Back in February 2024, I posted some example searches of the ZINC20 website at https://fold.it/forum/discussion/converting-general-compounds-into-ones-in-foldit-s-compound-library/page-2#post_78116
One was https://zinc20.docking.org/substances/?sub_id-matches=O=S=O (which back then listed many compounds containing SO2 groups). Another was https://zinc20.docking.org/substances/?sub_id-matches=O=S=O&mwt-ge=193 (which back then listed only twelve SO2 compounds with molecular weight (mwt) >= 193).
I just retried both of these searches and got "ZINC Error 500" with "'int' object has no attribute 'split'" for them. I have been trying similar rather simple searches for weeks now, but I keep receiving similar errors. Have others been having similar problems? Has anyone been able to make such searches work recently? Could it be that I just need to use a different web browser on a different machine to make the ZINC20 site work? Have they stopped supporting the ZINC20 site in favor of the ZINC22 site? Can the ZINC22 site do similar searches, and if so, how? See https://cartblanche22.docking.org/ for more details. I really don't itch to set up yet another account and password on another website, but maybe they've decided to only support searches done by people who login to the ZINC22 site (the ZINC20 site used to work for anyone, whether they had an account or not).
Other parts of the ZINC20 site still seem to work. For example, the molecule viewer at https://zinc20.docking.org/substances/home/ still lets you draw a compound and then convert that drawing into SMILES code.
Thanks for reading.
I checked both zinc15 and zinc20, they're not working for me either.
I think we can do something similar in arthor, e.g.
https://arthor.docking.org/index.html#real-database-22q1/sub:crqi/s/O=S=O
Haven't checked the exact meaning, just copied the URL from a random test.
Wow. Thanks rosie4loop. The Arthor query you gave me above seems to work. It says it found 40,000 matches in 176 ms. So far it has let me view 96 of them within my web browser. All compounds it shows seem to contain an SO2 group. Looks promising, but odds are there is more to learn. If you find any web pages that define things like /sub:crqi/s/, please let me know. Also, if you find examples that combine several search terms, like to find compounds with molecular weight > 200 that contain both -SO2 and -NH2, please let me know. Some other useful examples would be compounds containing 3 or more F atoms, compounds containing at least 2 hydrogen bond donors and at least 5 hydrogen bond acceptors, and compounds with >C=O and -NH- connected by 6 general atoms. ZINC20 used to be able to do searches like these, using a mix of html, SMILES, and SMARTS code. It would be nice if such searches could be done using Arthor instead.
Thanks again, Jeff.
ZINC is an academic project like Foldit is, so there's only a limited amount of debugging/testing time available. Which means certain bugs may slip through the cracks. – It looks like the error page has a contact address. If you're doing something which you think you should be able to do, but are still getting an error message, it may be worthwhile to contact the ZINC maintainers at the specified address with a bug report.
On both https://cartblanche22.docking.org/ and the ZINC20 site, I've seen the same e-mail address (chemistry4biology@gmail.com). I tried this e-mail address months ago but don't think I ever got a response that way. Maybe their spam filter didn't like my e-mail address or messages. The Arthor SO2 search https://arthor.docking.org/index.html#real-database-22q1/sub:crqi/s/O=S=O above gives a different e-mail address (jjiteam@googlegroups.com). I don't think I've tried this address yet. Nevertheless, it may be that an e-mail from someone on the Foldit staff or with an edu e-mail address would be more effective.
Wiki of Arthor api (https://wiki.docking.org/index.php/How_to_use_Arthor_API#Substructure_or_SMARTS_Search) gives some info on usage via url.
I've played around with the webpage a bit to see what's the meaning of sub:crqi. Basically it reflects the the search type in webpage's GUI:
sub = Substructure
sim = Similarity
sma = SMARTS
Substructure type has three more options that allows you to "lock":
c = Chain
r = Ring systems
qi = Properties
Thanks for the additional info, rosie4loop.
I've tried a few more Arthor searches. It seems like in the upper right corner for each search, a download icon always appears (it looks like an arrow pointing down to a horizontal bar). If you hover over the download icon, it shows the options "download limit", tsv, csv, and sdf. When I put 100 for the download limit and then picked tsv, csv, or sdf, it seemed to only download the first/top 100 ligands (hits) listed. tsv made a file with many tab-separated columns. csv made one with comma-separated columns. Both tsv & csv files had columns for the hit #'s (like 1-100), the SMILES code for each ligand, the Z# for each ligand, and the database each hit came from. The sdf file was much longer. I didn't see hit #'s or SMILES code there, but I did see each ligand's Z# and database listed. The rest of the sdf file looked like lists of heavy atoms and their coordinates for each ligand. I sure wish Arthor searches would have another download option that would make a pdf file containing images for the top ligands listed. So far, it seems like Arthor searches just show a handful of ligand images on the web browser screen at a time. If you scroll down and wait, it updates which handful are shown. If you try things like print to pdf, only that handful of images goes to the pdf file. It would be better if you could send 100's of images to the pdf file at once instead of just a few at a time.
The ZINC20 site, when it was working for me, would show images for up to 100 ligands at a time within my web browser. If I then printed this page to a pdf file, the resultant pdf file would contain all of these images, and if I clicked on any of them, it would open a web site for the ligand I clicked on. Having the pdf file on hand would save a lot of time because I wouldn't have to re-run old searches.
