mark atoms inside rings by ZontaNicola · Pull Request #12 · rdkit/molecular_templates

ZontaNicola · 2024-11-13T15:42:59Z

adding functions to mark atoms inside rings so that we don't match molecules that have substitutions to those atoms.

note that I'm doing something wrong, because the query does not show up in the smiles, but the atoms are correctly identified (i.e. the atomic number is changed to 4...)

PLEASE DO NOT MODIFY template_smiles.h DIRECTLY

Add your templates to the templates.smi file in this repository. Add one CXSMILES per line, and please do not insert any blank lines.
Commit your changes and create the PR. The CI workflow will run on the new templates and validate them, and once they are approved, they will be automatically added to the header.

For more information please check https://github.com/rdkit/molecular_templates/blob/main/README.md

rachelnwalker · 2024-11-14T01:14:50Z

+        if aidx in inner_atoms:
+            # this atom cannot have substituents, so we substitute it with a query with a fixed degree
+            atom = mol.GetAtomWithIdx(aidx)
+            query_atom = Chem.AtomFromSmarts(f"[#4&D{atom.GetDegree()}]")


Was there a reason for #4? I think dummy atoms/#0 would make sense but I remember you mentioning there being a reason for having specific atom types

oh that's because the &D2 part does not show up in CXSMILES (I need your help for this! It was working from C++), so I temporarily put #4 to make sure the atoms are correct... #0 works for me, #6 also works and maybe leads to slightly more readable SMILES. I saw in the code that there's a plan to use wildcards on the bonds too, that also makes sense to me

D2 is a SMARTS feature, so it's not going to show up in output SMILES.
You could try generating CXSMARTS by calling Chem.MolToCXSmarts()

rachelnwalker · 2024-11-14T01:16:12Z

To get the github actions bot to show pictures of the new/updated SMILES, you have to run this script and rewrite templates.smi -- since this is a change to the script directly, could you show a couple before/after resulting SMILES?

ZontaNicola · 2024-11-14T06:32:16Z

sure, as soon as I can get the wildcards to show up in the SMILES...

rachelnwalker · 2024-11-14T15:34:27Z

sure, as soon as I can get the wildcards to show up in the SMILES...

I think you could also create the 'gallery' using this script and attach it manually to the PR https://github.com/rdkit/molecular_templates/blob/main/src/update_gallery.py

I'm not sure how great the rdkit image generation will be though with the SMARTS

ZontaNicola · 2024-11-18T06:24:55Z

after talking with @rachelnwalker I reverted the changes to the header as @ricrogz suggested. Once this code is approved we can update the .smi file in a second review.

Here's how the last three smiles look with the new code

[#0&D2]1-[!#1]-[#0&D3]2-[#0&D3]3-[!#1]-[#0&D2]-[#0&D3]4-[#0&D3]-1-[#0&D3]1-[!#1]-[!#1]-[#0&D3]-2-[#0&D3]2-[#0&D2]-[!#1]-[!#1]-1-[!#1]1-[!#1]-[#0&D2]-[#0&D3]-2-[#0&D3]-3-[!#1]-[!#1]-[#0&D3]-4-1 |(1.72266,-1.8102,;1.72266,-5.53747,;0.674242,-4.23843,;-0.825758,-4.23843,;-1.94049,-5.53747,;-1.94049,-1.8102,;-0.825758,-0.511159,;0.674242,-0.511159,;1.42424,0.787879,;2.92424,0.787879,;2.92424,-2.93939,;1.42424,-2.93939,;0.674242,-1.64036,;1.72266,-0.341318,;1.72266,3.38596,;0.674242,2.08692,;-0.825758,2.08692,;-1.94049,3.38596,;-1.94049,-0.341318,;-0.825758,-1.64036,;-1.57576,-2.93939,;-3.07576,-2.93939,;-3.07576,0.787879,;-1.57576,0.787879,)|

[!#1]1-[!#1]2-[#0&D3]3-[!#1]-[!#1]4-[!#1]5-[!#1]-[#0&D3]-3-[!#1]-1-[#0&D3]-5-[#0&D3]-2-4 |(-1.68833,-1.6056,;-0.474804,-0.723925,;-1.13662,-0.251107,;0.360708,1.45266,;-0.93833,2.20266,;-2.43833,2.20266,;-3.73737,1.45266,;-2.19718,-0.251107,;-2.90185,-0.723925,;-2.43833,0.70266,;-0.93833,0.70266,)|

[!#1]1-[!#1]-[!#1]2-[!#1]-[!#1]-1-[#0&D2]-2 |(0.562,0.77,;0.562,-0.77,;-0.9027,-1.2459,;-1.8079,0,;-0.9027,1.2459,;-0.5912,0,)|

ZontaNicola · 2024-11-18T07:11:08Z

@greglandrum @rachelnwalker the new template atoms are marked as [!#1], which is fine by my, but I think will match the atoms we mark as outside of a ring system here https://github.dev/rdkit/rdkit/blob/eaf544ab6f7df12dd0f738474e44f9792d97d971/Code/GraphMol/Depictor/EmbeddedFrag.cpp#L429-L433.

