Behavior change in substruct match between 2024.9.6 and 2025.3.1

[dwhswenson]

Describe the bug
A change has occurred in substructure matching between version 2024.9.6 and 2025.3.1. The release notes mention a change that "AdjustQueryProperties now no longer ignores additional atom adjustments when makeAtomsGeneric is enabled", but I wouldn't have expected that to break this workflow:

To Reproduce

# mcve.py
# Usage: python mcve.py TEMPLATE.sdf MOLECULE.pdb
import sys
from rdkit import __version__ as v
from rdkit import Chem

print(f"rdkit {v}")

template_filename, molecule_filename = sys.argv[1:]

template = Chem.MolFromMolFile(template_filename, sanitize=True, removeHs=True)
params = Chem.AdjustQueryParameters()
params.makeAtomsGeneric = True
params.makeBondsGeneric = True
template = Chem.AdjustQueryProperties(template, params)

molecule = Chem.MolFromPDBFile(molecule_filename, removeHs=False)

assert molecule.HasSubstructMatch(template), \
        f"I thought this would work for {template_filename}"
print(f"It did work for {template_filename}")

Use pairs (hem.sdf / HEM.pdb) and (adamantine.sdf / ADM.pdb) as examples.

HEM.pdb.txt
ADM.pdb.txt
hem.sdf.txt
adamantane.sdf.txt

$ python mcve.py adamantane.sdf ADM.pdb
rdkit 2025.03.1
Traceback (most recent call last):
  File "/Users/dwhs/tmp/rdkit-pdbeccdutils-2025/mcve.py", line 19, in <module>
    assert molecule.HasSubstructMatch(template), \
           ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^
AssertionError: I thought this would work for adamantane.sdf
$ pixi shell -e passing
$ python mcve.py adamantane.sdf ADM.pdb
rdkit 2024.09.6
It did work for adamantane.sdf
$ exit
$ python mcve.py hem.sdf HEM.pdb
rdkit 2025.03.1
Traceback (most recent call last):
  File "/Users/dwhs/tmp/rdkit-pdbeccdutils-2025/mcve.py", line 19, in <module>
    assert molecule.HasSubstructMatch(template), \
           ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^
AssertionError: I thought this would work for hem.sdf
$ pixi shell -e passing
$ python mcve.py hem.sdf HEM.pdb
rdkit 2024.09.6
It did work for hem.sdf

Expected behavior
I expected these patterns to match, as they did in earlier RDKit versions.

Screenshots
Pairs of template, molecule for context:

Configuration (please complete the following information):

• RDKit version: 2024.9.6 / 2025.3.1 (behavior continues in later releases)
• OS: macOS 15.2
• Python version (if relevant): 3.13.5
• Are you using conda? yes (via pixi)
• If you are using conda, which channel did you install the rdkit from? conda-forge
• If you are not using conda: how did you install the RDKit?

[greglandrum]

The default behavior of AdjustQueryProperties() is to add degree queries to non-chain atoms. So your heme template becomes:

In [14]: Chem.MolToSmarts(adj_templ)
Out[14]: '[D3]12~[D2]~[D2]~[D3]3~[D2]~[D3]4~[D2]~[D2]~[D3]5~[D2]~[D3]6~[D2]~[D2]~[D3]7~[D2]~[D3]8~[D2]~[D2]~[D3](~[D2]~1)~[D3]~8~[D4](~[D3]~2~3)(~[D3]~4~5)~[D3]~6~7'

This will, of course, not match anything that has a substitution on the heme ring.
If you want to disable that, you can do:

In [17]: params.adjustDegree = False

In [18]: adj_templ2 = Chem.AdjustQueryProperties(templ,params)

and then you get:

In [19]: Chem.MolToSmarts(adj_templ2)
Out[19]: '*12~*~*~*3~*~*4~*~*~*5~*~*6~*~*~*7~*~*8~*~*~*(~*~1)~*~8~*(~*~2~3)(~*~4~5)~*~6~7'

In [20]: mol.HasSubstructMatch(adj_templ2)
Out[20]: True