How to access the Foldit models on the PDB-REDO site

Started by beta_helix

beta_helix Staff Lv 1

1) Go to https://pdb-redo.eu/archive
2) Select FIUSED from the Property Filter tab
3) Click on Add Property Filter to the left of FIUSED (you should now see "FIUSED value is true" above it)
4) Click the Submit button

All the entries below are the ones where Foldit players improved on the PDB and on PDB-REDO!

The tricky part is that all the files for each entry are .json files (or a .zip of json files)
So if you look at the first one, 104l, for example:

entry ID….Version Hash….Date…………………Links
104l………..e0703f246……..2024-09-07……..cif mtz data versions zip

You can download the 104l_e0703f246_versions.json file by clicking on "versions", and if you open it in any text editor, you'll see this at the top:

{
"data": {
"PDBID": "104l",
"coordinates_revision_date_pdb": null,
"coordinates_revision_major_mmCIF": "1",
"coordinates_revision_minor_mmCIF": "3",
"coordinates_edited": true,
"reflections_revision": "1_0",
"reflections_edited": false,
"foldit_used": true,
"foldit_id": 2013743,
"foldit_player": "Galaxie"
},

Where we can see that Galaxie submitted an improved model for PDB structure 104l :-)

LociOiling Lv 1

Thanks, one comment and a question or two.

As a comment, the search starts with a program and a property. The program named "binliner" appears first. The program name doesn't seem to matter, as long as you don't add a program filter. The property "FIUSED" and the associated property filter are all that matter.

For question one, I see a player "null" is listed in several entries. There is a player named "null", who has never completed a puzzle since 2008, so I don't think that's it. What does "null" mean in the PDB-redo database? (Is it a privacy thing?)

Second question, there's also a "RealPerson" listed, but that's an actual player, who has completed five puzzles. Puzzle 2514 was one of the five, and RealPerson is listed in 42nd place. How did RealPerson's result on that puzzle make the cut?

beta_helix Staff Lv 1

Yes, "null" means we did not get permission from the Foldit player to post their username.

I will look into "RealPerson", but it is very possible that the 42nd-ranked solution could be the "best" one. Surprising for sure, but possible!

LociOiling Lv 1

I'd also note that there's an entry "2lz2.tmp", where all the file links return zero bytes. The entry "2lz2" is also present, and everything looks good there. (Congratulations to dcrwheeler on 2lz2, also known as Puzzle 2508: Refine Density Reconstruction 6.)

The erroneous "2lz2.tmp" means there are only 64 matches at the moment, not 65.

beta_helix Staff Lv 1

These are all great points, that we will need to provide in order to publish the paper.

For now, we are constrained by the parameters of PDB-REDO's website. While this is better than nothing, I completely agree that it would be best if we could host a proper table on fold.it with all that information.

We didn't want to wait until then to share this with all of you, though.

Bletchley Park Lv 1

@beta_helix I second my actual question: if I see a fold_it number in the files on PDB-REDO, which puzzle number is that ? I can see the directories on my system with those numbers, but I cannot retrace that to a website puzzle number.

Ok, I found out just now that it is the number that shows when hovering over the puzzle URL. But a list (ID-puzzle vs number) would save a lot of scrolling and probing time.

LociOiling Lv 1

The wiki now has a PDB-REDO page with the total entries by player.

The PDB-REDO Foldit results page lists each entry, with a link to the PDB and PDB-REDO pages, and the Foldit puzzle page.

I see the most recent entry was added on 28 January 2025, so I'll check back periodically to see what's new.