Maybe [!#200] would be better? What do you think is the best solution in terms of speed/template readability?

rachelnwalker · 2024-11-22T00:33:39Z

@greglandrum @rachelnwalker the new template atoms are marked as [!#1], which is fine by my, but I think will match the atoms we mark as outside of a ring system here https://github.dev/rdkit/rdkit/blob/eaf544ab6f7df12dd0f738474e44f9792d97d971/Code/GraphMol/Depictor/EmbeddedFrag.cpp#L429-L433.

Maybe [!#200] would be better? What do you think is the best solution in terms of speed/template readability?

Yeah I think [!#200] is the best solution. Could we even mark the degree-specific atom queries with !#200 instead of #0? I guess it probably wouldn't make that much of a difference...

ZontaNicola · 2024-12-02T05:34:14Z

implemented [!#200] for every atom

rachelnwalker

lgtm, but it would be helpful to see the gallery of template depictions if possible.

ZontaNicola · 2024-12-03T05:53:05Z

Here's a few random examples of the new templates:

ZontaNicola · 2024-12-03T06:00:15Z

I've looked at a lot more and they all seem correct. The only thing is that centers with 3 substituents at 120° sometimes get marked as "internal" even if they could actually get a substitution: due to the symmetry of the system the average of the substituents falls exactly on the atom, so its direction is undetermined. I don't think that in general we would ever want a substitution there... maybe we can mark ALL centers with more than 2 bonds as [!#200&D3] so they never get substitutions. @rachelnwalker what do you think?

rachelnwalker · 2024-12-04T17:18:13Z

The third example you have has an atom with three substituents that we probably want to be able to attach to, so I don't think we should mark all of the atoms. I think its OK to not get a substitution with atoms with 3 equally spaced neighbors, though.

This lgtm though! In order to force the header update, I think we will need to make a second PR after this one (maybe add a template?) or otherwise trigger the GH action that generates the header and makes the PR

ZontaNicola · 2024-12-04T17:23:42Z

sounds good, thanks

greglandrum

One requested change

I would also suggest running yapf or black or some kind of python code formatter across this in order to cleanup the formatting and oddly indented comments. That's not a huge deal, but I do think it helps with maintainability to have consistently formatted code

greglandrum · 2025-01-30T08:55:00Z

@ZontaNicola just a quick ping in case you missed the notification of my review

ZontaNicola · 2025-01-30T09:07:16Z

@ZontaNicola just a quick ping in case you missed the notification of my review

thanks, I'm aware of it. I should be able to get back to this case soon

ZontaNicola · 2025-01-30T10:42:57Z

@greglandrum I ran yapf and removed the unused function

greglandrum

The code changes look good to me.
@rachelnwalker or @ricrogz will need to confirm, but I think we should not be modifying the template_smiles.h directly in a PR
(I know you included the changes here in order to show what will happen, but I think that needs to be reverted?)

rachelnwalker · 2025-02-06T18:21:29Z

@greglandrum is correct, @ZontaNicola I think if we just remove the changes to templates.h we should be able to merge this.

@ricrogz will the github action still trigger if no templates are added to the templates.smi file?

ricrogz · 2025-02-06T21:13:34Z

@ricrogz will the github action still trigger if no templates are added to the templates.smi file?

For the GHA to trigger, there needs to be a change to the templates.smi file. Changing the order of the smiles in it should work :)

…rigger automated update

ZontaNicola · 2025-03-27T17:04:18Z

reverted the changes and swapped the last two templates to trigger update as Ricardo suggested. Sorry this took so long!

ZontaNicola · 2025-04-02T08:14:31Z

@greglandrum does this look good to you?

rachelnwalker

Lgtm after updating the .smi file to trigger the GH action!

rachelnwalker · 2025-05-05T16:50:00Z

-    "C1C2C3CC4C5CC3C1C5C24 |(-1.68833,-1.6056,;-0.474804,-0.723925,;-1.13662,-0.251107,;0.360708,1.45266,;-0.93833,2.20266,;-2.43833,2.20266,;-3.73737,1.45266,;-2.19718,-0.251107,;-2.90185,-0.723925,;-2.43833,0.70266,;-0.93833,0.70266,)|",
    "C1CC2CC1C2 |(0.562,0.77,;0.562,-0.77,;-0.9027,-1.2459,;-1.8079,0,;-0.9027,1.2459,;-0.5912,0,)|",
+    "C1C2C3CC4C5CC3C1C5C24 |(-1.68833,-1.6056,;-0.474804,-0.723925,;-1.13662,-0.251107,;0.360708,1.45266,;-0.93833,2.20266,;-2.43833,2.20266,;-3.73737,1.45266,;-2.19718,-0.251107,;-2.90185,-0.723925,;-2.43833,0.70266,;-0.93833,0.70266,)|",
+


