btn181. And those may find an additional number of names, even with just the literature I already covered.

Well, you get the idea.

ICCS poster rejected

Unfortunately, the ICCS poster abstract did not make the cut. The score was high enough, but they received many abstracts and had to make a selection (of course, I am part of the ICCS organization, and have more details of how it came about). I really like the project, and eager to write up a paper around it.