If you did this to trigger the GH action -- I think you need to make the change in the .smi file and not the .h file

oh I see, thanks! it should be fixed now

github-actions · 2025-05-07T07:50:13Z

Something went wrong, please check https://github.com/rdkit/molecular_templates/actions/runs/14878048769 for more information.

github-actions · 2025-05-08T19:39:47Z

Something went wrong, please check https://github.com/rdkit/molecular_templates/actions/runs/14878048769 for more information.

ricrogz · 2025-05-08T19:49:22Z

Ok, this failed because Nick's PR didn't add or remove any SMILES from the file, he just moved them around.

github-actions · 2025-05-08T20:49:01Z

Something went wrong, please check https://github.com/rdkit/molecular_templates/actions/runs/14915849355 for more information.

github-actions · 2025-05-08T21:14:39Z

No new templates were found, so no images were generated.

ricrogz · 2025-05-08T21:15:12Z

No new templates were found, so no images were generated.

Ok, seems this type of changes is now handled

greglandrum

LGTM

mark atoms inside rings

3b29cec

rachelnwalker reviewed Nov 14, 2024

View reviewed changes

update templates.smi

4fac986

reset header

0d55a19

ZontaNicola mentioned this pull request Nov 18, 2024

fix problems with template matching rdkit/rdkit#7975

Merged

set alla toms to !#200

393d561

rachelnwalker reviewed Dec 3, 2024

View reviewed changes

Comment thread src/header_generation.py Outdated

Comment thread src/header_generation.py Outdated

Comment thread src/header_generation.py

pr suggestions

3c59a79

ZontaNicola requested review from greglandrum and rachelnwalker December 3, 2024 06:05

rachelnwalker approved these changes Dec 4, 2024

View reviewed changes

greglandrum requested changes Dec 19, 2024

View reviewed changes

Comment thread src/header_generation.py Outdated

ZontaNicola mentioned this pull request Jan 23, 2025

Clashes in 2D layout with bridged systems rdkit/rdkit#6753

Open

pr suggestions

d2bf684

ZontaNicola requested a review from greglandrum January 30, 2025 10:41

greglandrum reviewed Jan 30, 2025

View reviewed changes

revert template_smiles.h changes and swap the last two templates to t…

c1242ca

…rigger automated update

ZontaNicola requested review from greglandrum and rachelnwalker March 27, 2025 17:03

rachelnwalker approved these changes May 5, 2025

View reviewed changes

update templates.smi

b95e1fd

Merge branch 'main' into main

d08ec0b

Merge branch 'main' into main

3cfba99

greglandrum approved these changes May 9, 2025

View reviewed changes

greglandrum merged commit d24273e into rdkit:main May 9, 2025
1 check passed

Conversation

ZontaNicola commented Nov 13, 2024

PLEASE DO NOT MODIFY template_smiles.h DIRECTLY

Uh oh!

rachelnwalker Nov 14, 2024

Choose a reason for hiding this comment

Uh oh!

ZontaNicola Nov 14, 2024

Choose a reason for hiding this comment

Uh oh!

greglandrum Nov 14, 2024

Choose a reason for hiding this comment

Uh oh!

rachelnwalker commented Nov 14, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ZontaNicola commented Nov 14, 2024

Uh oh!

rachelnwalker commented Nov 14, 2024

Uh oh!

ZontaNicola commented Nov 18, 2024 • edited by greglandrum Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ZontaNicola commented Nov 18, 2024

Uh oh!

rachelnwalker commented Nov 22, 2024

Uh oh!

ZontaNicola commented Dec 2, 2024

Uh oh!

rachelnwalker left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ZontaNicola commented Dec 3, 2024

Uh oh!

ZontaNicola commented Dec 3, 2024

Uh oh!

rachelnwalker commented Dec 4, 2024

Uh oh!

ZontaNicola commented Dec 4, 2024

Uh oh!

greglandrum left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

greglandrum commented Jan 30, 2025

Uh oh!

ZontaNicola commented Jan 30, 2025

Uh oh!

ZontaNicola commented Jan 30, 2025

Uh oh!

greglandrum left a comment

Choose a reason for hiding this comment

Uh oh!

rachelnwalker commented Feb 6, 2025

Uh oh!

ricrogz commented Feb 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ZontaNicola commented Mar 27, 2025

Uh oh!

ZontaNicola commented Apr 2, 2025

Uh oh!

rachelnwalker left a comment

Choose a reason for hiding this comment

Uh oh!

rachelnwalker May 5, 2025

Choose a reason for hiding this comment

Uh oh!

ZontaNicola May 7, 2025

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented May 7, 2025

Uh oh!

github-actions bot commented May 8, 2025

Uh oh!

ricrogz commented May 8, 2025

Uh oh!

rachelnwalker commented Nov 14, 2024 •

edited

Loading

ZontaNicola commented Nov 18, 2024 •

edited by greglandrum

Loading

ricrogz commented Feb 6, 2025 •

edited

Loading